While the actual problem is caused by a ccxt change, the change itself makes sense.
once ccxt starts returning the correct status (open) for create-orders, we can remove the fix.
closes#8079
before calling `git`. otherwise it would display git commit id from the
directory where you are calling `freqtrade` from instead of freqtrade's
current commit id
Improve the RL learning process by selecting random start point for the agent, it can help to block the agent to only learn on the selected period of time, while improving the quality of the model.
_update_total_profit() must be executed before "self._position = Positions.Neutral" because _update_total_profit() calls get_unrealized_profit(), which returns 0 if position is neutral and total_profit is not updated
* Allow use of --strategy-list with freqai, with warning
* ensure populate_any_indicators is identical for resused identifiers
* use pair instead of metadata["pair"]
Co-authored-by: robcaulk <rob.caulk@gmail.com>
Now that a recent bug regarding selling BNB is fixed, it should be safe to trade it, but with a warning that the user may have to manually maintain extra BNB.
Also the old text implied those features are always unabled so this texts makes it clear those fee-related features can be also disabled.
I'm not sure if it's still true that an "eaten by fees" position becomes unsellable but I left that as it is.
@@ -28,7 +29,6 @@ Please read the [exchange specific notes](docs/exchanges.md) to learn about even
- [X] [Binance](https://www.binance.com/)
- [X] [Bittrex](https://bittrex.com/)
- [X] [FTX](https://ftx.com/#a=2258149)
- [X] [Gate.io](https://www.gate.io/ref/6266643)
- [X] [Huobi](http://huobi.com/)
- [X] [Kraken](https://kraken.com/)
@@ -39,7 +39,8 @@ Please read the [exchange specific notes](docs/exchanges.md) to learn about even
- [X] [Binance](https://www.binance.com/)
- [X] [Gate.io](https://www.gate.io/ref/6266643)
- [X] [OKX](https://okx.com/).
- [X] [OKX](https://okx.com/)
- [X] [Bybit](https://bybit.com/)
Please make sure to read the [exchange specific notes](docs/exchanges.md), as well as the [trading with leverage](docs/leverage.md) documentation before diving in.
@@ -164,6 +165,10 @@ first. If it hasn't been reported, please
ensure you follow the template guide so that the team can assist you as
quickly as possible.
For every [issue](https://github.com/freqtrade/freqtrade/issues/new/choose) created, kindly follow up and mark satisfaction or reminder to close issue when equilibrium ground is reached.
@@ -74,9 +75,11 @@ This function needs to return a floating point number (`float`). Smaller numbers
## Overriding pre-defined spaces
To override a pre-defined space (`roi_space`, `generate_roi_table`, `stoploss_space`, `trailing_space`), define a nested class called Hyperopt and define the required spaces as follows:
To override a pre-defined space (`roi_space`, `generate_roi_table`, `stoploss_space`, `trailing_space`, `max_open_trades_space`), define a nested class called Hyperopt and define the required spaces as follows:
```python
from freqtrade.optimize.space import Categorical, Dimension, Integer, SKDecimal
class MyAwesomeStrategy(IStrategy):
class HyperOpt:
# Define a custom stoploss space.
@@ -93,6 +96,39 @@ class MyAwesomeStrategy(IStrategy):
This needs the `systemd` python package installed as the dependency, which is not available on Windows. Hence, the whole journald logging functionality is not available for a bot running on Windows.
This needs the `cysystemd` python package installed as dependency (`pip install cysystemd`), which is not available on Windows. Hence, the whole journald logging functionality is not available for a bot running on Windows.
To send Freqtrade log messages to `journald` system service use the `--logfile` command line option with the value in the following format:
Now you have good Buy and Sell strategies and some historic data, you want to test it against
Now you have good Entry and exit strategies and some historic data, you want to test it against
real data. This is what we call [backtesting](https://en.wikipedia.org/wiki/Backtesting).
Backtesting will use the crypto-currencies (pairs) from your config file and load historical candle (OHLCV) data from `user_data/data/<exchange>` by default.
@@ -215,7 +215,7 @@ Sometimes your account has certain fee rebates (fee reductions starting with a c
To account for this in backtesting, you can use the `--fee` command line option to supply this value to backtesting.
This fee must be a ratio, and will be applied twice (once for trade entry, and once for trade exit).
For example, if the buying and selling commission fee is 0.1% (i.e., 0.001 written as ratio), then you would run backtesting as the following:
For example, if the commission fee per order is 0.1% (i.e., 0.001 written as ratio), then you would run backtesting as the following:
```bash
freqtrade backtesting --fee 0.001
@@ -252,41 +252,41 @@ The most important in the backtesting is to understand the result.
@@ -300,7 +300,11 @@ A backtesting result will look like that:
| Absolute profit | 0.00762792 BTC |
| Total profit % | 76.2% |
| CAGR % | 460.87% |
| Sortino | 1.88 |
| Sharpe | 2.97 |
| Calmar | 6.29 |
| Profit factor | 1.11 |
| Expectancy | -0.15 |
| Avg. stake amount | 0.001 BTC |
| Total trade volume | 0.429 BTC |
| | |
@@ -356,7 +360,7 @@ The column `Avg Profit %` shows the average profit for all trades made while the
The column `Tot Profit %` shows instead the total profit % in relation to the starting balance.
In the above results, we have a starting balance of 0.01 BTC and an absolute profit of 0.00762792 BTC - so the `Tot Profit %` will be `(0.00762792 / 0.01) * 100 ~= 76.2%`.
Your strategy performance is influenced by your buy strategy, your exit strategy, and also by the `minimal_roi` and `stop_loss` you have set.
Your strategy performance is influenced by your entry strategy, your exit strategy, and also by the `minimal_roi` and `stop_loss` you have set.
For example, if your `minimal_roi` is only `"0": 0.01` you cannot expect the bot to make more profit than 1% (because it will exit every time a trade reaches 1%).
@@ -400,7 +404,11 @@ It contains some useful key metrics about performance of your strategy on backte
| Absolute profit | 0.00762792 BTC |
| Total profit % | 76.2% |
| CAGR % | 460.87% |
| Sortino | 1.88 |
| Sharpe | 2.97 |
| Calmar | 6.29 |
| Profit factor | 1.11 |
| Expectancy | -0.15 |
| Avg. stake amount | 0.001 BTC |
| Total trade volume | 0.429 BTC |
| | |
@@ -447,6 +455,9 @@ It contains some useful key metrics about performance of your strategy on backte
-`Absolute profit`: Profit made in stake currency.
-`Total profit %`: Total profit. Aligned to the `TOTAL` row's `Tot Profit %` from the first table. Calculated as `(End capital − Starting capital) / Starting capital`.
-`CAGR %`: Compound annual growth rate.
-`Sortino`: Annualized Sortino ratio.
-`Sharpe`: Annualized Sharpe ratio.
-`Calmar`: Annualized Calmar ratio.
-`Profit factor`: profit / loss.
-`Avg. stake amount`: Average stake amount, either `stake_amount` or the average when using dynamic stake amount.
-`Total trade volume`: Volume generated on the exchange to reach the above profit.
@@ -515,20 +526,20 @@ You can then load the trades to perform further analysis as shown in the [data a
Since backtesting lacks some detailed information about what happens within a candle, it needs to take a few assumptions:
- Exchange [trading limits](#trading-limits-in-backtesting) are respected
- Buys happen at open-price
- Entries happen at open-price
- All orders are filled at the requested price (no slippage, no unfilled orders)
- Exit-signal exits happen at open-price of the consecutive candle
- Exit-signal is favored over Stoploss, because exit-signals are assumed to trigger on candle's open
- ROI
- exits are compared to high - but the ROI value is used (e.g. ROI = 2%, high=5% - so the exit will be at 2%)
- exits are never "below the candle", so a ROI of 2% may result in a exit at 2.4% if low was at 2.4% profit
- Forceexits caused by `<N>=-1` ROI entries use low as exit value, unless N falls on the candle open (e.g. `120: -1` for 1h candles)
- Force-exits caused by `<N>=-1` ROI entries use low as exit value, unless N falls on the candle open (e.g. `120: -1` for 1h candles)
- Stoploss exits happen exactly at stoploss price, even if low was lower, but the loss will be `2 * fees` higher than the stoploss price
- Stoploss is evaluated before ROI within one candle. So you can often see more trades with the `stoploss` exit reason comparing to the results obtained with the same strategy in the Dry Run/Live Trade modes
- Low happens before high for stoploss, protecting capital first
- Trailing stoploss
- Trailing Stoploss is only adjusted if it's below the candle's low (otherwise it would be triggered)
- On trade entry candles that trigger trailing stoploss, the "minimum offset" (`stop_positive_offset`) is assumed (instead of high) - and the stop is calculated from this point
- On trade entry candles that trigger trailing stoploss, the "minimum offset" (`stop_positive_offset`) is assumed (instead of high) - and the stop is calculated from this point. This rule is NOT applicable to custom-stoploss scenarios, since there's no information about the stoploss logic available.
- High happens first - adjusting stoploss
- Low uses the adjusted stoploss (so exits with large high-low difference are backtested correctly)
- ROI applies before trailing-stop, ensuring profits are "top-capped" at ROI if both ROI and trailing stop applies
@@ -546,8 +557,8 @@ In addition to the above assumptions, strategy authors should carefully read the
### Trading limits in backtesting
Exchanges have certain trading limits, like minimum base currency, or minimum stake (quote) currency.
These limits are usually listed in the exchange documentation as "trading rules" or similar.
Exchanges have certain trading limits, like minimum (and maximum) base currency, or minimum/maximum stake (quote) currency.
These limits are usually listed in the exchange documentation as "trading rules" or similar and can be quite different between different pairs.
Backtesting (as well as live and dry-run) does honor these limits, and will ensure that a stoploss can be placed below this value - so the value will be slightly higher than what the exchange specifies.
Freqtrade has however no information about historic limits.
@@ -583,7 +594,8 @@ To utilize this, you can append `--timeframe-detail 5m` to your regular backtest
This will load 1h data as well as 5m data for the timeframe. The strategy will be analyzed with the 1h timeframe - and for every "open trade candle" (candles where a trade is open) the 5m data will be used to simulate intra-candle movements.
This will load 1h data as well as 5m data for the timeframe. The strategy will be analyzed with the 1h timeframe, and Entry orders will only be placed at the main timeframe, however Order fills and exit signals will be evaluated at the 5m candle, simulating intra-candle movements.
All callback functions (`custom_exit()`, `custom_stoploss()`, ... ) will be running for each 5m candle once the trade is opened (so 12 times in the above example of 1h timeframe, and 5m detailed timeframe).
`--timeframe-detail` must be smaller than the original timeframe, otherwise backtesting will fail to start.
@@ -612,11 +624,11 @@ There will be an additional table comparing win/losses of the different strategi
Detailed output for all strategies one after the other will be available, so make sure to scroll up to see the details per strategy.
@@ -75,3 +75,7 @@ This loop will be repeated again and again until the bot is stopped.
!!! Note
Both Backtesting and Hyperopt include exchange default Fees in the calculation. Custom fees can be passed to backtesting / hyperopt by specifying the `--fee` argument.
!!! Warning "Callback call frequency"
Backtesting will call each callback at max. once per candle (`--timeframe-detail` modifies this behavior to once per detailed candle).
Most callbacks will be called once per iteration in live (usually every ~5s) - which can cause backtesting mismatches.
@@ -11,7 +11,7 @@ Per default, the bot loads the configuration from the `config.json` file, locate
You can specify a different configuration file used by the bot with the `-c/--config` command-line option.
If you used the [Quick start](installation.md/#quick-start) method for installing
If you used the [Quick start](docker_quickstart.md#docker-quick-start) method for installing
the bot, the installation script should have already created the default configuration file (`config.json`) for you.
If the default configuration file is not created we recommend to use `freqtrade new-config --config config.json` to generate a basic configuration file.
@@ -58,9 +58,20 @@ This is similar to using multiple `--config` parameters, but simpler in usage as
!!! Tip "Use multiple configuration files to keep secrets secret"
You can use a 2nd configuration file containing your secrets. That way you can share your "primary" configuration file, while still keeping your API keys for yourself.
The 2nd file should only specify what you intend to override.
If a key is in more than one of the configurations, then the "last specified configuration" wins (in the above example, `config-private.json`).
For one-off commands, you can also use the below syntax by specifying multiple "--config" parameters.
This is equivalent to the example above - but `config-private.json` is specified as cli argument.
??? Note "config collision handling"
If the same configuration setting takes place in both `config.json` and `config-import.json`, then the parent configuration wins.
In the below case, `max_open_trades` would be 3 after the merging - as the reusable "import" configuration has this key overwritten.
@@ -111,6 +111,8 @@ This is similar to using multiple `--config` parameters, but simpler in usage as
}
```
If multiple files are in the `add_config_files` section, then they will be assumed to be at identical levels, having the last occurrence override the earlier config (unless a parent already defined such a key).
## Configuration parameters
The table below will list all configuration parameters available.
@@ -132,7 +134,7 @@ Mandatory parameters are marked as **Required**, which means that they are requi
| Parameter | Description |
|------------|-------------|
| `max_open_trades` | **Required.** Number of open trades your bot is allowed to have. Only one open trade per pair is possible, so the length of your pairlist is another limitation that can apply. If -1 then it is ignored (i.e. potentially unlimited open trades, limited by the pairlist). [More information below](#configuring-amount-per-trade).<br> **Datatype:** Positive integer or -1.
| `max_open_trades` | **Required.** Number of open trades your bot is allowed to have. Only one open trade per pair is possible, so the length of your pairlist is another limitation that can apply. If -1 then it is ignored (i.e. potentially unlimited open trades, limited by the pairlist). [More information below](#configuring-amount-per-trade). [Strategy Override](#parameters-in-the-strategy).<br> **Datatype:** Positive integer or -1.
| `stake_currency` | **Required.** Crypto-currency used for trading. <br> **Datatype:** String
| `stake_amount` | **Required.** Amount of crypto-currency your bot will use for each trade. Set it to `"unlimited"` to allow the bot to use all available balance. [More information below](#configuring-amount-per-trade). <br> **Datatype:** Positive float or `"unlimited"`.
| `tradable_balance_ratio` | Ratio of the total account balance the bot is allowed to trade. [More information below](#configuring-amount-per-trade). <br>*Defaults to `0.99` 99%).*<br> **Datatype:** Positive float between `0.1` and `1.0`.
@@ -213,24 +215,28 @@ Mandatory parameters are marked as **Required**, which means that they are requi
| `telegram.balance_dust_level` | Dust-level (in stake currency) - currencies with a balance below this will not be shown by `/balance`. <br> **Datatype:** float
| `telegram.reload` | Allow "reload" buttons on telegram messages. <br>*Defaults to `True`.<br> **Datatype:** boolean
| `telegram.notification_settings.*` | Detailed notification settings. Refer to the [telegram documentation](telegram-usage.md) for details.<br> **Datatype:** dictionary
| `telegram.allow_custom_messages` | Enable the sending of Telegram messages from strategies via the dataprovider.send_msg() function. <br> **Datatype:** Boolean
| `webhook.url` | URL for the webhook. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.webhookentry` | Payload to send on entry. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.webhookentrycancel` | Payload to send on entry order cancel. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.webhookentryfill` | Payload to send on entry order filled. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.webhookexit` | Payload to send on exit. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.webhookexitcancel` | Payload to send on exit order cancel. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.webhookexitfill` | Payload to send on exit order filled. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.webhookstatus` | Payload to send on status calls. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| | **Rest API / FreqUI**
| `webhook.entry` | Payload to send on entry. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.entry_cancel` | Payload to send on entry order cancel. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.entry_fill` | Payload to send on entry order filled. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.exit` | Payload to send on exit. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.exit_cancel` | Payload to send on exit order cancel. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.exit_fill` | Payload to send on exit order filled. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.status` | Payload to send on status calls. Only required if `webhook.enabled` is `true`. See the [webhook documentation](webhook-config.md) for more details. <br> **Datatype:** String
| `webhook.allow_custom_messages` | Enable the sending of Webhook messages from strategies via the dataprovider.send_msg() function. <br> **Datatype:** Boolean
| | **Rest API / FreqUI / Producer-Consumer**
| `api_server.enabled` | Enable usage of API Server. See the [API Server documentation](rest-api.md) for more details. <br> **Datatype:** Boolean
| `api_server.listen_ip_address` | Bind IP address. See the [API Server documentation](rest-api.md) for more details. <br> **Datatype:** IPv4
| `api_server.listen_port` | Bind Port. See the [API Server documentation](rest-api.md) for more details. <br>**Datatype:** Integer between 1024 and 65535
| `api_server.verbosity` | Logging verbosity. `info` will print all RPC Calls, while "error" will only display errors. <br>**Datatype:** Enum, either `info` or `error`. Defaults to `info`.
| `api_server.username` | Username for API server. See the [API Server documentation](rest-api.md) for more details. <br>**Keep it in secret, do not disclose publicly.**<br> **Datatype:** String
| `api_server.password` | Password for API server. See the [API Server documentation](rest-api.md) for more details. <br>**Keep it in secret, do not disclose publicly.**<br> **Datatype:** String
| `api_server.ws_token` | API token for the Message WebSocket. See the [API Server documentation](rest-api.md) for more details. <br>**Keep it in secret, do not disclose publicly.** <br> **Datatype:** String
| `bot_name` | Name of the bot. Passed via API to a client - can be shown to distinguish / name bots.<br> *Defaults to `freqtrade`*<br> **Datatype:** String
| `external_message_consumer` | Enable [Producer/Consumer mode](producer-consumer.md) for more details. <br> **Datatype:** Dict
| | **Other**
| `initial_state` | Defines the initial application state. If set to stopped, then the bot has to be explicitly started via `/start` RPC command. <br>*Defaults to `stopped`.* <br> **Datatype:** Enum, either `stopped` or `running`
| `force_entry_enable` | Enables the RPC Commands to force a Trade entry. More information below. <br> **Datatype:** Boolean
@@ -247,6 +253,7 @@ Mandatory parameters are marked as **Required**, which means that they are requi
| `add_config_files` | Additional config files. These files will be loaded and merged with the current config file. The files are resolved relative to the initial file.<br> *Defaults to `[]`*. <br> **Datatype:** List of strings
| `dataformat_ohlcv` | Data format to use to store historical candle (OHLCV) data. <br> *Defaults to `json`*. <br> **Datatype:** String
| `dataformat_trades` | Data format to use to store historical trades data. <br> *Defaults to `jsongz`*. <br> **Datatype:** String
| `reduce_df_footprint` | Recast all numeric columns to float32/int32, with the objective of reducing ram/disk usage (and decreasing train/inference timing in FreqAI). (Currently only affects FreqAI use-cases) <br> **Datatype:** Boolean. <br> Default: `False`.
### Parameters in the strategy
@@ -256,6 +263,7 @@ Values set in the configuration file always overwrite values set in the strategy
* `minimal_roi`
* `timeframe`
* `stoploss`
* `max_open_trades`
* `trailing_stop`
* `trailing_stop_positive`
* `trailing_stop_positive_offset`
@@ -525,21 +533,28 @@ It means if the order is not executed immediately AND fully then it is cancelled
It is the same as FOK (above) except it can be partially fulfilled. The remaining part
is automatically cancelled by the exchange.
The `order_time_in_force` parameter contains a dict with buy and sell time in force policy values.
**PO (Post only):**
Post only order. The order is either placed as a maker order, or it is canceled.
This means the order must be placed on orderbook for at at least time in an unfilled state.
#### time_in_force config
The `order_time_in_force` parameter contains a dict with entry and exit time in force policy values.
This can be set in the configuration file or in the strategy.
Values set in the configuration file overwrites values set in the strategy.
The possible values are: `gtc` (default), `fok` or `ioc`.
The possible values are: `GTC` (default), `FOK` or `IOC`.
``` python
"order_time_in_force": {
"entry": "gtc",
"exit": "gtc"
"entry": "GTC",
"exit": "GTC"
},
```
!!! Warning
This is ongoing work. For now, it is supported only for binance and kucoin.
This is ongoing work. For now, it is supported only for binance, gate and kucoin.
Please don't change the default value unless you know what you are doing and have researched the impact of using different values for your particular exchange.
### What values can be used for fiat_display_currency?
@@ -650,17 +665,8 @@ You should also make sure to read the [Exchanges](exchanges.md) section of the d
### Using proxy with Freqtrade
To use a proxy with freqtrade, add the kwarg `"aiohttp_trust_env"=true` to the `"ccxt_async_kwargs"` dict in the exchange section of the configuration.
An example for this can be found in `config_examples/config_full.example.json`
``` json
"ccxt_async_config": {
"aiohttp_trust_env": true
}
```
Then, export your proxy settings using the variables `"HTTP_PROXY"` and `"HTTPS_PROXY"` set to the appropriate values
To use a proxy with freqtrade, export your proxy settings using the variables `"HTTP_PROXY"` and `"HTTPS_PROXY"` set to the appropriate values.
This will have the proxy settings applied to everything (telegram, coingecko, ...) **except** for exchange requests.
Freqtrade currently supports 3 data-formats for both OHLCV and trades data:
Freqtrade currently supports the following data-formats:
* `json` (plain "text" json files)
* `jsongz` (a gzip-zipped version of json files)
* `hdf5` (a high performance datastore)
* `json` - plain "text" json files
* `jsongz` - a gzip-zipped version of json files
* `hdf5` - a high performance datastore
* `feather` - a dataformat based on Apache Arrow (OHLCV only)
* `parquet` - columnar datastore (OHLCV only)
By default, OHLCV data is stored as `json` data, while trades data is stored as `jsongz` data.
@@ -200,38 +202,74 @@ If the default data-format has been changed during download, then the keys `data
!!! Note
You can convert between data-formats using the [convert-data](#sub-command-convert-data) and [convert-trade-data](#sub-command-convert-trade-data) methods.
#### Dataformat comparison
The following comparisons have been made with the following data, and by using the linux `time` command.
@@ -24,7 +24,7 @@ This will spin up a local server (usually on port 8000) so you can see if everyt
To configure a development environment, you can either use the provided [DevContainer](#devcontainer-setup), or use the `setup.sh` script and answer "y" when asked "Do you want to install dependencies for dev [y/N]? ".
Alternatively (e.g. if your system is not supported by the setup.sh script), follow the manual installation process and run `pip3 install -e .[all]`.
This will install all required tools for development, including `pytest`, `flake8`, `mypy`, and `coveralls`.
This will install all required tools for development, including `pytest`, `ruff`, `mypy`, and `coveralls`.
Then install the git hook scripts by running `pre-commit install`, so your changes will be verified locally before committing.
This avoids a lot of waiting for CI already, as some basic formatting checks are done locally on your machine.
@@ -49,6 +49,13 @@ For more information about the [Remote container extension](https://code.visuals
New code should be covered by basic unittests. Depending on the complexity of the feature, Reviewers may request more in-depth unittests.
If necessary, the Freqtrade team can assist and give guidance with writing good tests (however please don't expect anyone to write the tests for you).
#### How to run tests
Use `pytest` in root folder to run all available testcases and confirm your local environment is setup correctly
!!! Note "feature branches"
Tests are expected to pass on the `develop` and `stable` branches. Other branches may be work in progress with tests not working yet.
#### Checking log content in tests
Freqtrade uses 2 main methods to check log content in tests, `log_has()` and `log_has_re()` (to check using regex, in case of dynamic log-messages).
@@ -356,7 +363,7 @@ from pathlib import Path
exchange = ccxt.binance({
'apiKey': '<apikey>',
'secret': '<secret>'
'options': {'defaultType': 'future'}
'options': {'defaultType': 'swap'}
})
_ = exchange.load_markets()
@@ -409,8 +416,9 @@ Determine if crucial bugfixes have been made between this commit and the current
* Merge the release branch (stable) into this branch.
* Edit `freqtrade/__init__.py` and add the version matching the current date (for example `2019.7` for July 2019). Minor versions can be `2019.7.1` should we need to do a second release that month. Version numbers must follow allowed versions from PEP0440 to avoid failures pushing to pypi.
* Commit this part
* push that branch to the remote and create a PR against the stable branch
* Commit this part.
* push that branch to the remote and create a PR against the stable branch.
* Update develop version to next version following the pattern `2019.8-dev`.
### Create changelog from git commits
@@ -433,6 +441,11 @@ To keep the release-log short, best wrap the full git changelog into a collapsib
</details>
```
### FreqUI release
If FreqUI has been updated substantially, make sure to create a release before merging the release branch.
Make sure that freqUI CI on the release is finished and passed before merging the release.
### Create github release / tag
Once the PR against stable is merged (best right after merging):
To simplify running freqtrade, [`docker-compose`](https://docs.docker.com/compose/install/) should be installed and available to follow the below [docker quick start guide](#docker-quick-start).
!!! Info "Docker compose install"
Freqtrade documentation assumes the use of Docker desktop (or the docker compose plugin).
While the docker-compose standalone installation still works, it will require changing all `docker compose` commands from `docker compose` to `docker-compose` to work (e.g. `docker compose up -d` will become `docker-compose up -d`).
## Freqtrade with docker-compose
## Freqtrade with docker
Freqtrade provides an official Docker image on [Dockerhub](https://hub.docker.com/r/freqtradeorg/freqtrade/), as well as a [docker-compose file](https://github.com/freqtrade/freqtrade/blob/stable/docker-compose.yml) ready for usage.
Freqtrade provides an official Docker image on [Dockerhub](https://hub.docker.com/r/freqtradeorg/freqtrade/), as well as a [dockercompose file](https://github.com/freqtrade/freqtrade/blob/stable/docker-compose.yml) ready for usage.
!!! Note
- The following section assumes that `docker`and `docker-compose` are installed and available to the logged in user.
- The following section assumes that `docker`is installed and available to the logged in user.
- All below commands use relative directories and will have to be executed from the directory containing the `docker-compose.yml` file.
docker-compose run --rm freqtrade new-config --config user_data/config.json
dockercompose run --rm freqtrade new-config --config user_data/config.json
```
The above snippet creates a new directory called `ft_userdata`, downloads the latest compose file and pulls the freqtrade image.
@@ -64,7 +66,7 @@ The `SampleStrategy` is run by default.
Once this is done, you're ready to launch the bot in trading mode (Dry-run or Live-trading, depending on your answer to the corresponding question you made above).
``` bash
docker-compose up -d
dockercompose up -d
```
!!! Warning "Default configuration"
@@ -84,27 +86,27 @@ You can now access the UI by typing localhost:8080 in your browser.
#### Monitoring the bot
You can check for running instances with `docker-compose ps`.
You can check for running instances with `dockercompose ps`.
This should list the service `freqtrade` as `running`. If that's not the case, best check the logs (see next point).
#### Docker-compose logs
#### Dockercompose logs
Logs will be written to: `user_data/logs/freqtrade.log`.
You can also check the latest log with the command `docker-compose logs -f`.
You can also check the latest log with the command `dockercompose logs -f`.
#### Database
The database will be located at: `user_data/tradesv3.sqlite`
#### Updating freqtrade with docker-compose
#### Updating freqtrade with docker
Updating freqtrade when using `docker-compose` is as simple as running the following 2 commands:
Updating freqtrade when using `docker` is as simple as running the following 2 commands:
``` bash
# Download the latest image
docker-compose pull
dockercompose pull
# Restart the image
docker-compose up -d
dockercompose up -d
```
This will first pull the latest image, and will then restart the container with the just pulled version.
@@ -116,43 +118,43 @@ This will first pull the latest image, and will then restart the container with
Advanced users may edit the docker-compose file further to include all possible options or arguments.
All freqtrade arguments will be available by running `docker-compose run --rm freqtrade <command><optionalarguments>`.
All freqtrade arguments will be available by running `dockercompose run --rm freqtrade <command><optionalarguments>`.
!!! Warning "`docker-compose` for trade commands"
Trade commands (`freqtrade trade <...>`) should not be ran via `docker-compose run` - but should use `docker-compose up -d` instead.
!!! Warning "`dockercompose` for trade commands"
Trade commands (`freqtrade trade <...>`) should not be ran via `dockercompose run` - but should use `dockercompose up -d` instead.
This makes sure that the container is properly started (including port forwardings) and will make sure that the container will restart after a system reboot.
If you intend to use freqUI, please also ensure to adjust the [configuration accordingly](rest-api.md#configuration-with-docker), otherwise the UI will not be available.
!!! Note "`docker-compose run --rm`"
!!! Note "`dockercompose run --rm`"
Including `--rm` will remove the container after completion, and is highly recommended for all modes except trading mode (running with `freqtrade trade` command).
??? Note "Using docker without docker-compose"
"`docker-compose run --rm`" will require a compose file to be provided.
??? Note "Using docker without docker"
"`dockercompose run --rm`" will require a compose file to be provided.
Some freqtrade commands that don't require authentication such as `list-pairs` can be run with "`docker run --rm`" instead.
For example `docker run --rm freqtradeorg/freqtrade:stable list-pairs --exchange binance --quote BTC --print-json`.
This can be useful for fetching exchange information to add to your `config.json` without affecting your running containers.
#### Example: Download data with docker-compose
#### Example: Download data with docker
Download backtesting data for 5 days for the pair ETH/BTC and 1h timeframe from Binance. The data will be stored in the directory `user_data/data/` on the host.
Head over to the [Backtesting Documentation](backtesting.md) to learn more.
### Additional dependencies with docker-compose
### Additional dependencies with docker
If your strategy requires dependencies not included in the default image - it will be necessary to build the image on your host.
For this, please create a Dockerfile containing installation steps for the additional dependencies (have a look at [docker/Dockerfile.custom](https://github.com/freqtrade/freqtrade/blob/develop/docker/Dockerfile.custom) for an example).
@@ -166,15 +168,15 @@ You'll then also need to modify the `docker-compose.yml` file and uncomment the
dockerfile: "./Dockerfile.<yourextension>"
```
You can then run `docker-compose build --pull` to build the docker image, and run it using the commands described above.
You can then run `dockercompose build --pull` to build the docker image, and run it using the commands described above.
### Plotting with docker-compose
### Plotting with docker
Commands `freqtrade plot-profit` and `freqtrade plot-dataframe` ([Documentation](plotting.md)) are available by changing the image to `*_plot` in your docker-compose.yml file.
You can then use these commands as follows:
``` bash
docker-compose run --rm freqtrade plot-dataframe --strategy AwesomeStrategy -p BTC/ETH --timerange=20180801-20180805
dockercompose run --rm freqtrade plot-dataframe --strategy AwesomeStrategy -p BTC/ETH --timerange=20180801-20180805
```
The output will be stored in the `user_data/plot` directory, and can be opened with any modern browser.
@@ -185,7 +187,7 @@ Freqtrade provides a docker-compose file which starts up a jupyter lab server.
You can run this server using the following command:
``` bash
docker-compose -f docker/docker-compose-jupyter.yml up
dockercompose -f docker/docker-compose-jupyter.yml up
```
This will create a docker-container running jupyter lab, which will be accessible using `https://127.0.0.1:8888/lab`.
@@ -194,7 +196,7 @@ Please use the link that's printed in the console after startup for simplified l
Since part of this image is built on your machine, it is recommended to rebuild the image from time to time to keep freqtrade (and dependencies) up-to-date.
@@ -54,15 +54,45 @@ This configuration enables kraken, as well as rate-limiting to avoid bans from t
## Binance
!!! Warning "Server location and geo-ip restrictions"
Please be aware that binance restrict api access regarding the server country. The currents and non exhaustive countries blocked are United States, Malaysia (Singapour), Ontario (Canada). Please go to [binance terms > b. Eligibility](https://www.binance.com/en/terms) to find up to date list.
Binance supports `stoploss_on_exchange` and uses `stop-loss-limit` orders. It provides great advantages, so we recommend to benefit from it by enabling stoploss on exchange..
Binance supports `stoploss_on_exchange` and uses `stop-loss-limit` orders. It provides great advantages, so we recommend to benefit from it by enabling stoploss on exchange.
On futures, Binance supports both `stop-limit` as well as `stop-market` orders. You can use either `"limit"` or `"market"` in the `order_types.stoploss` configuration setting to decide which type to use.
### Binance Blacklist
### Binance Blacklist recommendation
For Binance, please add `"BNB/<STAKE>"` to your blacklist to avoid issues.
Accounts having BNB accounts use this to pay for fees - if your first trade happens to be on `BNB`, further trades will consume this position and make the initial BNB trade unsellable as the expected amount is not there anymore.
For Binance, it is suggested to add `"BNB/<STAKE>"` to your blacklist to avoid issues, unless you are willing to maintain enough extra `BNB` on the account or unless you're willing to disable using `BNB` for fees.
Binance accounts may use `BNB` for fees, and if a trade happens to be on `BNB`, further trades may consume this position and make the initial BNB trade unsellable as the expected amount is not there anymore.
### Binance sites
Binance has been split into 2, and users must use the correct ccxt exchange ID for their exchange, otherwise API keys are not recognized.
* [binance.com](https://www.binance.com/) - International users. Use exchange id: `binance`.
* [binance.us](https://www.binance.us/) - US based users. Use exchange id: `binanceus`.
They can however also be configured via configuration file. Since json doesn't support multi-line strings, you'll have to replace all newlines with `\n` to have a valid json file.
Freqtrade will not attempt to change these settings.
## Kraken
@@ -163,26 +195,6 @@ res = [p for p, x in lm.items() if 'US' in x['info']['prohibitedIn']]
print(res)
```
## FTX
!!! Tip "Stoploss on Exchange"
FTX supports `stoploss_on_exchange` and can use both stop-loss-market and stop-loss-limit orders. It provides great advantages, so we recommend to benefit from it.
You can use either `"limit"` or `"market"` in the `order_types.stoploss` configuration setting to decide which type of stoploss shall be used.
### Using subaccounts
To use subaccounts with FTX, you need to edit the configuration and add the following:
``` json
"exchange": {
"ccxt_config": {
"headers": {
"FTX-SUBACCOUNT": "name"
}
},
}
```
## Kucoin
Kucoin requires a passphrase for each api key, you will therefore need to add this key into the configuration so your exchange section looks as follows:
For Kucoin, please add `"KCS/<STAKE>"` to your blacklist to avoid issues.
Accounts having KCS accounts use this to pay for fees - if your first trade happens to be on `KCS`, further trades will consume this position and make the initial KCS trade unsellable as the expected amount is not there anymore.
For Kucoin, it is suggested to add `"KCS/<STAKE>"` to your blacklist to avoid issues, unless you are willing to maintain enough extra `KCS` on the account or unless you're willing to disable using `KCS` for fees.
Kucoin accounts may use `KCS` for fees, and if a trade happens to be on `KCS`, further trades may consume this position and make the initial `KCS` trade unsellable as the expected amount is not there anymore.
## Huobi
@@ -231,8 +243,8 @@ OKX requires a passphrase for each api key, you will therefore need to add this
OKX only provides 100 candles per api call. Therefore, the strategy will only have a pretty low amount of data available in backtesting mode.
!!! Warning "Futures"
OKX Futures has the concept of "position mode" - which can be Net or long/short (hedge mode).
Freqtrade supports both modes - but changing the mode mid-trading is not supported and will lead to exceptions and failures to place trades.
OKX Futures has the concept of "position mode" - which can be "Buy/Sell" or long/short (hedge mode).
Freqtrade supports both modes (we recommend to use Buy/Sell mode) - but changing the mode mid-trading is not supported and will lead to exceptions and failures to place trades.
OKX also only provides MARK candles for the past ~3 months. Backtesting futures prior to that date will therefore lead to slight deviations, as funding-fees cannot be calculated correctly without this data.
## Gate.io
@@ -243,6 +255,18 @@ OKX requires a passphrase for each api key, you will therefore need to add this
Gate.io allows the use of `POINT` to pay for fees. As this is not a tradable currency (no regular market available), automatic fee calculations will fail (and default to a fee of 0).
The configuration parameter `exchange.unknown_fee_rate` can be used to specify the exchange rate between Point and the stake currency. Obviously, changing the stake-currency will also require changes to this value.
## Bybit
Futures trading on bybit is currently supported for USDT markets, and will use isolated futures mode.
Users with unified accounts (there's no way back) can create a Sub-account which will start as "non-unified", and can therefore use isolated futures.
On startup, freqtrade will set the position mode to "One-way Mode" for the whole (sub)account. This avoids making this call over and over again (slowing down bot operations), but means that changes to this setting may result in exceptions and errors.
As bybit doesn't provide funding rate history, the dry-run calculation is used for live trades as well.
!!! Tip "Stoploss on Exchange"
Bybit (futures only) supports `stoploss_on_exchange` and uses `stop-loss-limit` orders. It provides great advantages, so we recommend to benefit from it by enabling stoploss on exchange.
On futures, Bybit supports both `stop-limit` as well as `stop-market` orders. You can use either `"limit"` or `"market"` in the `order_types.stoploss` configuration setting to decide which type to use.
## All exchanges
Should you experience constant errors with Nonce (like `InvalidNonce`), it is best to regenerate the API keys. Resetting Nonce is difficult and it's usually easier to regenerate the API keys.
@@ -278,7 +302,7 @@ For example, to test the order type `FOK` with Kraken, and modify candle limit t
Freqtrade supports spot trading, as well as (isolated) futures trading for some selected exchanges. Please refer to the [documentation start page](index.md#supported-futures-exchanges-experimental) for an uptodate list of supported exchanges.
### Can I open short positions?
### Can my bot open short positions?
Freqtrade can open short positions in futures markets.
This requires the strategy to be made for this - and `"trading_mode": "futures"` in the configuration.
@@ -12,9 +12,9 @@ Please make sure to read the [relevant documentation page](leverage.md) first.
In spot markets, you can in some cases use leveraged spot tokens, which reflect an inverted pair (eg. BTCUP/USD, BTCDOWN/USD, ETHBULL/USD, ETHBEAR/USD,...) which can be traded with Freqtrade.
### Can I trade options or futures?
### Can my bot trade options or futures?
Futures trading is supported for selected exchanges.
Futures trading is supported for selected exchanges. Please refer to the [documentation start page](index.md#supported-futures-exchanges-experimental) for an uptodate list of supported exchanges.
## Beginner Tips & Tricks
@@ -22,6 +22,13 @@ Futures trading is supported for selected exchanges.
## Freqtrade common issues
### Can freqtrade open multiple positions on the same pair in parallel?
No. Freqtrade will only open one position per pair at a time.
You can however use the [`adjust_trade_position()` callback](strategy-callbacks.md#adjust-trade-position) to adjust an open position.
Backtesting provides an option for this in `--eps` - however this is only there to highlight "hidden" signals, and will not work in live.
### The bot does not start
Running the bot with `freqtrade trade --config config.json` shows the output `freqtrade: command not found`.
@@ -30,7 +37,7 @@ This could be caused by the following reasons:
* The virtual environment is not active.
* Run `source .env/bin/activate` to activate the virtual environment.
* The installation did not work correctly.
* The installation did not complete successfully.
* Please check the [Installation documentation](installation.md).
### I have waited 5 minutes, why hasn't the bot made any trades yet?
@@ -95,6 +102,12 @@ If this happens for all pairs in the pairlist, this might indicate a recent exch
Irrespectively of the reason, Freqtrade will fill up these candles with "empty" candles, where open, high, low and close are set to the previous candle close - and volume is empty. In a chart, this will look like a `_` - and is aligned with how exchanges usually represent 0 volume candles.
### I'm getting "Price jump between 2 candles detected"
This message is a warning that the candles had a price jump of > 30%.
This might be a sign that the pair stopped trading, and some token exchange took place (e.g. COCOS in 2021 - where price jumped from 0.0000154 to 0.01621).
This message is often accompanied by ["Missing data fillup"](#im-getting-missing-data-fillup-messages-in-the-log) - as trading on such pairs is often stopped for some time.
### I'm getting "Outdated history for pair xxx" in the log
The bot is trying to tell you that it got an outdated last candle (not the last complete candle).
@@ -235,8 +248,26 @@ The Edge module is mostly a result of brainstorming of [@mishaker](https://githu
You can find further info on expectancy, win rate, risk management and position size in the following sources:
Nobody affiliated with the freqtrade project will ask you about your exchange keys or anything else exposing your funds to exploitation.
Should you be asked to expose your exchange keys or send funds to some random wallet, then please don't follow these instructions.
Failing to follow these guidelines will not be responsibility of freqtrade.
## "Freqtrade token"
Freqtrade does not have a Crypto token offering.
Token offerings you find on the internet referring Freqtrade, FreqAI or freqUI must be considered to be a scam, trying to exploit freqtrade's popularity for their own, nefarious gains.
FreqAI is configured through the typical [Freqtrade config file](configuration.md) and the standard [Freqtrade strategy](strategy-customization.md). Examples of FreqAI config and strategy files can be found in `config_examples/config_freqai.example.json` and `freqtrade/templates/FreqaiExampleStrategy.py`, respectively.
## Setting up the configuration file
Although there are plenty of additional parameters to choose from, as highlighted in the [parameter table](freqai-parameter-table.md#parameter-table), a FreqAI config must at minimum include the following parameters (the parameter values are only examples):
```json
"freqai":{
"enabled":true,
"purge_old_models":2,
"train_period_days":30,
"backtest_period_days":7,
"identifier":"unique-id",
"feature_parameters":{
"include_timeframes":["5m","15m","4h"],
"include_corr_pairlist":[
"ETH/USD",
"LINK/USD",
"BNB/USD"
],
"label_period_candles":24,
"include_shifted_candles":2,
"indicator_periods_candles":[10,20]
},
"data_split_parameters":{
"test_size":0.25
}
}
```
A full example config is available in `config_examples/config_freqai.example.json`.
## Building a FreqAI strategy
The FreqAI strategy requires including the following lines of code in the standard [Freqtrade strategy](strategy-customization.md):
```python
# user should define the maximum startup candle count (the largest number of candles
Notice how the `feature_engineering_*()` is where [features](freqai-feature-engineering.md#feature-engineering) are added. Meanwhile `set_freqai_targets()` adds the labels/targets. A full example strategy is available in `templates/FreqaiExampleStrategy.py`.
!!! Note
The `self.freqai.start()` function cannot be called outside the `populate_indicators()`.
!!! Note
Features **must** be defined in `feature_engineering_*()`. Defining FreqAI features in `populate_indicators()`
will cause the algorithm to fail in live/dry mode. In order to add generalized features that are not associated with a specific pair or timeframe, you should use `feature_engineering_standard()`
(as exemplified in `freqtrade/templates/FreqaiExampleStrategy.py`).
## Important dataframe key patterns
Below are the values you can expect to include/use inside a typical strategy dataframe (`df[]`):
| DataFrame Key | Description |
|------------|-------------|
| `df['&*']` | Any dataframe column prepended with `&` in `set_freqai_targets()` is treated as a training target (label) inside FreqAI (typically following the naming convention `&-s*`). For example, to predict the close price 40 candles into the future, you would set `df['&-s_close'] = df['close'].shift(-self.freqai_info["feature_parameters"]["label_period_candles"])` with `"label_period_candles": 40` in the config. FreqAI makes the predictions and gives them back under the same key (`df['&-s_close']`) to be used in `populate_entry/exit_trend()`. <br>**Datatype:** Depends on the output of the model.
| `df['&*_std/mean']` | Standard deviation and mean values of the defined labels during training (or live tracking with `fit_live_predictions_candles`). Commonly used to understand the rarity of a prediction (use the z-score as shown in `templates/FreqaiExampleStrategy.py` and explained [here](#creating-a-dynamic-target-threshold) to evaluate how often a particular prediction was observed during training or historically with `fit_live_predictions_candles`). <br>**Datatype:** Float.
| `df['do_predict']` | Indication of an outlier data point. The return value is integer between -2 and 2, which lets you know if the prediction is trustworthy or not. `do_predict==1` means that the prediction is trustworthy. If the Dissimilarity Index (DI, see details [here](freqai-feature-engineering.md#identifying-outliers-with-the-dissimilarity-index-di)) of the input data point is above the threshold defined in the config, FreqAI will subtract 1 from `do_predict`, resulting in `do_predict==0`. If `use_SVM_to_remove_outliers()` is active, the Support Vector Machine (SVM, see details [here](freqai-feature-engineering.md#identifying-outliers-using-a-support-vector-machine-svm)) may also detect outliers in training and prediction data. In this case, the SVM will also subtract 1 from `do_predict`. If the input data point was considered an outlier by the SVM but not by the DI, or vice versa, the result will be `do_predict==0`. If both the DI and the SVM considers the input data point to be an outlier, the result will be `do_predict==-1`. As with the SVM, if `use_DBSCAN_to_remove_outliers` is active, DBSCAN (see details [here](freqai-feature-engineering.md#identifying-outliers-with-dbscan)) may also detect outliers and subtract 1 from `do_predict`. Hence, if both the SVM and DBSCAN are active and identify a datapoint that was above the DI threshold as an outlier, the result will be `do_predict==-2`. A particular case is when `do_predict == 2`, which means that the model has expired due to exceeding `expired_hours`. <br>**Datatype:** Integer between -2 and 2.
| `df['DI_values']` | Dissimilarity Index (DI) values are proxies for the level of confidence FreqAI has in the prediction. A lower DI means the prediction is close to the training data, i.e., higher prediction confidence. See details about the DI [here](freqai-feature-engineering.md#identifying-outliers-with-the-dissimilarity-index-di). <br>**Datatype:** Float.
| `df['%*']` | Any dataframe column prepended with `%` in `feature_engineering_*()` is treated as a training feature. For example, you can include the RSI in the training feature set (similar to in `templates/FreqaiExampleStrategy.py`) by setting `df['%-rsi']`. See more details on how this is done [here](freqai-feature-engineering.md). <br>**Note:** Since the number of features prepended with `%` can multiply very quickly (10s of thousands of features are easily engineered using the multiplictative functionality of, e.g., `include_shifted_candles` and `include_timeframes` as described in the [parameter table](freqai-parameter-table.md)), these features are removed from the dataframe that is returned from FreqAI to the strategy. To keep a particular type of feature for plotting purposes, you would prepend it with `%%`. <br>**Datatype:** Depends on the output of the model.
## Setting the `startup_candle_count`
The `startup_candle_count` in the FreqAI strategy needs to be set up in the same way as in the standard Freqtrade strategy (see details [here](strategy-customization.md#strategy-startup-period)). This value is used by Freqtrade to ensure that a sufficient amount of data is provided when calling the `dataprovider`, to avoid any NaNs at the beginning of the first training. You can easily set this value by identifying the longest period (in candle units) which is passed to the indicator creation functions (e.g., TA-Lib functions). In the presented example, `startup_candle_count` is 20 since this is the maximum value in `indicators_periods_candles`.
!!! Note
There are instances where the TA-Lib functions actually require more data than just the passed `period` or else the feature dataset gets populated with NaNs. Anecdotally, multiplying the `startup_candle_count` by 2 always leads to a fully NaN free training dataset. Hence, it is typically safest to multiply the expected `startup_candle_count` by 2. Look out for this log message to confirm that the data is clean:
```
2022-08-31 15:14:04 - freqtrade.freqai.data_kitchen - INFO - dropped 0 training points due to NaNs in populated dataset 4319.
```
## Creating a dynamic target threshold
Deciding when to enter or exit a trade can be done in a dynamic way to reflect current market conditions. FreqAI allows you to return additional information from the training of a model (more info [here](freqai-feature-engineering.md#returning-additional-info-from-training)). For example, the `&*_std/mean` return values describe the statistical distribution of the target/label *during the most recent training*. Comparing a given prediction to these values allows you to know the rarity of the prediction. In `templates/FreqaiExampleStrategy.py`, the `target_roi` and `sell_roi` are defined to be 1.25 z-scores away from the mean which causes predictions that are closer to the mean to be filtered out.
To consider the population of *historical predictions* for creating the dynamic target instead of information from the training as discussed above, you would set `fit_live_predictions_candles` in the config to the number of historical prediction candles you wish to use to generate target statistics.
```json
"freqai": {
"fit_live_predictions_candles": 300,
}
```
If this value is set, FreqAI will initially use the predictions from the training data and subsequently begin introducing real prediction data as it is generated. FreqAI will save this historical data to be reloaded if you stop and restart a model with the same `identifier`.
## Using different prediction models
FreqAI has multiple example prediction model libraries that are ready to be used as is via the flag `--freqaimodel`. These libraries include `CatBoost`, `LightGBM`, and `XGBoost` regression, classification, and multi-target models, and can be found in `freqai/prediction_models/`.
Regression and classification models differ in what targets they predict - a regression model will predict a target of continuous values, for example what price BTC will be at tomorrow, whilst a classifier will predict a target of discrete values, for example if the price of BTC will go up tomorrow or not. This means that you have to specify your targets differently depending on which model type you are using (see details [below](#setting-model-targets)).
All of the aforementioned model libraries implement gradient boosted decision tree algorithms. They all work on the principle of ensemble learning, where predictions from multiple simple learners are combined to get a final prediction that is more stable and generalized. The simple learners in this case are decision trees. Gradient boosting refers to the method of learning, where each simple learner is built in sequence - the subsequent learner is used to improve on the error from the previous learner. If you want to learn more about the different model libraries you can find the information in their respective docs:
There are also numerous online articles describing and comparing the algorithms. Some relatively lightweight examples would be [CatBoost vs. LightGBM vs. XGBoost — Which is the best algorithm?](https://towardsdatascience.com/catboost-vs-lightgbm-vs-xgboost-c80f40662924#:~:text=In%20CatBoost%2C%20symmetric%20trees%2C%20or,the%20same%20depth%20can%20differ.) and [XGBoost, LightGBM or CatBoost — which boosting algorithm should I use?](https://medium.com/riskified-technology/xgboost-lightgbm-or-catboost-which-boosting-algorithm-should-i-use-e7fda7bb36bc). Keep in mind that the performance of each model is highly dependent on the application and so any reported metrics might not be true for your particular use of the model.
Apart from the models already available in FreqAI, it is also possible to customize and create your own prediction models using the `IFreqaiModel` class. You are encouraged to inherit `fit()`, `train()`, and `predict()` to customize various aspects of the training procedures. You can place custom FreqAI models in `user_data/freqaimodels` - and freqtrade will pick them up from there based on the provided `--freqaimodel` name - which has to correspond to the class name of your custom model.
Make sure to use unique names to avoid overriding built-in models.
### Setting model targets
#### Regressors
If you are using a regressor, you need to specify a target that has continuous values. FreqAI includes a variety of regressors, such as the `CatboostRegressor`via the flag `--freqaimodel CatboostRegressor`. An example of how you could set a regression target for predicting the price 100 candles into the future would be
```python
df['&s-close_price'] = df['close'].shift(-100)
```
If you want to predict multiple targets, you need to define multiple labels using the same syntax as shown above.
#### Classifiers
If you are using a classifier, you need to specify a target that has discrete values. FreqAI includes a variety of classifiers, such as the `CatboostClassifier` via the flag `--freqaimodel CatboostClassifier`. If you elects to use a classifier, the classes need to be set using strings. For example, if you want to predict if the price 100 candles into the future goes up or down you would set
If you want to predict multiple targets you must specify all labels in the same label column. You could, for example, add the label `same` to define where the price was unchanged by setting
The architecture and functions of FreqAI are generalized to encourages development of unique features, functions, models, etc.
The class structure and a detailed algorithmic overview is depicted in the following diagram:
As shown, there are three distinct objects comprising FreqAI:
* **IFreqaiModel** - A singular persistent object containing all the necessary logic to collect, store, and process data, engineer features, run training, and inference models.
* **FreqaiDataKitchen** - A non-persistent object which is created uniquely for each unique asset/model. Beyond metadata, it also contains a variety of data processing tools.
* **FreqaiDataDrawer** - A singular persistent object containing all the historical predictions, models, and save/load methods.
There are a variety of built-in [prediction models](freqai-configuration.md#using-different-prediction-models) which inherit directly from `IFreqaiModel`. Each of these models have full access to all methods in `IFreqaiModel` and can therefore override any of those functions at will. However, advanced users will likely stick to overriding `fit()`, `train()`, `predict()`, and `data_cleaning_train/predict()`.
## Data handling
FreqAI aims to organize model files, prediction data, and meta data in a way that simplifies post-processing and enhances crash resilience by automatic data reloading. The data is saved in a file structure,`user_data_dir/models/`, which contains all the data associated with the trainings and backtests. The `FreqaiDataKitchen()` relies heavily on the file structure for proper training and inferencing and should therefore not be manually modified.
### File structure
The file structure is automatically generated based on the model `identifier` set in the [config](freqai-configuration.md#setting-up-the-configuration-file). The following structure shows where the data is stored for post processing:
| Structure | Description |
|-----------|-------------|
| `config_*.json` | A copy of the model specific configuration file. |
| `historic_predictions.pkl` | A file containing all historic predictions generated during the lifetime of the `identifier` model during live deployment. `historic_predictions.pkl` is used to reload the model after a crash or a config change. A backup file is always held in case of corruption on the main file. FreqAI **automatically** detects corruption and replaces the corrupted file with the backup. |
| `pair_dictionary.json` | A file containing the training queue as well as the on disk location of the most recently trained model. |
| `sub-train-*_TIMESTAMP` | A folder containing all the files associated with a single model, such as: <br>
|| `*_metadata.json` - Metadata for the model, such as normalization max/min, expected training feature list, etc. <br>
|| `*_model.*` - The model file saved to disk for reloading from a crash. Can be `joblib` (typical boosting libs), `zip` (stable_baselines), `hd5` (keras type), etc. <br>
|| `*_pca_object.pkl` - The [Principal component analysis (PCA)](freqai-feature-engineering.md#data-dimensionality-reduction-with-principal-component-analysis) transform (if `principal_component_analysis: True` is set in the config) which will be used to transform unseen prediction features. <br>
|| `*_svm_model.pkl` - The [Support Vector Machine (SVM)](freqai-feature-engineering.md#identifying-outliers-using-a-support-vector-machine-svm) model (if `use_SVM_to_remove_outliers: True` is set in the config) which is used to detect outliers in unseen prediction features. <br>
|| `*_trained_df.pkl` - The dataframe containing all the training features used to train the `identifier` model. This is used for computing the [Dissimilarity Index (DI)](freqai-feature-engineering.md#identifying-outliers-with-the-dissimilarity-index-di) and can also be used for post-processing. <br>
|| `*_trained_dates.df.pkl` - The dates associated with the `trained_df.pkl`, which is useful for post-processing. |
Low level feature engineering is performed in the user strategy within a set of functions called `feature_engineering_*`. These function set the `base features` such as, `RSI`, `MFI`, `EMA`, `SMA`, time of day, volume, etc. The `base features` can be custom indicators or they can be imported from any technical-analysis library that you can find. FreqAI is equipped with a set of functions to simplify rapid large-scale feature engineering:
| Function | Description |
|---------------|-------------|
| `feature_engineering__expand_all()` | This optional function will automatically expand the defined features on the config defined `indicator_periods_candles`, `include_timeframes`, `include_shifted_candles`, and `include_corr_pairs`.
| `feature_engineering__expand_basic()` | This optional function will automatically expand the defined features on the config defined `include_timeframes`, `include_shifted_candles`, and `include_corr_pairs`. Note: this function does *not* expand across `include_periods_candles`.
| `feature_engineering_standard()` | This optional function will be called once with the dataframe of the base timeframe. This is the final function to be called, which means that the dataframe entering this function will contain all the features and columns from the base asset created by the other `feature_engineering_expand` functions. This function is a good place to do custom exotic feature extractions (e.g. tsfresh). This function is also a good place for any feature that should not be auto-expanded upon (e.g., day of the week).
| `set_freqai_targets()` | Required function to set the targets for the model. All targets must be prepended with `&` to be recognized by the FreqAI internals.
Meanwhile, high level feature engineering is handled within `"feature_parameters":{}` in the FreqAI config. Within this file, it is possible to decide large scale feature expansions on top of the `base_features` such as "including correlated pairs" or "including informative timeframes" or even "including recent candles."
It is advisable to start from the template `feature_engineering_*` functions in the source provided example strategy (found in `templates/FreqaiExampleStrategy.py`) to ensure that the feature definitions are following the correct conventions. Here is an example of how to set the indicators and labels in the strategy:
In the presented example, the user does not wish to pass the `bb_lowerband` as a feature to the model,
and has therefore not prepended it with `%`. The user does, however, wish to pass `bb_width` to the
model for training/prediction and has therefore prepended it with `%`.
After having defined the `base features`, the next step is to expand upon them using the powerful `feature_parameters` in the configuration file:
```json
"freqai":{
//...
"feature_parameters":{
"include_timeframes":["5m","15m","4h"],
"include_corr_pairlist":[
"ETH/USD",
"LINK/USD",
"BNB/USD"
],
"label_period_candles":24,
"include_shifted_candles":2,
"indicator_periods_candles":[10,20]
},
//...
}
```
The `include_timeframes` in the config above are the timeframes (`tf`) of each call to `feature_engineering_expand_*()` in the strategy. In the presented case, the user is asking for the `5m`, `15m`, and `4h` timeframes of the `rsi`, `mfi`, `roc`, and `bb_width` to be included in the feature set.
You can ask for each of the defined features to be included also for informative pairs using the `include_corr_pairlist`. This means that the feature set will include all the features from `feature_engineering_expand_*()` on all the `include_timeframes` for each of the correlated pairs defined in the config (`ETH/USD`, `LINK/USD`, and `BNB/USD` in the presented example).
`include_shifted_candles` indicates the number of previous candles to include in the feature set. For example, `include_shifted_candles: 2` tells FreqAI to include the past 2 candles for each of the features in the feature set.
In total, the number of features the user of the presented example strat has created is: length of `include_timeframes`* no. features in `feature_engineering_expand_*()` * length of `include_corr_pairlist` * no. `include_shifted_candles` * length of `indicator_periods_candles`
$= 3 * 3 * 3 * 2 * 2 = 108$.
### Gain finer control over `feature_engineering_*` functions with `metadata`
All `feature_engineering_*` and `set_freqai_targets()` functions are passed a `metadata` dictionary which contains information about the `pair`, `tf` (timeframe), and `period` that FreqAI is automating for feature building. As such, a user can use `metadata` inside `feature_engineering_*` functions as criteria for blocking/reserving features for certain timeframes, periods, pairs etc.
This will block `ta.ROC()` from being added to any timeframes other than `"1h"`.
### Returning additional info from training
Important metrics can be returned to the strategy at the end of each model training by assigning them to `dk.data['extra_returns_per_train']['my_new_value'] = XYZ` inside the custom prediction model class.
FreqAI takes the `my_new_value` assigned in this dictionary and expands it to fit the dataframe that is returned to the strategy. You can then use the returned metrics in your strategy through `dataframe['my_new_value']`. An example of how return values can be used in FreqAI are the `&*_mean` and `&*_std` values that are used to [created a dynamic target threshold](freqai-configuration.md#creating-a-dynamic-target-threshold).
Another example, where the user wants to use live metrics from the trade database, is shown below:
```json
"freqai": {
"extra_returns_per_train": {"total_profit": 4}
}
```
You need to set the standard dictionary in the config so that FreqAI can return proper dataframe shapes. These values will likely be overridden by the prediction model, but in the case where the model has yet to set them, or needs a default initial value, the pre-set values are what will be returned.
## Feature normalization
FreqAI is strict when it comes to data normalization. The train features, $X^{train}$, are always normalized to [-1, 1] using a shifted min-max normalization:
All other data (test data and unseen prediction data in dry/live/backtest) is always automatically normalized to the training feature space according to industry standards. FreqAI stores all the metadata required to ensure that test and prediction features will be properly normalized and that predictions are properly denormalized. For this reason, it is not recommended to eschew industry standards and modify FreqAI internals - however - advanced users can do so by inheriting `train()` in their custom `IFreqaiModel` and using their own normalization functions.
## Data dimensionality reduction with Principal Component Analysis
You can reduce the dimensionality of your features by activating the `principal_component_analysis` in the config:
```json
"freqai": {
"feature_parameters" : {
"principal_component_analysis": true
}
}
```
This will perform PCA on the features and reduce their dimensionality so that the explained variance of the data set is >= 0.999. Reducing data dimensionality makes training the model faster and hence allows for more up-to-date models.
## Inlier metric
The `inlier_metric` is a metric aimed at quantifying how similar the features of a data point are to the most recent historical data points.
You define the lookback window by setting `inlier_metric_window` and FreqAI computes the distance between the present time point and each of the previous `inlier_metric_window` lookback points. A Weibull function is fit to each of the lookback distributions and its cumulative distribution function (CDF) is used to produce a quantile for each lookback point. The `inlier_metric` is then computed for each time point as the average of the corresponding lookback quantiles. The figure below explains the concept for an `inlier_metric_window` of 5.
FreqAI adds the `inlier_metric` to the training features and hence gives the model access to a novel type of temporal information.
This function does **not** remove outliers from the data set.
## Weighting features for temporal importance
FreqAI allows you to set a `weight_factor` to weight recent data more strongly than past data via an exponential function:
$$ W_i = \exp(\frac{-i}{\alpha*n}) $$
where $W_i$ is the weight of data point $i$ in a total set of $n$ data points. Below is a figure showing the effect of different weight factors on the data points in a feature set.
## Outlier detection
Equity and crypto markets suffer from a high level of non-patterned noise in the form of outlier data points. FreqAI implements a variety of methods to identify such outliers and hence mitigate risk.
### Identifying outliers with the Dissimilarity Index (DI)
The Dissimilarity Index (DI) aims to quantify the uncertainty associated with each prediction made by the model.
You can tell FreqAI to remove outlier data points from the training/test data sets using the DI by including the following statement in the config:
```json
"freqai": {
"feature_parameters" : {
"DI_threshold": 1
}
}
```
The DI allows predictions which are outliers (not existent in the model feature space) to be thrown out due to low levels of certainty. To do so, FreqAI measures the distance between each training data point (feature vector), $X_{a}$, and all other training data points:
where $d_{ab}$ is the distance between the normalized points $a$ and $b$, and $p$ is the number of features, i.e., the length of the vector $X$. The characteristic distance, $\overline{d}$, for a set of training data points is simply the mean of the average distances:
$\overline{d}$ quantifies the spread of the training data, which is compared to the distance between a new prediction feature vectors, $X_k$ and all the training data:
$$ d_k = \arg \min d_{k,i} $$
This enables the estimation of the Dissimilarity Index as:
$$ DI_k = d_k/\overline{d} $$
You can tweak the DI through the `DI_threshold` to increase or decrease the extrapolation of the trained model. A higher `DI_threshold` means that the DI is more lenient and allows predictions further away from the training data to be used whilst a lower `DI_threshold` has the opposite effect and hence discards more predictions.
Below is a figure that describes the DI for a 3D data set.
### Identifying outliers using a Support Vector Machine (SVM)
You can tell FreqAI to remove outlier data points from the training/test data sets using a Support Vector Machine (SVM) by including the following statement in the config:
```json
"freqai": {
"feature_parameters" : {
"use_SVM_to_remove_outliers": true
}
}
```
The SVM will be trained on the training data and any data point that the SVM deems to be beyond the feature space will be removed.
FreqAI uses `sklearn.linear_model.SGDOneClassSVM` (details are available on scikit-learn's webpage [here](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.SGDOneClassSVM.html) (external website)) and you can elect to provide additional parameters for the SVM, such as `shuffle`, and `nu`.
The parameter `shuffle` is by default set to `False` to ensure consistent results. If it is set to `True`, running the SVM multiple times on the same data set might result in different outcomes due to `max_iter` being to low for the algorithm to reach the demanded `tol`. Increasing `max_iter` solves this issue but causes the procedure to take longer time.
The parameter `nu`, *very* broadly, is the amount of data points that should be considered outliers and should be between 0 and 1.
### Identifying outliers with DBSCAN
You can configure FreqAI to use DBSCAN to cluster and remove outliers from the training/test data set or incoming outliers from predictions, by activating `use_DBSCAN_to_remove_outliers` in the config:
```json
"freqai": {
"feature_parameters" : {
"use_DBSCAN_to_remove_outliers": true
}
}
```
DBSCAN is an unsupervised machine learning algorithm that clusters data without needing to know how many clusters there should be.
Given a number of data points $N$, and a distance $\varepsilon$, DBSCAN clusters the data set by setting all data points that have $N-1$ other data points within a distance of $\varepsilon$ as *core points*. A data point that is within a distance of $\varepsilon$ from a *core point* but that does not have $N-1$ other data points within a distance of $\varepsilon$ from itself is considered an *edge point*. A cluster is then the collection of *core points* and *edge points*. Data points that have no other data points at a distance $<\varepsilon$ are considered outliers. The figure below shows a cluster with $N = 3$.
FreqAI uses `sklearn.cluster.DBSCAN` (details are available on scikit-learn's webpage [here](https://scikit-learn.org/stable/modules/generated/sklearn.cluster.DBSCAN.html) (external website)) with `min_samples` ($N$) taken as 1/4 of the no. of time points (candles) in the feature set. `eps` ($\varepsilon$) is computed automatically as the elbow point in the *k-distance graph* computed from the nearest neighbors in the pairwise distances of all data points in the feature set.
The table below will list all configuration parameters available for FreqAI. Some of the parameters are exemplified in `config_examples/config_freqai.example.json`.
Mandatory parameters are marked as **Required** and have to be set in one of the suggested ways.
### General configuration parameters
| Parameter | Description |
|------------|-------------|
| | **General configuration parameters within the `config.freqai` tree**
| `freqai` | **Required.**<br> The parent dictionary containing all the parameters for controlling FreqAI. <br>**Datatype:** Dictionary.
| `train_period_days` | **Required.**<br> Number of days to use for the training data (width of the sliding window). <br>**Datatype:** Positive integer.
| `backtest_period_days` | **Required.**<br> Number of days to inference from the trained model before sliding the `train_period_days` window defined above, and retraining the model during backtesting (more info [here](freqai-running.md#backtesting)). This can be fractional days, but beware that the provided `timerange` will be divided by this number to yield the number of trainings necessary to complete the backtest. <br>**Datatype:** Float.
| `identifier` | **Required.**<br> A unique ID for the current model. If models are saved to disk, the `identifier` allows for reloading specific pre-trained models/data. <br>**Datatype:** String.
| `live_retrain_hours` | Frequency of retraining during dry/live runs. <br>**Datatype:** Float > 0. <br> Default: `0` (models retrain as often as possible).
| `expiration_hours` | Avoid making predictions if a model is more than `expiration_hours` old. <br>**Datatype:** Positive integer. <br> Default: `0` (models never expire).
| `purge_old_models` | Number of models to keep on disk (not relevant to backtesting). Default is 2, which means that dry/live runs will keep the latest 2 models on disk. Setting to 0 keeps all models. This parameter also accepts a boolean to maintain backwards compatibility. <br>**Datatype:** Integer. <br> Default: `2`.
| `save_backtest_models` | Save models to disk when running backtesting. Backtesting operates most efficiently by saving the prediction data and reusing them directly for subsequent runs (when you wish to tune entry/exit parameters). Saving backtesting models to disk also allows to use the same model files for starting a dry/live instance with the same model `identifier`. <br>**Datatype:** Boolean. <br> Default: `False` (no models are saved).
| `fit_live_predictions_candles` | Number of historical candles to use for computing target (label) statistics from prediction data, instead of from the training dataset (more information can be found [here](freqai-configuration.md#creating-a-dynamic-target-threshold)). <br>**Datatype:** Positive integer.
| `continual_learning` | Use the final state of the most recently trained model as starting point for the new model, allowing for incremental learning (more information can be found [here](freqai-running.md#continual-learning)). <br>**Datatype:** Boolean. <br> Default: `False`.
| `write_metrics_to_disk` | Collect train timings, inference timings and cpu usage in json file. <br>**Datatype:** Boolean. <br> Default: `False`
| `data_kitchen_thread_count` | <br> Designate the number of threads you want to use for data processing (outlier methods, normalization, etc.). This has no impact on the number of threads used for training. If user does not set it (default), FreqAI will use max number of threads - 2 (leaving 1 physical core available for Freqtrade bot and FreqUI) <br>**Datatype:** Positive integer.
### Feature parameters
| Parameter | Description |
|------------|-------------|
| | **Feature parameters within the `freqai.feature_parameters` sub dictionary**
| `feature_parameters` | A dictionary containing the parameters used to engineer the feature set. Details and examples are shown [here](freqai-feature-engineering.md). <br>**Datatype:** Dictionary.
| `include_timeframes` | A list of timeframes that all indicators in `feature_engineering_expand_*()` will be created for. The list is added as features to the base indicators dataset. <br>**Datatype:** List of timeframes (strings).
| `include_corr_pairlist` | A list of correlated coins that FreqAI will add as additional features to all `pair_whitelist` coins. All indicators set in `feature_engineering_expand_*()` during feature engineering (see details [here](freqai-feature-engineering.md)) will be created for each correlated coin. The correlated coins features are added to the base indicators dataset. <br>**Datatype:** List of assets (strings).
| `label_period_candles` | Number of candles into the future that the labels are created for. This is used in `feature_engineering_expand_all()` (see `templates/FreqaiExampleStrategy.py` for detailed usage). You can create custom labels and choose whether to make use of this parameter or not. <br>**Datatype:** Positive integer.
| `include_shifted_candles` | Add features from previous candles to subsequent candles with the intent of adding historical information. If used, FreqAI will duplicate and shift all features from the `include_shifted_candles` previous candles so that the information is available for the subsequent candle. <br>**Datatype:** Positive integer.
| `weight_factor` | Weight training data points according to their recency (see details [here](freqai-feature-engineering.md#weighting-features-for-temporal-importance)). <br>**Datatype:** Positive float (typically <1).
|`indicator_max_period_candles`|**No longer used (#7325)**.Replacedby`startup_candle_count`whichissetinthe [strategy](freqai-configuration.md#building-a-freqai-strategy).`startup_candle_count`istimeframeindependentanddefinesthemaximum*period*usedin`feature_engineering_*()`forindicatorcreation.FreqAIusesthisparametertogetherwiththemaximumtimeframein`include_time_frames`tocalculatehowmanydatapointstodownloadsuchthatthefirstdatapointdoesnotincludeaNaN.<br>**Datatype:** Positive integer.
| `indicator_periods_candles` | Time periods to calculate indicators for. The indicators are added to the base indicator dataset. <br>**Datatype:** List of positive integers.
| `principal_component_analysis` | Automatically reduce the dimensionality of the data set using Principal Component Analysis. See details about how it works [here](#reducing-data-dimensionality-with-principal-component-analysis) <br>**Datatype:** Boolean. <br> Default: `False`.
| `plot_feature_importances` | Create a feature importance plot for each model for the top/bottom `plot_feature_importances` number of features. Plot is stored in `user_data/models/<identifier>/sub-train-<COIN>_<timestamp>.html`. <br>**Datatype:** Integer. <br> Default: `0`.
| `DI_threshold` | Activates the use of the Dissimilarity Index for outlier detection when set to > 0. See details about how it works [here](freqai-feature-engineering.md#identifying-outliers-with-the-dissimilarity-index-di). <br>**Datatype:** Positive float (typically <1).
| `svm_params` | All parameters available in Sklearn's `SGDOneClassSVM()`. See details about some select parameters [here](freqai-feature-engineering.md#identifying-outliers-using-a-support-vector-machine-svm). <br>**Datatype:** Dictionary.
| `use_DBSCAN_to_remove_outliers` | Cluster data using the DBSCAN algorithm to identify and remove outliers from training and prediction data. See details about how it works [here](freqai-feature-engineering.md#identifying-outliers-with-dbscan). <br>**Datatype:** Boolean.
| `inlier_metric_window` | If set, FreqAI adds an `inlier_metric` to the training feature set and set the lookback to be the `inlier_metric_window`, i.e., the number of previous time points to compare the current candle to. Details of how the `inlier_metric` is computed can be found [here](freqai-feature-engineering.md#inlier-metric). <br>**Datatype:** Integer. <br> Default: `0`.
| `noise_standard_deviation` | If set, FreqAI adds noise to the training features with the aim of preventing overfitting. FreqAI generates random deviates from a gaussian distribution with a standard deviation of `noise_standard_deviation` and adds them to all data points. `noise_standard_deviation` should be kept relative to the normalized space, i.e., between -1 and 1. In other words, since data in FreqAI is always normalized to be between -1 and 1, `noise_standard_deviation: 0.05` would result in 32% of the data being randomly increased/decreased by more than 2.5% (i.e., the percent of data falling within the first standard deviation). <br>**Datatype:** Integer. <br> Default: `0`.
| `outlier_protection_percentage` | Enable to prevent outlier detection methods from discarding too much data. If more than `outlier_protection_percentage` % of points are detected as outliers by the SVM or DBSCAN, FreqAI will log a warning message and ignore outlier detection, i.e., the original dataset will be kept intact. If the outlier protection is triggered, no predictions will be made based on the training dataset. <br>**Datatype:** Float. <br> Default: `30`.
| `reverse_train_test_order` | Split the feature dataset (see below) and use the latest data split for training and test on historical split of the data. This allows the model to be trained up to the most recent data point, while avoiding overfitting. However, you should be careful to understand the unorthodox nature of this parameter before employing it. <br>**Datatype:** Boolean. <br> Default: `False` (no reversal).
| `shuffle_after_split` | Split the data into train and test sets, and then shuffle both sets individually. <br>**Datatype:** Boolean. <br> Default: `False`.
| `buffer_train_data_candles` | Cut `buffer_train_data_candles` off the beginning and end of the training data *after* the indicators were populated. The main example use is when predicting maxima and minima, the argrelextrema function cannot know the maxima/minima at the edges of the timerange. To improve model accuracy, it is best to compute argrelextrema on the full timerange and then use this function to cut off the edges (buffer) by the kernel. In another case, if the targets are set to a shifted price movement, this buffer is unnecessary because the shifted candles at the end of the timerange will be NaN and FreqAI will automatically cut those off of the training dataset.<br>**Datatype:** Boolean. <br> Default: `False`.
### Data split parameters
| Parameter | Description |
|------------|-------------|
| | **Data split parameters within the `freqai.data_split_parameters` sub dictionary**
| `data_split_parameters` | Include any additional parameters available from scikit-learn `test_train_split()`, which are shown [here](https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.train_test_split.html) (external website). <br>**Datatype:** Dictionary.
| `test_size` | The fraction of data that should be used for testing instead of training. <br>**Datatype:** Positive float <1.
| | **Model training parameters within the `freqai.model_training_parameters` sub dictionary**
| `model_training_parameters` | A flexible dictionary that includes all parameters available by the selected model library. For example, if you use `LightGBMRegressor`, this dictionary can contain any parameter available by the `LightGBMRegressor` [here](https://lightgbm.readthedocs.io/en/latest/pythonapi/lightgbm.LGBMRegressor.html) (external website). If you select a different model, this dictionary can contain any parameter from that model. A list of the currently available models can be found [here](freqai-configuration.md#using-different-prediction-models). <br>**Datatype:** Dictionary.
| `n_estimators` | The number of boosted trees to fit in the training of the model. <br>**Datatype:** Integer.
| `learning_rate` | Boosting learning rate during training of the model. <br>**Datatype:** Float.
| `n_jobs`, `thread_count`, `task_type` | Set the number of threads for parallel processing and the `task_type` (`gpu` or `cpu`). Different model libraries use different parameter names. <br>**Datatype:** Float.
### Reinforcement Learning parameters
| Parameter | Description |
|------------|-------------|
| | **Reinforcement Learning Parameters within the `freqai.rl_config` sub dictionary**
| `rl_config` | A dictionary containing the control parameters for a Reinforcement Learning model. <br>**Datatype:** Dictionary.
| `train_cycles` | Training time steps will be set based on the `train_cycles * number of training data points. <br> **Datatype:** Integer.
| `cpu_count` | Number of processors to dedicate to the Reinforcement Learning training process. <br> **Datatype:** int.
| `max_trade_duration_candles`| Guides the agent training to keep trades below desired length. Example usage shown in `prediction_models/ReinforcementLearner.py` within the customizable `calculate_reward()` function. <br> **Datatype:** int.
| `model_type` | Model string from stable_baselines3 or SBcontrib. Available strings include: `'TRPO', 'ARS', 'RecurrentPPO', 'MaskablePPO', 'PPO', 'A2C', 'DQN'`. User should ensure that `model_training_parameters` match those available to the corresponding stable_baselines3 model by visiting their documentaiton. [PPO doc](https://stable-baselines3.readthedocs.io/en/master/modules/ppo.html) (external website) <br> **Datatype:** string.
| `policy_type` | One of the available policy types from stable_baselines3 <br> **Datatype:** string.
| `max_training_drawdown_pct` | The maximum drawdown that the agent is allowed to experience during training. <br> **Datatype:** float. <br> Default: 0.8
| `cpu_count` | Number of threads/cpus to dedicate to the Reinforcement Learning training process (depending on if `ReinforcementLearning_multiproc` is selected or not). Recommended to leave this untouched, by default, this value is set to the total number of physical cores minus 1. <br> **Datatype:** int.
| `model_reward_parameters` | Parameters used inside the customizable `calculate_reward()` function in `ReinforcementLearner.py` <br> **Datatype:** int.
| `add_state_info` | Tell FreqAI to include state information in the feature set for training and inferencing. The current state variables include trade duration, current profit, trade position. This is only available in dry/live runs, and is automatically switched to false for backtesting. <br> **Datatype:** bool. <br> Default: `False`.
| `net_arch` | Network architecture which is well described in [`stable_baselines3` doc](https://stable-baselines3.readthedocs.io/en/master/guide/custom_policy.html#examples). In summary: `[<sharedlayers>, dict(vf=[<non-sharedvaluenetworklayers>], pi=[<non-sharedpolicynetworklayers>])]`. By default this is set to `[128, 128]`, which defines 2 shared hidden layers with 128 units each.
| `randomize_starting_position` | Randomize the starting point of each episode to avoid overfitting. <br> **Datatype:** bool. <br> Default: `False`.
### Additional parameters
| Parameter | Description |
|------------|-------------|
| | **Extraneous parameters**
| `freqai.keras` | If the selected model makes use of Keras (typical for TensorFlow-based prediction models), this flag needs to be activated so that the model save/loading follows Keras standards. <br> **Datatype:** Boolean. <br> Default: `False`.
| `freqai.conv_width` | The width of a convolutional neural network input tensor. This replaces the need for shifting candles (`include_shifted_candles`) by feeding in historical data points as the second dimension of the tensor. Technically, this parameter can also be used for regressors, but it only adds computational overhead and does not change the model training/prediction. <br> **Datatype:** Integer. <br> Default: `2`.
| `freqai.reduce_df_footprint` | Recast all numeric columns to float32/int32, with the objective of reducing ram/disk usage and decreasing train/inference timing. This parameter is set in the main level of the Freqtrade configuration file (not inside FreqAI). <br> **Datatype:** Boolean. <br> Default: `False`.
Reinforcement learning dependencies include large packages such as `torch`, which should be explicitly requested during `./setup.sh -i` by answering "y" to the question "Do you also want dependencies for freqai-rl (~700mb additional space required) [y/N]?".
Users who prefer docker should ensure they use the docker image appended with `_freqairl`.
## Background and terminology
### What is RL and why does FreqAI need it?
Reinforcement learning involves two important components, the *agent* and the training *environment*. During agent training, the agent moves through historical data candle by candle, always making 1 of a set of actions: Long entry, long exit, short entry, short exit, neutral). During this training process, the environment tracks the performance of these actions and rewards the agent according to a custom user made `calculate_reward()` (here we offer a default reward for users to build on if they wish [details here](#creating-a-custom-reward-function)). The reward is used to train weights in a neural network.
A second important component of the FreqAI RL implementation is the use of *state* information. State information is fed into the network at each step, including current profit, current position, and current trade duration. These are used to train the agent in the training environment, and to reinforce the agent in dry/live (this functionality is not available in backtesting). *FreqAI + Freqtrade is a perfect match for this reinforcing mechanism since this information is readily available in live deployments.*
Reinforcement learning is a natural progression for FreqAI, since it adds a new layer of adaptivity and market reactivity that Classifiers and Regressors cannot match. However, Classifiers and Regressors have strengths that RL does not have such as robust predictions. Improperly trained RL agents may find "cheats" and "tricks" to maximize reward without actually winning any trades. For this reason, RL is more complex and demands a higher level of understanding than typical Classifiers and Regressors.
### The RL interface
With the current framework, we aim to expose the training environment via the common "prediction model" file, which is a user inherited `BaseReinforcementLearner` object (e.g. `freqai/prediction_models/ReinforcementLearner`). Inside this user class, the RL environment is available and customized via `MyRLEnv` as [shown below](#creating-a-custom-reward-function).
We envision the majority of users focusing their effort on creative design of the `calculate_reward()` function [details here](#creating-a-custom-reward-function), while leaving the rest of the environment untouched. Other users may not touch the environment at all, and they will only play with the configuration settings and the powerful feature engineering that already exists in FreqAI. Meanwhile, we enable advanced users to create their own model classes entirely.
The framework is built on stable_baselines3 (torch) and OpenAI gym for the base environment class. But generally speaking, the model class is well isolated. Thus, the addition of competing libraries can be easily integrated into the existing framework. For the environment, it is inheriting from `gym.env` which means that it is necessary to write an entirely new environment in order to switch to a different library.
### Important considerations
As explained above, the agent is "trained" in an artificial trading "environment". In our case, that environment may seem quite similar to a real Freqtrade backtesting environment, but it is *NOT*. In fact, the RL training environment is much more simplified. It does not incorporate any of the complicated strategy logic, such as callbacks like `custom_exit`, `custom_stoploss`, leverage controls, etc. The RL environment is instead a very "raw" representation of the true market, where the agent has free will to learn the policy (read: stoploss, take profit, etc.) which is enforced by the `calculate_reward()`. Thus, it is important to consider that the agent training environment is not identical to the real world.
## Running Reinforcement Learning
Setting up and running a Reinforcement Learning model is the same as running a Regressor or Classifier. The same two flags, `--freqaimodel` and `--strategy`, must be defined on the command line:
where `ReinforcementLearner` will use the templated `ReinforcementLearner` from `freqai/prediction_models/ReinforcementLearner` (or a custom user defined one located in `user_data/freqaimodels`). The strategy, on the other hand, follows the same base [feature engineering](freqai-feature-engineering.md) with `feature_engineering_*` as a typical Regressor. The difference lies in the creation of the targets, Reinforcement Learning doesn't require them. However, FreqAI requires a default (neutral) value to be set in the action column:
```python
defset_freqai_targets(self,dataframe,**kwargs):
"""
*Only functional with FreqAI enabled strategies*
Required function to set the targets for the model.
All targets must be prepended with `&` to be recognized by the FreqAI internals.
# For RL, there are no direct targets to set. This is filler (neutral)
# until the agent sends an action.
dataframe["&-action"]=0
```
Most of the function remains the same as for typical Regressors, however, the function above shows how the strategy must pass the raw price data to the agent so that it has access to raw OHLCV in the training environment:
# The following features are necessary for RL models
dataframe[f"%-raw_close"]=dataframe["close"]
dataframe[f"%-raw_open"]=dataframe["open"]
dataframe[f"%-raw_high"]=dataframe["high"]
dataframe[f"%-raw_low"]=dataframe["low"]
```
Finally, there is no explicit "label" to make - instead it is necessary to assign the `&-action` column which will contain the agent's actions when accessed in `populate_entry/exit_trends()`. In the present example, the neutral action to 0. This value should align with the environment used. FreqAI provides two environments, both use 0 as the neutral action.
After users realize there are no labels to set, they will soon understand that the agent is making its "own" entry and exit decisions. This makes strategy construction rather simple. The entry and exit signals come from the agent in the form of an integer - which are used directly to decide entries and exits in the strategy:
It is important to consider that `&-action` depends on which environment they choose to use. The example above shows 5 actions, where 0 is neutral, 1 is enter long, 2 is exit long, 3 is enter short and 4 is exit short.
## Configuring the Reinforcement Learner
In order to configure the `Reinforcement Learner` the following dictionary must exist in the `freqai` config:
```json
"rl_config":{
"train_cycles":25,
"add_state_info":true,
"max_trade_duration_candles":300,
"max_training_drawdown_pct":0.02,
"cpu_count":8,
"model_type":"PPO",
"policy_type":"MlpPolicy",
"model_reward_parameters":{
"rr":1,
"profit_aim":0.025
}
}
```
Parameter details can be found [here](freqai-parameter-table.md), but in general the `train_cycles` decides how many times the agent should cycle through the candle data in its artificial environment to train weights in the model. `model_type` is a string which selects one of the available models in [stable_baselines](https://stable-baselines3.readthedocs.io/en/master/)(external link).
!!! Note
If you would like to experiment with `continual_learning`, then you should set that value to `true` in the main `freqai` configuration dictionary. This will tell the Reinforcement Learning library to continue training new models from the final state of previous models, instead of retraining new models from scratch each time a retrain is initiated.
!!! Note
Remember that the general `model_training_parameters` dictionary should contain all the model hyperparameter customizations for the particular `model_type`. For example, `PPO` parameters can be found [here](https://stable-baselines3.readthedocs.io/en/master/modules/ppo.html).
## Creating a custom reward function
As you begin to modify the strategy and the prediction model, you will quickly realize some important differences between the Reinforcement Learner and the Regressors/Classifiers. Firstly, the strategy does not set a target value (no labels!). Instead, you set the `calculate_reward()` function inside the `MyRLEnv` class (see below). A default `calculate_reward()` is provided inside `prediction_models/ReinforcementLearner.py` to demonstrate the necessary building blocks for creating rewards, but users are encouraged to create their own custom reinforcement learning model class (see below) and save it to `user_data/freqaimodels`. It is inside the `calculate_reward()` where creative theories about the market can be expressed. For example, you can reward your agent when it makes a winning trade, and penalize the agent when it makes a losing trade. Or perhaps, you wish to reward the agent for entering trades, and penalize the agent for sitting in trades too long. Below we show examples of how these rewards are all calculated:
Reinforcement Learning models benefit from tracking training metrics. FreqAI has integrated Tensorboard to allow users to track training and evaluation performance across all coins and across all retrainings. Tensorboard is activated via the following command:
```bash
cd freqtrade
tensorboard --logdir user_data/models/unique-id
```
where `unique-id` is the `identifier` set in the `freqai` configuration file. This command must be run in a separate shell to view the output in their browser at 127.0.0.1:6006 (6006 is the default port used by Tensorboard).
### Custom logging
FreqAI also provides a built in episodic summary logger called `self.tensorboard_log` for adding custom information to the Tensorboard log. By default, this function is already called once per step inside the environment to record the agent actions. All values accumulated for all steps in a single episode are reported at the conclusion of each episode, followed by a full reset of all metrics to 0 in preparation for the subsequent episode.
`self.tensorboard_log` can also be used anywhere inside the environment, for example, it can be added to the `calculate_reward` function to collect more detailed information about how often various parts of the reward were called:
```py
classMyRLEnv(Base5ActionRLEnv):
"""
User made custom environment. This class inherits from BaseEnvironment and gym.env.
Users can override any functions from those parent classes. Here is an example
of a user customized `calculate_reward()` function.
"""
defcalculate_reward(self,action:int)->float:
ifnotself._is_valid(action):
self.tensorboard_log("is_valid")
return-2
```
!!! Note
The `self.tensorboard_log()` function is designed for tracking incremented objects only i.e. events, actions inside the training environment. If the event of interest is a float, the float can be passed as the second argument e.g. `self.tensorboard_log("float_metric1", 0.23)` would add 0.23 to `float_metric`. In this case you can also disable incrementing using `inc=False` parameter.
### Choosing a base environment
FreqAI provides three base environments, `Base3ActionRLEnvironment`, `Base4ActionEnvironment` and `Base5ActionEnvironment`. As the names imply, the environments are customized for agents that can select from 3, 4 or 5 actions. The `Base3ActionEnvironment` is the simplest, the agent can select from hold, long, or short. This environment can also be used for long-only bots (it automatically follows the `can_short` flag from the strategy), where long is the enter condition and short is the exit condition. Meanwhile, in the `Base4ActionEnvironment`, the agent can enter long, enter short, hold neutral, or exit position. Finally, in the `Base5ActionEnvironment`, the agent has the same actions as Base4, but instead of a single exit action, it separates exit long and exit short. The main changes stemming from the environment selection include:
* the actions available in the `calculate_reward`
* the actions consumed by the user strategy
All of the FreqAI provided environments inherit from an action/position agnostic environment object called the `BaseEnvironment`, which contains all shared logic. The architecture is designed to be easily customized. The simplest customization is the `calculate_reward()` (see details [here](#creating-a-custom-reward-function)). However, the customizations can be further extended into any of the functions inside the environment. You can do this by simply overriding those functions inside your `MyRLEnv` in the prediction model file. Or for more advanced customizations, it is encouraged to create an entirely new environment inherited from `BaseEnvironment`.
!!! Note
Only the `Base3ActionRLEnv` can do long-only training/trading (set the user strategy attribute `can_short = False`).
There are two ways to train and deploy an adaptive machine learning model - live deployment and historical backtesting. In both cases, FreqAI runs/simulates periodic retraining of models as shown in the following figure:
## Live deployments
FreqAI can be run dry/live using the following command:
When launched, FreqAI will start training a new model, with a new `identifier`, based on the config settings. Following training, the model will be used to make predictions on incoming candles until a new model is available. New models are typically generated as often as possible, with FreqAI managing an internal queue of the coin pairs to try to keep all models equally up to date. FreqAI will always use the most recently trained model to make predictions on incoming live data. If you do not want FreqAI to retrain new models as often as possible, you can set `live_retrain_hours` to tell FreqAI to wait at least that number of hours before training a new model. Additionally, you can set `expired_hours` to tell FreqAI to avoid making predictions on models that are older than that number of hours.
Trained models are by default saved to disk to allow for reuse during backtesting or after a crash. You can opt to [purge old models](#purging-old-model-data) to save disk space by setting `"purge_old_models": true` in the config.
To start a dry/live run from a saved backtest model (or from a previously crashed dry/live session), you only need to specify the `identifier` of the specific model:
```json
"freqai":{
"identifier":"example",
"live_retrain_hours":0.5
}
```
In this case, although FreqAI will initiate with a pre-trained model, it will still check to see how much time has elapsed since the model was trained. If a full `live_retrain_hours` has elapsed since the end of the loaded model, FreqAI will start training a new model.
### Automatic data download
FreqAI automatically downloads the proper amount of data needed to ensure training of a model through the defined `train_period_days` and `startup_candle_count` (see the [parameter table](freqai-parameter-table.md) for detailed descriptions of these parameters).
### Saving prediction data
All predictions made during the lifetime of a specific `identifier` model are stored in `historic_predictions.pkl` to allow for reloading after a crash or changes made to the config.
### Purging old model data
FreqAI stores new model files after each successful training. These files become obsolete as new models are generated to adapt to new market conditions. If you are planning to leave FreqAI running for extended periods of time with high frequency retraining, you should enable `purge_old_models` in the config:
```json
"freqai":{
"purge_old_models":true,
}
```
This will automatically purge all models older than the two most recently trained ones to save disk space.
## Backtesting
The FreqAI backtesting module can be executed with the following command:
If this command has never been executed with the existing config file, FreqAI will train a new model
for each pair, for each backtesting window within the expanded `--timerange`.
Backtesting mode requires [downloading the necessary data](#downloading-data-to-cover-the-full-backtest-period) before deployment (unlike in dry/live mode where FreqAI handles the data downloading automatically). You should be careful to consider that the time range of the downloaded data is more than the backtesting time range. This is because FreqAI needs data prior to the desired backtesting time range in order to train a model to be ready to make predictions on the first candle of the set backtesting time range. More details on how to calculate the data to download can be found [here](#deciding-the-size-of-the-sliding-training-window-and-backtesting-duration).
!!! Note "Model reuse"
Once the training is completed, you can execute the backtesting again with the same config file and
FreqAI will find the trained models and load them instead of spending time training. This is useful
if you want to tweak (or even hyperopt) buy and sell criteria inside the strategy. If you
*want* to retrain a new model with the same config file, you should simply change the `identifier`.
This way, you can return to using any model you wish by simply specifying the `identifier`.
!!! Note
Backtesting calls `set_freqai_targets()` one time for each backtest window (where the number of windows is the full backtest timerange divided by the `backtest_period_days` parameter). Doing this means that the targets simulate dry/live behavior without look ahead bias. However, the definition of the features in `feature_engineering_*()` is performed once on the entire backtest timerange. This means that you should be sure that features do look-ahead into the future.
More details about look-ahead bias can be found in [Common Mistakes](strategy-customization.md#common-mistakes-when-developing-strategies).
---
### Saving prediction data
To allow for tweaking your strategy (**not** the features!), FreqAI will automatically save the predictions during backtesting so that they can be reused for future backtests and live runs using the same `identifier` model. This provides a performance enhancement geared towards enabling **high-level hyperopting** of entry/exit criteria.
An additional directory called `backtesting_predictions`, which contains all the predictions stored in `hdf` format, will be created in the `unique-id` folder.
To change your **features**, you **must** set a new `identifier` in the config to signal to FreqAI to train new models.
To save the models generated during a particular backtest so that you can start a live deployment from one of them instead of training a new model, you must set `save_backtest_models` to `True` in the config.
### Backtest live collected predictions
FreqAI allow you to reuse live historic predictions through the backtest parameter `--freqai-backtest-live-models`. This can be useful when you want to reuse predictions generated in dry/run for comparison or other study.
The `--timerange` parameter must not be informed, as it will be automatically calculated through the data in the historic predictions file.
### Downloading data to cover the full backtest period
For live/dry deployments, FreqAI will download the necessary data automatically. However, to use backtesting functionality, you need to download the necessary data using `download-data` (details [here](data-download.md#data-downloading)). You need to pay careful attention to understanding how much *additional* data needs to be downloaded to ensure that there is a sufficient amount of training data *before* the start of the backtesting time range. The amount of additional data can be roughly estimated by moving the start date of the time range backwards by `train_period_days` and the `startup_candle_count` (see the [parameter table](freqai-parameter-table.md) for detailed descriptions of these parameters) from the beginning of the desired backtesting time range.
As an example, to backtest the `--timerange 20210501-20210701` using the [example config](freqai-configuration.md#setting-up-the-configuration-file) which sets `train_period_days` to 30, together with `startup_candle_count: 40` on a maximum `include_timeframes` of 1h, the start date for the downloaded data needs to be `20210501` - 30 days - 40 * 1h / 24 hours = 20210330 (31.7 days earlier than the start of the desired training time range).
### Deciding the size of the sliding training window and backtesting duration
The backtesting time range is defined with the typical `--timerange` parameter in the configuration file. The duration of the sliding training window is set by `train_period_days`, whilst `backtest_period_days` is the sliding backtesting window, both in number of days (`backtest_period_days` can be
a float to indicate sub-daily retraining in live/dry mode). In the presented [example config](freqai-configuration.md#setting-up-the-configuration-file) (found in `config_examples/config_freqai.example.json`), the user is asking FreqAI to use a training period of 30 days and backtest on the subsequent 7 days. After the training of the model, FreqAI will backtest the subsequent 7 days. The "sliding window" then moves one week forward (emulating FreqAI retraining once per week in live mode) and the new model uses the previous 30 days (including the 7 days used for backtesting by the previous model) to train. This is repeated until the end of `--timerange`. This means that if you set `--timerange 20210501-20210701`, FreqAI will have trained 8 separate models at the end of `--timerange` (because the full range comprises 8 weeks).
!!! Note
Although fractional `backtest_period_days` is allowed, you should be aware that the `--timerange` is divided by this value to determine the number of models that FreqAI will need to train in order to backtest the full range. For example, by setting a `--timerange` of 10 days, and a `backtest_period_days` of 0.1, FreqAI will need to train 100 models per pair to complete the full backtest. Because of this, a true backtest of FreqAI adaptive training would take a *very* long time. The best way to fully test a model is to run it dry and let it train constantly. In this case, backtesting would take the exact same amount of time as a dry run.
## Defining model expirations
During dry/live mode, FreqAI trains each coin pair sequentially (on separate threads/GPU from the main Freqtrade bot). This means that there is always an age discrepancy between models. If you are training on 50 pairs, and each pair requires 5 minutes to train, the oldest model will be over 4 hours old. This may be undesirable if the characteristic time scale (the trade duration target) for a strategy is less than 4 hours. You can decide to only make trade entries if the model is less than a certain number of hours old by setting the `expiration_hours` in the config file:
```json
"freqai":{
"expiration_hours":0.5,
}
```
In the presented example config, the user will only allow predictions on models that are less than 1/2 hours old.
## Controlling the model learning process
Model training parameters are unique to the selected machine learning library. FreqAI allows you to set any parameter for any library using the `model_training_parameters` dictionary in the config. The example config (found in `config_examples/config_freqai.example.json`) shows some of the example parameters associated with `Catboost` and `LightGBM`, but you can add any parameters available in those libraries or any other machine learning library you choose to implement.
Data split parameters are defined in `data_split_parameters` which can be any parameters associated with scikit-learn's `train_test_split()` function. `train_test_split()` has a parameters called `shuffle` which allows to shuffle the data or keep it unshuffled. This is particularly useful to avoid biasing training with temporally auto-correlated data. More details about these parameters can be found the [scikit-learn website](https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.train_test_split.html) (external website).
The FreqAI specific parameter `label_period_candles` defines the offset (number of candles into the future) used for the `labels`. In the presented [example config](freqai-configuration.md#setting-up-the-configuration-file), the user is asking for `labels` that are 24 candles in the future.
## Continual learning
You can choose to adopt a continual learning scheme by setting `"continual_learning": true` in the config. By enabling `continual_learning`, after training an initial model from scratch, subsequent trainings will start from the final model state of the preceding training. This gives the new model a "memory" of the previous state. By default, this is set to `False` which means that all new models are trained from scratch, without input from previous models.
## Hyperopt
You can hyperopt using the same command as for [typical Freqtrade hyperopt](hyperopt.md):
`hyperopt` requires you to have the data pre-downloaded in the same fashion as if you were doing [backtesting](#backtesting). In addition, you must consider some restrictions when trying to hyperopt FreqAI strategies:
- The `--analyze-per-epoch` hyperopt parameter is not compatible with FreqAI.
- It's not possible to hyperopt indicators in the `feature_engineering_*()` and `set_freqai_targets()` functions. This means that you cannot optimize model parameters using hyperopt. Apart from this exception, it is possible to optimize all other [spaces](hyperopt.md#running-hyperopt-with-smaller-search-space).
- The backtesting instructions also apply to hyperopt.
The best method for combining hyperopt and FreqAI is to focus on hyperopting entry/exit thresholds/criteria. You need to focus on hyperopting parameters that are not used in your features. For example, you should not try to hyperopt rolling window lengths in the feature creation, or any part of the FreqAI config which changes predictions. In order to efficiently hyperopt the FreqAI strategy, FreqAI stores predictions as dataframes and reuses them. Hence the requirement to hyperopt entry/exit thresholds/criteria only.
A good example of a hyperoptable parameter in FreqAI is a threshold for the [Dissimilarity Index (DI)](freqai-feature-engineering.md#identifying-outliers-with-the-dissimilarity-index-di) `DI_values` beyond which we consider data points as outliers:
This specific hyperopt would help you understand the appropriate `DI_values` for your particular parameter space.
## Using Tensorboard
CatBoost models benefit from tracking training metrics via Tensorboard. You can take advantage of the FreqAI integration to track training and evaluation performance across all coins and across all retrainings. Tensorboard is activated via the following command:
```bash
cd freqtrade
tensorboard --logdir user_data/models/unique-id
```
where `unique-id` is the `identifier` set in the `freqai` configuration file. This command must be run in a separate shell if you wish to view the output in your browser at 127.0.0.1:6060 (6060 is the default port used by Tensorboard).
FreqAI is a module designed to automate a variety of tasks associated with training a predictive machine learning model to generate market forecasts given a set of input features.
## Introduction
FreqAI is a software designed to automate a variety of tasks associated with training a predictive machine learning model to generate market forecasts given a set of input signals. In general, FreqAI aims to be a sandbox for easily deploying robust machine learning libraries on real-time data ([details](#freqai-position-in-open-source-machine-learning-landscape)).
!!! Note
FreqAI is, and always will be, a not-for-profit, open-source project. FreqAI does *not* have a crypto token, FreqAI does *not* sell signals, and FreqAI does not have a domain besides the present [freqtrade documentation](https://www.freqtrade.io/en/latest/freqai/).
Features include:
* **Self-adaptive retraining**: retrain models during [live deployments](#running-the-model-live) to self-adapt to the market in an unsupervised manner.
* **Rapid feature engineering**: create large rich [feature sets](#feature-engineering) (10k+ features) based on simple user-created strategies.
* **High performance**: adaptive retraining occurs on a separate thread (or on GPU if available) from inferencing and bot trade operations. Newest models and data are kept in memory for rapid inferencing.
* **Realistic backtesting**: emulate self-adaptive retraining with a [backtesting module](#backtesting) that automates past retraining.
* **Modifiability**: use the generalized and robust architecture for incorporating any [machine learning library/method](#building-a-custom-prediction-model) available in Python. Eight examples are currently available, including classifiers, regressors, and a convolutional neural network.
* **Smart outlier removal**: remove outliers from training and prediction data sets using a variety of [outlier detection techniques](#outlier-removal).
* **Crash resilience**: store model to disk to make reloading from a crash fast and easy, and [purge obsolete files](#purging-old-model-data) for sustained dry/live runs.
* **Automatic data normalization**: [normalize the data](#feature-normalization) in a smart and statistically safe way.
* **Automatic data download**: compute the data download timerange and update historic data (in live deployments).
* **Cleaning of incoming data**: handle NaNs safely before training and prediction.
* **Dimensionality reduction**: reduce the size of the training data via [Principal Component Analysis](#reducing-data-dimensionality-with-principal-component-analysis).
* **Deploying bot fleets**: set one bot to train models while a fleet of [follower bots](#setting-up-a-follower) inference the models and handle trades.
* **Self-adaptive retraining** - Retrain models during [live deployments](freqai-running.md#live-deployments) to self-adapt to the market in asupervised manner
* **Rapid feature engineering** - Create large rich [feature sets](freqai-feature-engineering.md#feature-engineering) (10k+ features) based on simple user-created strategies
* **High performance** - Threading allows for adaptive model retraining on a separate thread (or on GPU if available) from model inferencing (prediction) and bot trade operations. Newest models and data are kept in RAM for rapid inferencing
* **Realistic backtesting** - Emulate self-adaptive training on historic data with a [backtesting module](freqai-running.md#backtesting) that automates retraining
* **Extensibility** - The generalized and robust architecture allows for incorporating any [machine learning library/method](freqai-configuration.md#using-different-prediction-models) available in Python. Eight examples are currently available, including classifiers, regressors, and a convolutional neural network
* **Smart outlier removal** - Remove outliers from training and prediction data sets using a variety of [outlier detection techniques](freqai-feature-engineering.md#outlier-detection)
* **Crash resilience** - Store trained models to disk to make reloading from a crash fast and easy, and [purge obsolete files](freqai-running.md#purging-old-model-data) for sustained dry/live runs
* **Automatic data normalization** - [Normalize the data](freqai-feature-engineering.md#feature-normalization) in a smart and statistically safe way
* **Automatic data download** - Compute timeranges for data downloads and update historic data (in live deployments)
* **Cleaning of incoming data** - Handle NaNs safely before training and model inferencing
* **Dimensionality reduction** - Reduce the size of the training data via [Principal Component Analysis](freqai-feature-engineering.md#data-dimensionality-reduction-with-principal-component-analysis)
* **Deploying bot fleets** - Set one bot to train models while a fleet of [consumers](producer-consumer.md) use signals.
## Quick start
The easiest way to quickly test FreqAI is to run it in dry mode with the following command
The easiest way to quickly test FreqAI is to run it in dry mode with the following command:
The user provides FreqAI with a set of custom *base* indicators (the same way as in a typical Freqtrade strategy) as well as target values (*labels*).
FreqAI trains a model to predict the target values based on the input of custom indicators, for each pair in the whitelist. These models are consistently retrained to adapt to market conditions. FreqAI offers the ability to both backtest strategies (emulating reality with periodic retraining) and deploy dry/live runs. In dry/live conditions, FreqAI can be set to constant retraining in a background thread in an effort to keep models as up to date as possible.
You provide FreqAI with a set of custom *base indicators* (the same way as in a [typical Freqtrade strategy](strategy-customization.md)) as well as target values (*labels*). For each pair in the whitelist, FreqAI trains a model to predict the target values based on the input of custom indicators. The models are then consistently retrained, with a predetermined frequency, to adapt to market conditions. FreqAI offers the ability to both backtest strategies (emulating reality with periodic retraining on historic data) and deploy dry/live runs. In dry/live conditions, FreqAI can be set to constant retraining in a background thread to keep models as up to date as possible.
An overview of the algorithm is shown below, explaining the data processing pipeline and the model usage.
An overview of the algorithm, explaining the data processing pipeline and model usage, is shown below.
### Important machine learning vocabulary
**Features** - the quantities with which a model is trained. All features for a single candle is stored as a vector. In FreqAI, the user
builds the feature sets from anything they can construct in the strategy.
**Features** - the parameters, based on historic data, on which a model is trained. All features for a single candle are stored as a vector. In FreqAI, you build a feature data set from anything you can construct in the strategy.
**Labels** - the target values that a model is trained
toward. Each set of features is associated with a single label that is
defined by the user within the strategy. These labels intentionally look into the
future, and are not available to the model during dry/live/backtesting.
**Labels** - the target values that the model is trained toward. Each feature vector is associated with a single label that is defined by you within the strategy. These labels intentionally look into the future and are what you are training the model to be able to predict.
**Training** - the process of feeding individual feature sets, composed of historic data, with associated labels into the
model with the goal of matching input feature sets to associated labels.
**Training** - the process of "teaching" the model to match the feature sets to the associated labels. Different types of models "learn" in different ways which means that one might be better than another for a specific application. More information about the different models that are already implemented in FreqAI can be found [here](freqai-configuration.md#using-different-prediction-models).
**Train data** - a subset of the historic data that is fed to the model during
training. This data directly influences weight connections in the model.
**Train data** - a subset of the feature data set that is fed to the model during training to "teach" the model how to predict the targets. This data directly influences weight connections in the model.
**Test data** - a subset of the historic data that is used to evaluate the performance of the model after training. This data does not influence nodal weights within the model.
**Test data** - a subset of the feature data set that is used to evaluate the performance of the model after training. This data does not influence nodal weights within the model.
**Inferencing** - the process of feeding a trained model new unseen data on which it will make a prediction.
## Install prerequisites
The normal Freqtrade install process will ask the user if they wish to install FreqAI dependencies. The user should reply "yes" to this question if they wish to use FreqAI. If the user did not reply yes, they can manually install these dependencies after the install with:
The normal Freqtrade install process will ask if you wish to install FreqAI dependencies. You should reply "yes" to this question if you wish to use FreqAI. If you did not reply yes, you can manually install these dependencies after the install with:
``` bash
pip install -r requirements-freqai.txt
```
!!! Note
Catboost will not be installed on arm devices (raspberry, Mac M1, ARM based VPS, ...), since Catboost does not provide wheels for this platform.
Catboost will not be installed on arm devices (raspberry, Mac M1, ARM based VPS, ...), since it does not provide wheels for this platform.
### Usage with docker
For docker users, a dedicated tag with freqAI dependencies is available as `:freqai`.
As such - you can replace the image line in your docker-compose file with `image: freqtradeorg/freqtrade:develop_freqai`.
This image contains the regular freqAI dependencies. Similar to native installs, Catboost will not be available on ARM based devices.
If you are using docker, a dedicated tag with FreqAI dependencies is available as `:freqai`. As such - you can replace the image line in your docker compose file with `image: freqtradeorg/freqtrade:develop_freqai`. This image contains the regular FreqAI dependencies. Similar to native installs, Catboost will not be available on ARM based devices.
## Setting up FreqAI
### FreqAI position in open-source machine learning landscape
### Parameter table
Forecasting chaotic time-series based systems, such as equity/cryptocurrency markets, requires a broad set of tools geared toward testing a wide range of hypotheses. Fortunately, a recent maturation of robust machine learning libraries (e.g. `scikit-learn`) has opened up a wide range of research possibilities. Scientists from a diverse range of fields can now easily prototype their studies on an abundance of established machine learning algorithms. Similarly, these user-friendly libraries enable "citzen scientists" to use their basic Python skills for data exploration. However, leveraging these machine learning libraries on historical and live chaotic data sources can be logistically difficult and expensive. Additionally, robust data collection, storage, and handling presents a disparate challenge. [`FreqAI`](#freqai) aims to provide a generalized and extensible open-sourced framework geared toward live deployments of adaptive modeling for market forecasting. The `FreqAI` framework is effectively a sandbox for the rich world of open-source machine learning libraries. Inside the `FreqAI` sandbox, users find they can combine a wide variety of third-party libraries to test creative hypotheses on a free live 24/7 chaotic data source - cryptocurrency exchange data.
The table below will list all configuration parameters available for FreqAI, presented in the same order as `config_examples/config_freqai.example.json`.
### Citing FreqAI
Mandatory parameters are marked as **Required**, which means that they are required to be set in one of the possible ways.
FreqAI is [published in the Journal of Open Source Software](https://joss.theoj.org/papers/10.21105/joss.04864). If you find FreqAI useful in your research, please use the following citation:
| Parameter | Description |
|------------|-------------|
||**General configuration parameters**
| `freqai` | **Required.** <br> The parent dictionary containing all the parameters for controlling FreqAI. <br> **Datatype:** Dictionary.
| `startup_candles` | Number of candles needed for *backtesting only* to ensure all indicators are non NaNs at the start of the first train period. <br> **Datatype:** Positive integer.
| `purge_old_models` | Delete obsolete models (otherwise, all historic models will remain on disk). <br> **Datatype:** Boolean. Default: `False`.
| `train_period_days` | **Required.** <br> Number of days to use for the training data (width of the sliding window). <br> **Datatype:** Positive integer.
| `backtest_period_days` | **Required.** <br> Number of days to inference from the trained model before sliding the window defined above, and retraining the model. This can be fractional days, but beware that the user-provided `timerange` will be divided by this number to yield the number of trainings necessary to complete the backtest. <br> **Datatype:** Float.
| `identifier` | **Required.** <br> A unique name for the current model. This can be reused to reload pre-trained models/data. <br> **Datatype:** String.
| `live_retrain_hours` | Frequency of retraining during dry/live runs. <br> Default set to 0, which means the model will retrain as often as possible. <br> **Datatype:** Float > 0.
| `expiration_hours` | Avoid making predictions if a model is more than `expiration_hours` old. <br> Defaults set to 0, which means models never expire. <br> **Datatype:** Positive integer.
| `fit_live_predictions_candles` | Number of historical candles to use for computing target (label) statistics from prediction data, instead of from the training data set. <br> **Datatype:** Positive integer.
| `follow_mode` | If true, this instance of FreqAI will look for models associated with `identifier` and load those for inferencing. A `follower` will **not** train new models. <br> **Datatype:** Boolean. Default: `False`.
| | **Feature parameters**
| `feature_parameters` | A dictionary containing the parameters used to engineer the feature set. Details and examples are shown [here](#feature-engineering). <br> **Datatype:** Dictionary.
| `include_timeframes` | A list of timeframes that all indicators in `populate_any_indicators` will be created for. The list is added as features to the base asset feature set. <br> **Datatype:** List of timeframes (strings).
| `include_corr_pairlist` | A list of correlated coins that FreqAI will add as additional features to all `pair_whitelist` coins. All indicators set in `populate_any_indicators` during feature engineering (see details [here](#feature-engineering)) will be created for each coin in this list, and that set of features is added to the base asset feature set. <br> **Datatype:** List of assets (strings).
| `label_period_candles` | Number of candles into the future that the labels are created for. This is used in `populate_any_indicators` (see `templates/FreqaiExampleStrategy.py` for detailed usage). The user can create custom labels, making use of this parameter or not. <br> **Datatype:** Positive integer.
| `include_shifted_candles` | Add features from previous candles to subsequent candles to add historical information. FreqAI takes all features from the `include_shifted_candles` previous candles, duplicates and shifts them so that the information is available for the subsequent candle. <br> **Datatype:** Positive integer.
| `weight_factor` | Used to set weights for training data points according to their recency. See details about how it works [here](#controlling-the-model-learning-process). <br> **Datatype:** Positive float (typically < 1).
| `indicator_max_period_candles` | The maximum period used in `populate_any_indicators()` for indicator creation. FreqAI uses this information in combination with the maximum timeframe to calculate how many data points that should be downloaded so that the first data point does not have a NaN. <br> **Datatype:** Positive integer.
| `indicator_periods_candles` | Calculate indicators for `indicator_periods_candles` time periods and add them to the feature set. <br> **Datatype:** List of positive integers.
| `stratify_training_data` | This value is used to indicate the grouping of the data. For example, 2 would set every 2nd data point into a separate dataset to be pulled from during training/testing. See details about how it works [here](#stratifying-the-data-for-training-and-testing-the-model) <br> **Datatype:** Positive integer.
| `principal_component_analysis` | Automatically reduce the dimensionality of the data set using Principal Component Analysis. See details about how it works [here](#reducing-data-dimensionality-with-principal-component-analysis) <br> **Datatype:** Boolean.
| `DI_threshold` | Activates the Dissimilarity Index for outlier detection when > 0. See details about how it works [here](#removing-outliers-with-the-dissimilarity-index). <br> **Datatype:** Positive float (typically < 1).
| `use_SVM_to_remove_outliers` | Train a support vector machine to detect and remove outliers from the training data set, as well as from incoming data points. See details about how it works [here](#removing-outliers-using-a-support-vector-machine-svm). <br> **Datatype:** Boolean.
| `svm_params` | All parameters available in Sklearn's `SGDOneClassSVM()`. See details about some select parameters [here](#removing-outliers-using-a-support-vector-machine-svm). <br> **Datatype:** Dictionary.
| `use_DBSCAN_to_remove_outliers` | Cluster data using DBSCAN to identify and remove outliers from training and prediction data. See details about how it works [here](#removing-outliers-with-dbscan). <br> **Datatype:** Boolean.
| `outlier_protection_percentage` | If more than `outlier_protection_percentage` fraction of points are removed as outliers, FreqAI will log a warning message and ignore outlier detection while keeping the original dataset intact. <br> **Datatype:** float. Default: `30`
| | **Data split parameters**
| `data_split_parameters` | Include any additional parameters available from Scikit-learn `test_train_split()`, which are shown [here](https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.train_test_split.html) (external website). <br> **Datatype:** Dictionary.
| `test_size` | Fraction of data that should be used for testing instead of training. <br> **Datatype:** Positive float < 1.
| `shuffle` | Shuffle the training data points during training. Typically, for time-series forecasting, this is set to `False`. <br>
| | **Model training parameters**
| `model_training_parameters` | A flexible dictionary that includes all parameters available by the user selected model library. For example, if the user uses `LightGBMRegressor`, this dictionary can contain any parameter available by the `LightGBMRegressor` [here](https://lightgbm.readthedocs.io/en/latest/pythonapi/lightgbm.LGBMRegressor.html) (external website). If the user selects a different model, this dictionary can contain any parameter from that model. <br> **Datatype:** Dictionary.**Datatype:** Boolean.
| `n_estimators` | The number of boosted trees to fit in regression. <br> **Datatype:** Integer.
| `n_jobs`, `thread_count`, `task_type` | Set the number of threads for parallel processing and the `task_type` (`gpu` or `cpu`). Different model libraries use different parameter names. <br> **Datatype:** Float.
| | **Extraneous parameters**
| `keras` | If your model makes use of Keras (typical for Tensorflow-based prediction models), activate this flag so that the model save/loading follows Keras standards. <br> **Datatype:** Boolean. Default: `False`.
| `conv_width` | The width of a convolutional neural network input tensor. This replaces the need for shifting candles (`include_shifted_candles`) by feeding in historical data points as the second dimension of the tensor. Technically, this parameter can also be used for regressors, but it only adds computational overhead and does not change the model training/prediction. <br> **Datatype:** Integer. Default: 2.
### Important dataframe key patterns
Below are the values the user can expect to include/use inside a typical strategy dataframe (`df[]`):
| DataFrame Key | Description |
|------------|-------------|
| `df['&*']` | Any dataframe column prepended with `&` in `populate_any_indicators()` is treated as a training target (label) inside FreqAI (typically following the naming convention `&-s*`). The names of these dataframe columns are fed back to the user as the predictions. For example, if the user wishes to predict the price change in the next 40 candles (similar to `templates/FreqaiExampleStrategy.py`), they set `df['&-s_close']`. FreqAI makes the predictions and gives them back under the same key (`df['&-s_close']`) to be used in `populate_entry/exit_trend()`. <br> **Datatype:** Depends on the output of the model.
| `df['&*_std/mean']` | Standard deviation and mean values of the user-defined labels during training (or live tracking with `fit_live_predictions_candles`). Commonly used to understand the rarity of a prediction (use the z-score as shown in `templates/FreqaiExampleStrategy.py` to evaluate how often a particular prediction was observed during training or historically with `fit_live_predictions_candles`). <br> **Datatype:** Float.
| `df['do_predict']` | Indication of an outlier data point. The return value is integer between -1 and 2, which lets the user know if the prediction is trustworthy or not. `do_predict==1` means the prediction is trustworthy. If the Dissimilarity Index (DI, see details [here](#removing-outliers-with-the-dissimilarity-index)) of the input data point is above the user-defined threshold, FreqAI will subtract 1 from `do_predict`, resulting in `do_predict==0`. If `use_SVM_to_remove_outliers()` is active, the Support Vector Machine (SVM) may also detect outliers in training and prediction data. In this case, the SVM will also subtract 1 from `do_predict`. If the input data point was considered an outlier by the SVM but not by the DI, the result will be `do_predict==0`. If both the DI and the SVM considers the input data point to be an outlier, the result will be `do_predict==-1`. A particular case is when `do_predict == 2`, which means that the model has expired due to exceeding `expired_hours`. <br> **Datatype:** Integer between -1 and 2.
| `df['DI_values']` | Dissimilarity Index values are proxies to the level of confidence FreqAI has in the prediction. A lower DI means the prediction is close to the training data, i.e., higher prediction confidence. <br> **Datatype:** Float.
| `df['%*']` | Any dataframe column prepended with `%` in `populate_any_indicators()` is treated as a training feature. For example, the user can include the RSI in the training feature set (similar to in `templates/FreqaiExampleStrategy.py`) by setting `df['%-rsi']`. See more details on how this is done [here](#feature-engineering). <br> **Note**: Since the number of features prepended with `%` can multiply very quickly (10s of thousands of features is easily engineered using the multiplictative functionality described in the `feature_parameters` table shown above), these features are removed from the dataframe upon return from FreqAI. If the user wishes to keep a particular type of feature for plotting purposes, they can prepend it with `%%`. <br> **Datatype:** Depends on the output of the model.
### File structure
`user_data_dir/models/` contains all the data associated with the trainings and backtests.
This file structure is heavily controlled and inferenced by the `FreqaiDataKitchen()`
and should therefore not be modified.
### Example config file
The user interface is isolated to the typical Freqtrade config file. A FreqAI config should include:
```json
"freqai": {
"enabled": true,
"startup_candles": 10000,
"purge_old_models": true,
"train_period_days": 30,
"backtest_period_days": 7,
"identifier" : "unique-id",
"feature_parameters" : {
"include_timeframes": ["5m","15m","4h"],
"include_corr_pairlist": [
"ETH/USD",
"LINK/USD",
"BNB/USD"
],
"label_period_candles": 24,
"include_shifted_candles": 2,
"indicator_max_period_candles": 20,
"indicator_periods_candles": [10, 20]
},
"data_split_parameters" : {
"test_size": 0.25
},
"model_training_parameters" : {
"n_estimators": 100
},
}
```bibtex
@article{Caulk2022,
doi = {10.21105/joss.04864},
url = {https://doi.org/10.21105/joss.04864},
year = {2022}, publisher = {The Open Journal},
volume = {7}, number = {80}, pages = {4864},
author = {Robert A. Caulk and Elin Törnquist and Matthias Voppichler and Andrew R. Lawless and Ryan McMullan and Wagner Costa Santos and Timothy C. Pogue and Johan van der Vlugt and Stefan P. Gehring and Pascal Schmidt},
title = {FreqAI: generalizing adaptive modeling for chaotic time-series market forecasts},
journal = {Journal of Open Source Software} }
```
## Building a FreqAI strategy
The FreqAI strategy requires the user to include the following lines of code in the standard Freqtrade strategy:
Notice how the `populate_any_indicators()` is where the user adds their own features ([more information](#feature-engineering)) and labels ([more information](#setting-classifier-targets)). See a full example at `templates/FreqaiExampleStrategy.py`.
## Creating a dynamic target
The `&*_std/mean` return values describe the statistical fit of the user defined label *during the most recent training*. This value allows the user to know the rarity of a given prediction. For example, `templates/FreqaiExampleStrategy.py`, creates a `target_roi` which is based on filtering out predictions that are below a given z-score of 1.25.
of *historical predictions* for creating the dynamic target instead of the trained labels, (as discussed above) the user
can do so by setting `fit_live_prediction_candles` in the config to the number of historical prediction candles
the user wishes to use to generate target statistics.
```json
"freqai": {
"fit_live_prediction_candles": 300,
}
```
If the user sets this value, FreqAI will initially use the predictions from the training data
and subsequently begin introducing real prediction data as it is generated. FreqAI will save
this historical data to be reloaded if the user stops and restarts a model with the same `identifier`.
## Building a custom prediction model
FreqAI has multiple example prediction model libraries, such as `Catboost` regression (`freqai/prediction_models/CatboostRegressor.py`) and `LightGBM` regression.
However, the user can customize and create their own prediction models using the `IFreqaiModel` class.
The user is encouraged to inherit `train()` and `predict()` to let them customize various aspects of their training procedures.
## Feature engineering
Features are added by the user inside the `populate_any_indicators()` method of the strategy
by prepending indicators with `%`, and labels with `&`.
There are some important components/structures that the user *must* include when building their feature set; the use of these is shown below:
In the presented example strategy, the user does not wish to pass the `bb_lowerband` as a feature to the model,
and has therefore not prepended it with `%`. The user does, however, wish to pass `bb_width` to the
model for training/prediction and has therefore prepended it with `%`.
The `include_timeframes` in the example config above are the timeframes (`tf`) of each call to `populate_any_indicators()` in the strategy. In the present case, the user is asking for the
`5m`, `15m`, and `4h` timeframes of the `rsi`, `mfi`, `roc`, and `bb_width` to be included in the feature set.
The user can ask for each of the defined features to be included also from
informative pairs using the `include_corr_pairlist`. This means that the feature
set will include all the features from `populate_any_indicators` on all the `include_timeframes` for each of the correlated pairs defined in the config (`ETH/USD`, `LINK/USD`, and `BNB/USD`).
`include_shifted_candles` indicates the number of previous
candles to include in the feature set. For example, `include_shifted_candles: 2` tells
FreqAI to include the past 2 candles for each of the features in the feature set.
In total, the number of features the user of the presented example strat has created is:
length of `include_timeframes` * no. features in `populate_any_indicators()` * length of `include_corr_pairlist` * no. `include_shifted_candles` * length of `indicator_periods_candles`
$= 3 * 3 * 3 * 2 * 2 = 108$.
Another structure to consider is the location of the labels at the bottom of the example function (below `if set_generalized_indicators:`).
This is where the user will add single features and labels to their feature set to avoid duplication of them from
various configuration parameters that multiply the feature set, such as `include_timeframes`.
!!! Note
Features **must** be defined in `populate_any_indicators()`. Definining FreqAI features in `populate_indicators()`
will cause the algorithm to fail in live/dry mode. If the user wishes to add generalized features that are not associated with
a specific pair or timeframe, they should use the following structure inside `populate_any_indicators()`
(as exemplified in `freqtrade/templates/FreqaiExampleStrategy.py`):
Additionally, the example classifier models do not accommodate multiple labels, but they do allow multi-class classification within a single label column.
## Running FreqAI
There are two ways to train and deploy an adaptive machine learning model. FreqAI enables live deployment as well as backtesting analyses. In both cases, a model is trained periodically, as shown in the following figure.
### Running the model live
FreqAI can be run dry/live using the following command:
By default, FreqAI will not find any existing models and will start by training a new one
based on the user's configuration settings. Following training, the model will be used to make predictions on incoming candles until a new model is available. New models are typically generated as often as possible, with FreqAI managing an internal queue of the coin pairs to try to keep all models equally up to date. FreqAI will always use the most recently trained model to make predictions on incoming live data. If the user does not want FreqAI to retrain new models as often as possible, they can set `live_retrain_hours` to tell FreqAI to wait at least that number of hours before training a new model. Additionally, the user can set `expired_hours` to tell FreqAI to avoid making predictions on models that are older than that number of hours.
If the user wishes to start a dry/live run from a saved backtest model (or from a previously crashed dry/live session), the user only needs to reuse
the same `identifier` parameter:
```json
"freqai": {
"identifier": "example",
"live_retrain_hours": 0.5
}
```
In this case, although FreqAI will initiate with a
pre-trained model, it will still check to see how much time has elapsed since the model was trained,
and if a full `live_retrain_hours` has elapsed since the end of the loaded model, FreqAI will retrain.
### Backtesting
The FreqAI backtesting module can be executed with the following command:
Backtesting mode requires the user to have the data pre-downloaded (unlike in dry/live mode where FreqAI automatically downloads the necessary data). The user should be careful to consider that the time range of the downloaded data is more than the backtesting time range. This is because FreqAI needs data prior to the desired backtesting time range in order to train a model to be ready to make predictions on the first candle of the user-set backtesting time range. More details on how to calculate the data to download can be found [here](#deciding-the-sliding-training-window-and-backtesting-duration).
If this command has never been executed with the existing config file, it will train a new model
for each pair, for each backtesting window within the expanded `--timerange`.
!!! Note "Model reuse"
Once the training is completed, the user can execute the backtesting again with the same config file and
FreqAI will find the trained models and load them instead of spending time training. This is useful
if the user wants to tweak (or even hyperopt) buy and sell criteria inside the strategy. If the user
*wants* to retrain a new model with the same config file, then they should simply change the `identifier`.
This way, the user can return to using any model they wish by simply specifying the `identifier`.
---
### Deciding the size of the sliding training window and backtesting duration
The user defines the backtesting timerange with the typical `--timerange` parameter in the
configuration file. The duration of the sliding training window is set by `train_period_days`, whilst
`backtest_period_days` is the sliding backtesting window, both in number of days (`backtest_period_days` can be
a float to indicate sub-daily retraining in live/dry mode). In the presented example config,
the user is asking FreqAI to use a training period of 30 days and backtest on the subsequent 7 days.
This means that if the user sets `--timerange 20210501-20210701`,
FreqAI will train have trained 8 separate models at the end of `--timerange` (because the full range comprises 8 weeks). After the training of the model, FreqAI will backtest the subsequent 7 days. The "sliding window" then moves one week forward (emulating FreqAI retraining once per week in live mode) and the new model uses the previous 30 days (including the 7 days used for backtesting by the previous model) to train. This is repeated until the end of `--timerange`.
In live mode, the required training data is automatically computed and downloaded. However, in backtesting mode,
the user must manually enter the required number of `startup_candles` in the config. This value
is used to increase the data to FreqAI, which should be sufficient to enable all indicators
to be NaN free at the beginning of the first training. This is done by identifying the
longest timeframe (`4h` in presented example config) and the longest indicator period (`20` days in presented example config)
and adding this to the `train_period_days`. The units need to be in the base candle time frame:
`startup_candles` = ( 4 hours * 20 max period * 60 minutes/hour + 30 day train_period_days * 1440 minutes per day ) / 5 min (base time frame) = 9360.
!!! Note
In dry/live mode, this is all precomputed and handled automatically. Thus, `startup_candle` has no influence on dry/live mode.
!!! Note
Although fractional `backtest_period_days` is allowed, the user should be aware that the `--timerange` is divided by this value to determine the number of models that FreqAI will need to train in order to backtest the full range. For example, if the user wants to set a `--timerange` of 10 days, and asks for a `backtest_period_days` of 0.1, FreqAI will need to train 100 models per pair to complete the full backtest. Because of this, a true backtest of FreqAI adaptive training would take a *very* long time. The best way to fully test a model is to run it dry and let it constantly train. In this case, backtesting would take the exact same amount of time as a dry run.
### Defining model expirations
During dry/live mode, FreqAI trains each coin pair sequentially (on separate threads/GPU from the main Freqtrade bot). This means that there is always an age discrepancy between models. If a user is training on 50 pairs, and each pair requires 5 minutes to train, the oldest model will be over 4 hours old. This may be undesirable if the characteristic time scale (the trade duration target) for a strategy is less than 4 hours. The user can decide to only make trade entries if the model is less than
a certain number of hours old by setting the `expiration_hours` in the config file:
```json
"freqai": {
"expiration_hours": 0.5,
}
```
In the presented example config, the user will only allow predictions on models that are less than 1/2 hours old.
### Purging old model data
FreqAI stores new model files each time it retrains. These files become obsolete as new models are trained and FreqAI adapts to new market conditions. Users planning to leave FreqAI running for extended periods of time with high frequency retraining should enable `purge_old_models` in their config:
```json
"freqai": {
"purge_old_models": true,
}
```
This will automatically purge all models older than the two most recently trained ones.
### Returning additional info from training
The user may find that there are some important metrics that they'd like to return to the strategy at the end of each model training.
The user can include these metrics by assigning them to `dk.data['extra_returns_per_train']['my_new_value'] = XYZ` inside their custom prediction model class. FreqAI takes the `my_new_value` assigned in this dictionary and expands it to fit the return dataframe to the strategy.
The user can then use the value in the strategy with `dataframe['my_new_value']`. An example of how this is already used in FreqAI is
the `&*_mean` and `&*_std` values, which indicate the mean and standard deviation of the particular target (label) during the most recent training.
An example, where the user wants to use live metrics from the trade database, is shown below:
```json
"freqai": {
"extra_returns_per_train": {"total_profit": 4}
}
```
The user needs to set the standard dictionary in the config so that FreqAI can return proper dataframe shapes. These values will likely be overridden by the prediction model, but in the case where the model has yet to set them, or needs a default initial value, this is the value that will be returned.
### Setting up a follower
The user can define:
```json
"freqai": {
"follow_mode": true,
"identifier": "example"
}
```
to indicate to the bot that it should not train models, but instead should look for models trained by a leader with the same `identifier`. In this example, the user has a leader bot with the `identifier: "example"`. The leader bot is already running or launching simultaneously as the follower.
The follower will load models created by the leader and inference them to obtain predictions.
## Data manipulation techniques
### Feature normalization
The feature set created by the user is automatically normalized to the training data. This includes all test data and unseen prediction data (dry/live/backtest).
### Reducing data dimensionality with Principal Component Analysis
Users can reduce the dimensionality of their features by activating the `principal_component_analysis` in the config:
```json
"freqai": {
"feature_parameters" : {
"principal_component_analysis": true
}
}
```
This will perform PCA on the features and reduce the dimensionality of the data so that the explained variance of the data set is >= 0.999.
### Stratifying the data for training and testing the model
The user can stratify (group) the training/testing data using:
```json
"freqai": {
"feature_parameters" : {
"stratify_training_data": 3
}
}
```
This will split the data chronologically so that every Xth data point is used to test the model after training. In the
example above, the user is asking for every third data point in the dataframe to be used for
testing; the other points are used for training.
The test data is used to evaluate the performance of the model after training. If the test score is high, the model is able to capture the behavior of the data well. If the test score is low, either the model either does not capture the complexity of the data, the test data is significantly different from the train data, or a different model should be used.
### Controlling the model learning process
Model training parameters are unique to the machine learning library selected by the user. FreqAI allows the user to set any parameter for any library using the `model_training_parameters` dictionary in the user configuration file. The example configuration file (found in `config_examples/config_freqai.example.json`) show some of the example parameters associated with `Catboost` and `LightGBM`, but the user can add any parameters available in those libraries.
Data split parameters are defined in `data_split_parameters` which can be any parameters associated with `Sklearn`'s `train_test_split()` function.
FreqAI includes some additional parameters such as `weight_factor`, which allows the user to weight more recent data more strongly
than past data via an exponential function:
$$ W_i = \exp(\frac{-i}{\alpha*n}) $$
where $W_i$ is the weight of data point $i$ in a total set of $n$ data points. Below is a figure showing the effect of different weight factors on the data points (candles) in a feature set.
`train_test_split()` has a parameters called `shuffle` that allows the user to keep the data unshuffled. This is particularly useful to avoid biasing training with temporally auto-correlated data.
Finally, `label_period_candles` defines the offset (number of candles into the future) used for the `labels`. In the presented example config,
the user is asking for `labels` that are 24 candles in the future.
### Outlier removal
#### Removing outliers with the Dissimilarity Index
The user can tell FreqAI to remove outlier data points from the training/test data sets using a Dissimilarity Index by including the following statement in the config:
```json
"freqai": {
"feature_parameters" : {
"DI_threshold": 1
}
}
```
Equity and crypto markets suffer from a high level of non-patterned noise in the form of outlier data points. The Dissimilarity Index (DI) aims to quantify the uncertainty associated with each prediction made by the model. The DI allows predictions which are outliers (not existent in the model feature space) to be thrown out due to low levels of certainty.
To do so, FreqAI measures the distance between each training data point (feature vector), $X_{a}$, and all other training data points:
where $d_{ab}$ is the distance between the normalized points $a$ and $b$. $p$ is the number of features, i.e., the length of the vector $X$. The characteristic distance, $\overline{d}$ for a set of training data points is simply the mean of the average distances:
$\overline{d}$ quantifies the spread of the training data, which is compared to the distance between a new prediction feature vectors, $X_k$ and all the training data:
$$ d_k = \arg \min d_{k,i} $$
which enables the estimation of the Dissimilarity Index as:
$$ DI_k = d_k/\overline{d} $$
The user can tweak the DI through the `DI_threshold` to increase or decrease the extrapolation of the trained model.
Below is a figure that describes the DI for a 3D data set.
#### Removing outliers using a Support Vector Machine (SVM)
The user can tell FreqAI to remove outlier data points from the training/test data sets using a SVM by setting:
```json
"freqai": {
"feature_parameters" : {
"use_SVM_to_remove_outliers": true
}
}
```
FreqAI will train an SVM on the training data (or components of it if the user activated
`principal_component_analysis`) and remove any data point that the SVM deems to be beyond the feature space.
The parameter `shuffle` is by default set to `False` to ensure consistent results. If it is set to `True`, running the SVM multiple times on the same data set might result in different outcomes due to `max_iter` being to low for the algorithm to reach the demanded `tol`. Increasing `max_iter` solves this issue but causes the procedure to take longer time.
The parameter `nu`, *very* broadly, is the amount of data points that should be considered outliers.
#### Removing outliers with DBSCAN
The user can configure FreqAI to use DBSCAN to cluster and remove outliers from the training/test data set or incoming outliers from predictions, by activating `use_DBSCAN_to_remove_outliers` in the config:
```json
"freqai": {
"feature_parameters" : {
"use_DBSCAN_to_remove_outliers": true
}
}
```
DBSCAN is an unsupervised machine learning algorithm that clusters data without needing to know how many clusters there should be.
Given a number of data points $N$, and a distance $\varepsilon$, DBSCAN clusters the data set by setting all data points that have $N-1$ other data points within a distance of $\varepsilon$ as *core points*. A data point that is within a distance of $\varepsilon$ from a *core point* but that does not have $N-1$ other data points within a distance of $\varepsilon$ from itself is considered an *edge point*. A cluster is then the collection of *core points* and *edge points*. Data points that have no other data points at a distance $<\varepsilon$ are considered outliers. The figure below shows a cluster with $N = 3$.
FreqAI uses `sklearn.cluster.DBSCAN` (details are available on scikit-learn's webpage [here](#https://scikit-learn.org/stable/modules/generated/sklearn.cluster.DBSCAN.html)) with `min_samples` ($N$) taken as double the no. of user-defined features, and `eps` ($\varepsilon$) taken as the longest distance in the *k-distance graph* computed from the nearest neighbors in the pairwise distances of all data points in the feature set.
## Additional information
### Common pitfalls
## Common pitfalls
FreqAI cannot be combined with dynamic `VolumePairlists` (or any pairlist filter that adds and removes pairs dynamically).
This is for performance reasons - FreqAI relies on making quick predictions/retrains. To do this effectively,
@@ -743,17 +103,21 @@ new candles automatically for future retrains. This means that if new pairs arri
## Credits
FreqAI was developed by a group of individuals who all contributed specific skillsets to the project.
FreqAI is developed by a group of individuals who all contribute specific skillsets to the project.
Conception and software development:
Robert Caulk @robcaulk
Theoretical brainstorming, data analysis:
Theoretical brainstorming and data analysis:
Elin Törnquist @th0rntwig
Code review, software architecture brainstorming:
Code review and software architecture brainstorming:
@xmatthias
Software development:
Wagner Costa @wagnercosta
Emre Suzen @aemr3
Timothy Pogue @wizrds
Beta testing and bug reporting:
@bloodhunter4rc, Salah Lamkadem @ikonx, @ken11o2, @longyu, @paranoidandy, @smidelis, @smarm
Juha Nykänen @suikula, Wagner Costa @wagnercosta
Stefan Gehring @bloodhunter4rc, @longyu, Andrew Lawless @paranoidandy, Pascal Schmidt @smidelis, Ryan McMullan @smarmau, Juha Nykänen @suikula, Johan van der Vlugt @jooopiert, Richárd Józsa @richardjosza
Specify which parameters to hyperopt. Space-separated
list.
--print-all Print all results, not only the best ones.
@@ -180,6 +180,7 @@ Rarely you may also need to create a [nested class](advanced-hyperopt.md#overrid
*`generate_roi_table` - for custom ROI optimization (if you need the ranges for the values in the ROI table that differ from default or the number of entries (steps) in the ROI table which differs from the default 4 steps)
*`stoploss_space` - for custom stoploss optimization (if you need the range for the stoploss parameter in the optimization hyperspace that differs from default)
*`trailing_space` - for custom trailing stop optimization (if you need the ranges for the trailing stop parameters in the optimization hyperspace that differ from default)
*`max_open_trades_space` - for custom max_open_trades optimization (if you need the ranges for the max_open_trades parameter in the optimization hyperspace that differ from default)
!!! Tip "Quickly optimize ROI, stoploss and trailing stoploss"
You can quickly optimize the spaces `roi`, `stoploss` and `trailing` without changing anything in your strategy.
@@ -365,7 +366,7 @@ class MyAwesomeStrategy(IStrategy):
timeframe = '15m'
minimal_roi = {
"0": 0.10
},
}
# Define the parameter spaces
buy_ema_short = IntParameter(3, 50, default=5)
buy_ema_long = IntParameter(15, 200, default=50)
@@ -400,7 +401,7 @@ class MyAwesomeStrategy(IStrategy):
* `roi`: just optimize the minimal profit table for your strategy
* `stoploss`: search for the best stoploss value
* `trailing`: search for the best trailing stop values
* `trades`: search for the best max open trades values
* `protection`: search for the best protection parameters (read the [protections section](#optimizing-protections) on how to properly define these)
* `default`: `all` except `trailing` and `protection`
* space-separated list of any of the above values for example `--spaces roi stoploss`
@@ -916,5 +918,5 @@ Once the optimized strategy has been implemented into your strategy, you should
To achieve same the results (number of trades, their durations, profit, etc.) as during Hyperopt, please use the same configuration and parameters (timerange, timeframe, ...) used for hyperopt `--dmmp`/`--disable-max-market-positions` and `--eps`/`--enable-position-stacking` for Backtesting.
Should results not match, please double-check to make sure you transferred all conditions correctly.
Pay special care to the stoploss (and trailing stoploss) parameters, as these are often set in configuration files, which override changes to the strategy.
You should also carefully review the log of your backtest to ensure that there were no parameters inadvertently set by the configuration (like `stoploss` or `trailing_stop`).
Pay special care to the stoploss, max_open_trades and trailing stoploss parameters, as these are often set in configuration files, which override changes to the strategy.
You should also carefully review the log of your backtest to ensure that there were no parameters inadvertently set by the configuration (like `stoploss`, `max_open_trades` or `trailing_stop`).
@@ -22,6 +22,8 @@ You may also use something like `.*DOWN/BTC` or `.*UP/BTC` to exclude leveraged
* [`StaticPairList`](#static-pair-list) (default, if not configured differently)
* [`VolumePairList`](#volume-pair-list)
* [`ProducerPairList`](#producerpairlist)
* [`RemotePairList`](#remotepairlist)
* [`AgeFilter`](#agefilter)
* [`OffsetFilter`](#offsetfilter)
* [`PerformanceFilter`](#performancefilter)
@@ -84,7 +86,7 @@ Filtering instances (not the first position in the list) will not apply any cach
You can define a minimum volume with `min_value` - which will filter out pairs with a volume lower than the specified value in the specified timerange.
### VolumePairList Advanced mode
##### VolumePairList Advanced mode
`VolumePairList` can also operate in an advanced mode to build volume over a given timerange of specified candle size. It utilizes exchange historical candle data, builds a typical price (calculated by (open+high+low)/3) and multiplies the typical price with every candle's volume. The sum is the `quoteVolume` over the given range. This allows different scenarios, for a more smoothened volume, when using longer ranges with larger candle sizes, or the opposite when using a short range with small candles.
@@ -146,6 +148,74 @@ More sophisticated approach can be used, by using `lookback_timeframe` for candl
!!! Note
`VolumePairList` does not support backtesting mode.
#### ProducerPairList
With `ProducerPairList`, you can reuse the pairlist from a [Producer](producer-consumer.md) without explicitly defining the pairlist on each consumer.
[Consumer mode](producer-consumer.md) is required for this pairlist to work.
The pairlist will perform a check on active pairs against the current exchange configuration to avoid attempting to trade on invalid markets.
You can limit the length of the pairlist with the optional parameter `number_assets`. Using `"number_assets"=0` or omitting this key will result in the reuse of all producer pairs valid for the current setup.
```json
"pairlists": [
{
"method": "ProducerPairList",
"number_assets": 5,
"producer_name": "default",
}
],
```
!!! Tip "Combining pairlists"
This pairlist can be combined with all other pairlists and filters for further pairlist reduction, and can also act as an "additional" pairlist, on top of already defined pairs.
`ProducerPairList` can also be used multiple times in sequence, combining the pairs from multiple producers.
Obviously in complex such configurations, the Producer may not provide data for all pairs, so the strategy must be fit for this.
#### RemotePairList
It allows the user to fetch a pairlist from a remote server or a locally stored json file within the freqtrade directory, enabling dynamic updates and customization of the trading pairlist.
The RemotePairList is defined in the pairlists section of the configuration settings. It uses the following configuration options:
```json
"pairlists": [
{
"method": "RemotePairList",
"pairlist_url": "https://example.com/pairlist",
"number_assets": 10,
"refresh_period": 1800,
"keep_pairlist_on_failure": true,
"read_timeout": 60,
"bearer_token": "my-bearer-token"
}
]
```
The `pairlist_url` option specifies the URL of the remote server where the pairlist is located, or the path to a local file (if file:/// is prepended). This allows the user to use either a remote server or a local file as the source for the pairlist.
The user is responsible for providing a server or local file that returns a JSON object with the following structure:
```json
{
"pairs": ["XRP/USDT", "ETH/USDT", "LTC/USDT"],
"refresh_period": 1800,
}
```
The `pairs` property should contain a list of strings with the trading pairs to be used by the bot. The `refresh_period` property is optional and specifies the number of seconds that the pairlist should be cached before being refreshed.
The optional `keep_pairlist_on_failure` specifies whether the previous received pairlist should be used if the remote server is not reachable or returns an error. The default value is true.
The optional `read_timeout` specifies the maximum amount of time (in seconds) to wait for a response from the remote source, The default value is 60.
The optional `bearer_token` will be included in the requests Authorization Header.
!!! Note
In case of a server error the last received pairlist will be kept if `keep_pairlist_on_failure` is set to true, when set to false a empty pairlist is returned.
#### AgeFilter
Removes pairs that have been listed on the exchange for less than `min_days_listed` days (defaults to `10`) or more than `max_days_listed` days (defaults `None` mean infinity).
@@ -241,7 +311,7 @@ This option is disabled by default, and will only apply if set to > 0.
The `max_value` setting removes pairs where the minimum value change is above a specified value.
This is useful when an exchange has unbalanced limits. For example, if step-size = 1 (so you can only buy 1, or 2, or 3, but not 1.1 Coins) - and the price is pretty high (like 20\$) as the coin has risen sharply since the last limit adaption.
As a result of the above, you can only buy for 20\$, or 40\$ - but not for 25\$.
On exchanges that deduct fees from the receiving currency (e.g. FTX) - this can result in high value coins / amounts that are unsellable as the amount is slightly below the limit.
On exchanges that deduct fees from the receiving currency (e.g. binance) - this can result in high value coins / amounts that are unsellable as the amount is slightly below the limit.
The `low_price_ratio` setting removes pairs where a raise of 1 price unit (pip) is above the `low_price_ratio` ratio.
This option is disabled by default, and will only apply if set to > 0.
@@ -259,6 +329,18 @@ Min price precision for SHITCOIN/BTC is 8 decimals. If its price is 0.00000011 -
Shuffles (randomizes) pairs in the pairlist. It can be used for preventing the bot from trading some of the pairs more frequently then others when you want all pairs be treated with the same priority.
By default, ShuffleFilter will shuffle pairs once per candle.
To shuffle on every iteration, set `"shuffle_frequency"` to `"iteration"` instead of the default of `"candle"`.
``` json
{
"method": "ShuffleFilter",
"shuffle_frequency": "candle",
"seed": 42
}
```
!!! Tip
You may set the `seed` value for this Pairlist to obtain reproducible results, which can be useful for repeated backtesting sessions. If `seed` is not set, the pairs are shuffled in the non-repeatable random order. ShuffleFilter will automatically detect runmodes and apply the `seed` only for backtesting modes - if a `seed` value is set.
@@ -32,7 +33,7 @@ Freqtrade is a free and open source crypto trading bot written in Python. It is
- Run: Test your strategy with simulated money (Dry-Run mode) or deploy it with real money (Live-Trade mode).
- Run using Edge (optional module): The concept is to find the best historical [trade expectancy](edge.md#expectancy) by markets based on variation of the stop-loss and then allow/reject markets to trade. The sizing of the trade is based on a risk of a percentage of your capital.
- Control/Monitor: Use Telegram or a WebUI (start/stop the bot, show profit/loss, daily summary, current open trades results, etc.).
- Analyse: Further analysis can be performed on either Backtesting data or Freqtrade trading history (SQL database), including automated standard plots, and methods to load the data into [interactive environments](data-analysis.md).
- Analyze: Further analysis can be performed on either Backtesting data or Freqtrade trading history (SQL database), including automated standard plots, and methods to load the data into [interactive environments](data-analysis.md).
## Supported exchange marketplaces
@@ -40,7 +41,6 @@ Please read the [exchange specific notes](exchanges.md) to learn about eventual,
- [X] [Binance](https://www.binance.com/)
- [X] [Bittrex](https://bittrex.com/)
- [X] [FTX](https://ftx.com/#a=2258149)
- [X] [Gate.io](https://www.gate.io/ref/6266643)
- [X] [Huobi](http://huobi.com/)
- [X] [Kraken](https://kraken.com/)
@@ -51,7 +51,8 @@ Please read the [exchange specific notes](exchanges.md) to learn about eventual,
- [X] [Binance](https://www.binance.com/)
- [X] [Gate.io](https://www.gate.io/ref/6266643)
- [X] [OKX](https://okx.com/).
- [X] [OKX](https://okx.com/)
- [X] [Bybit](https://bybit.com/)
Please make sure to read the [exchange specific notes](exchanges.md), as well as the [trading with leverage](leverage.md) documentation before diving in.
@@ -30,6 +30,12 @@ The easiest way to install and run Freqtrade is to clone the bot Github reposito
!!! Warning "Up-to-date clock"
The clock on the system running the bot must be accurate, synchronized to a NTP server frequently enough to avoid problems with communication to the exchanges.
!!! Error "Running setup.py install for gym did not run successfully."
If you get an error related with gym we suggest you to downgrade setuptools it to version 65.5.0 you can do it with the following command:
Please only use advanced trading modes when you know how freqtrade (and your strategy) works.
Also, never risk more than what you can afford to lose.
Please read the [strategy migration guide](strategy_migration.md#strategy-migration-between-v2-and-v3) to migrate your strategy from a freqtrade v2 strategy, to v3 strategy that can short and trade futures.
If you already have an existing strategy, please read the [strategy migration guide](strategy_migration.md#strategy-migration-between-v2-and-v3) to migrate your strategy from a freqtrade v2 strategy, to strategy of version 3 which can short and trade futures.
## Shorting
@@ -62,6 +62,11 @@ You will also have to pick a "margin mode" (explanation below) - with freqtrade
"margin_mode": "isolated"
```
##### Pair namings
Freqtrade follows the [ccxt naming conventions for futures](https://docs.ccxt.com/en/latest/manual.html?#perpetual-swap-perpetual-future).
A futures pair will therefore have the naming of `base/quote:settle` (e.g. `ETH/USDT:USDT`).
### Margin mode
On top of `trading_mode` - you will also have to configure your `margin_mode`.
@@ -85,6 +90,8 @@ One account is used to share collateral between markets (trading pairs). Margin
"margin_mode": "cross"
```
Please read the [exchange specific notes](exchanges.md) for exchanges that support this mode and how they differ.
## Set leverage to use
Different strategies and risk profiles will require different levels of leverage.
freqtrade provides a mechanism whereby an instance (also called `consumer`) may listen to messages from an upstream freqtrade instance (also called `producer`) using the message websocket. Mainly, `analyzed_df` and `whitelist` messages. This allows the reuse of computed indicators (and signals) for pairs in multiple bots without needing to compute them multiple times.
See [Message Websocket](rest-api.md#message-websocket) in the Rest API docs for setting up the `api_server` configuration for your message websocket (this will be your producer).
!!! Note
We strongly recommend to set `ws_token` to something random and known only to yourself to avoid unauthorized access to your bot.
## Configuration
Enable subscribing to an instance by adding the `external_message_consumer` section to the consumer's config file.
```json
{
//...
"external_message_consumer":{
"enabled":true,
"producers":[
{
"name":"default",// This can be any name you'd like, default is "default"
"host":"127.0.0.1",// The host from your producer's api_server config
"port":8080,// The port from your producer's api_server config
"secure":false,// Use a secure websockets connection, default false
"ws_token":"sercet_Ws_t0ken"// The ws_token from your producer's api_server config
}
],
// The following configurations are optional, and usually not required
// "wait_timeout": 300,
// "ping_timeout": 10,
// "sleep_time": 10,
// "remove_entry_exit_signals": false,
// "message_size_limit": 8
}
//...
}
```
| Parameter | Description |
|------------|-------------|
| `enabled` | **Required.** Enable consumer mode. If set to false, all other settings in this section are ignored.<br>*Defaults to `false`.*<br>**Datatype:** boolean .
| `producers` | **Required.** List of producers <br>**Datatype:** Array.
| `producers.name` | **Required.** Name of this producer. This name must be used in calls to `get_producer_pairs()` and `get_producer_df()` if more than one producer is used.<br>**Datatype:** string
| `producers.host` | **Required.** The hostname or IP address from your producer.<br>**Datatype:** string
| `producers.port` | **Required.** The port matching the above host.<br>**Datatype:** string
| `producers.secure` | **Optional.** Use ssl in websockets connection. Default False.<br>**Datatype:** string
| `producers.ws_token` | **Required.**`ws_token` as configured on the producer.<br>**Datatype:** string
| | **Optional settings**
| `wait_timeout` | Timeout until we ping again if no message is received. <br>*Defaults to `300`.*<br>**Datatype:** Integer - in seconds.
| `wait_timeout` | Ping timeout <br>*Defaults to `10`.*<br>**Datatype:** Integer - in seconds.
| `sleep_time` | Sleep time before retrying to connect.<br>*Defaults to `10`.*<br>**Datatype:** Integer - in seconds.
| `remove_entry_exit_signals` | Remove signal columns from the dataframe (set them to 0) on dataframe receipt.<br>*Defaults to `10`.*<br>**Datatype:** Integer - in seconds.
| `message_size_limit` | Size limit per message<br>*Defaults to `8`.*<br>**Datatype:** Integer - Megabytes.
Instead of (or as well as) calculating indicators in `populate_indicators()` the follower instance listens on the connection to a producer instance's messages (or multiple producer instances in advanced configurations) and requests the producer's most recently analyzed dataframes for each pair in the active whitelist.
A consumer instance will then have a full copy of the analyzed dataframes without the need to calculate them itself.
## Examples
### Example - Producer Strategy
A simple strategy with multiple indicators. No special considerations are required in the strategy itself.
You can use this to setup [FreqAI](freqai.md) on a powerful machine, while you run consumers on simple machines like raspberries, which can interpret the signals generated from the producer in different ways.
### Example - Consumer Strategy
A logically equivalent strategy which calculates no indicators itself, but will have the same analyzed dataframes available to make trading decisions based on the indicators calculated in the producer. In this example the consumer has the same entry criteria, however this is not necessary. The consumer may use different logic to enter/exit trades, and only use the indicators as specified.
```py
classConsumerStrategy(IStrategy):
#...
process_only_new_candles=False# required for consumers
By setting `remove_entry_exit_signals=false`, you can also use the producer's signals directly. They should be available as `enter_long_default` (assuming `suffix="default"` was used) - and can be used as either signal directly, or as additional indicator.
@@ -93,7 +94,6 @@ Make sure that the following 2 lines are available in your docker-compose file:
!!! Danger "Security warning"
By using `8080:8080` in the docker port mapping, the API will be available to everyone connecting to the server under the correct port, so others may be able to control your bot.
| `strategy <strategy>` | Get specific Strategy content. **Alpha**
| `available_pairs` | List available backtest data. **Alpha**
| `version` | Show version.
| `sysinfo` | Show informations about the system load.
| `sysinfo` | Show information about the system load.
| `health` | Show bot health (last bot loop).
!!! Warning "Alpha status"
@@ -192,6 +192,11 @@ blacklist
:param add: List of coins to add (example: "BNB/BTC")
cancel_open_order
Cancel open order for trade.
:param trade_id: Cancels open orders for this trade.
count
Return the amount of open trades.
@@ -274,7 +279,6 @@ reload_config
Reload configuration.
show_config
Returns part of the configuration, relevant for trading operations.
start
@@ -320,8 +324,114 @@ version
whitelist
Show the current whitelist.
```
### Message WebSocket
The API Server includes a websocket endpoint for subscribing to RPC messages from the freqtrade Bot.
This can be used to consume real-time data from your bot, such as entry/exit fill messages, whitelist changes, populated indicators for pairs, and more.
This is also used to setup [Producer/Consumer mode](producer-consumer.md) in Freqtrade.
Assuming your rest API is set to `127.0.0.1` on port `8080`, the endpoint is available at `http://localhost:8080/api/v1/message/ws`.
To access the websocket endpoint, the `ws_token` is required as a query parameter in the endpoint URL.
To generate a safe `ws_token` you can run the following code:
``` python
>>> import secrets
>>> secrets.token_urlsafe(25)
'hZ-y58LXyX_HZ8O1cJzVyN6ePWrLpNQv4Q'
```
You would then add that token under `ws_token` in your `api_server` config. Like so:
You can now connect to the endpoint at `http://localhost:8080/api/v1/message/ws?token=hZ-y58LXyX_HZ8O1cJzVyN6ePWrLpNQv4Q`.
!!! Danger "Reuse of example tokens"
Please do not use the above example token. To make sure you are secure, generate a completely new token.
#### Using the WebSocket
Once connected to the WebSocket, the bot will broadcast RPC messages to anyone who is subscribed to them. To subscribe to a list of messages, you must send a JSON request through the WebSocket like the one below. The `data` key must be a list of message type strings.
``` json
{
"type": "subscribe",
"data": ["whitelist", "analyzed_df"] // A list of string message types
}
```
For a list of message types, please refer to the RPCMessageType enum in `freqtrade/enums/rpcmessagetype.py`
Now anytime those types of RPC messages are sent in the bot, you will receive them through the WebSocket as long as the connection is active. They typically take the same form as the request:
``` json
{
"type": "analyzed_df",
"data": {
"key": ["NEO/BTC", "5m", "spot"],
"df": {}, // The dataframe
"la": "2022-09-08 22:14:41.457786+00:00"
}
}
```
#### Reverse Proxy setup
When using [Nginx](https://nginx.org/en/docs/), the following configuration is required for WebSockets to work (Note this configuration is incomplete, it's missing some information and can not be used as is):
Please make sure to replace `<freqtrade_listen_ip>` (and the subsequent port) with the IP and Port matching your configuration/setup.
```
http {
map $http_upgrade $connection_upgrade {
default upgrade;
'' close;
}
#...
server {
#...
location / {
proxy_http_version 1.1;
proxy_pass http://<freqtrade_listen_ip>:8080;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection $connection_upgrade;
proxy_set_header Host $host;
}
}
}
```
To properly configure your reverse proxy (securely), please consult it's documentation for proxying websockets.
- **Traefik**: Traefik supports websockets out of the box, see the [documentation](https://doc.traefik.io/traefik/)
- **Caddy**: Caddy v2 supports websockets out of the box, see the [documentation](https://caddyserver.com/docs/v2-upgrade#proxy)
!!! Tip "SSL certificates"
You can use tools like certbot to setup ssl certificates to access your bot's UI through encrypted connection by using any fo the above reverse proxies.
While this will protect your data in transit, we do not recommend to run the freqtrade API outside of your private network (VPN, SSH tunnel).
### OpenAPI interface
To enable the builtin openAPI interface (Swagger UI), specify `"enable_openapi": true` in the api_server configuration.
@@ -24,7 +24,7 @@ These modes can be configured with these values:
```
!!! Note
Stoploss on exchange is only supported for Binance (stop-loss-limit), Huobi (stop-limit), Kraken (stop-loss-market, stop-loss-limit), FTX (stop limit and stop-market) Gateio (stop-limit), and Kucoin (stop-limit and stop-market) as of now.
Stoploss on exchange is only supported for Binance (stop-loss-limit), Huobi (stop-limit), Kraken (stop-loss-market, stop-loss-limit), Gate (stop-limit), and Kucoin (stop-limit and stop-market) as of now.
<ins>Do not set too low/tight stoploss value if using stop loss on exchange!</ins>
If set to low/tight then you have greater risk of missing fill on the order and stoploss will not work.
@@ -52,6 +52,18 @@ The bot cannot do these every 5 seconds (at each iteration), otherwise it would
So this parameter will tell the bot how often it should update the stoploss order. The default value is 60 (1 minute).
This same logic will reapply a stoploss order on the exchange should you cancel it accidentally.
### stoploss_price_type
!!! Warning "Only applies to futures"
`stoploss_price_type` only applies to futures markets (on exchanges where it's available).
Freqtrade will perform a validation of this setting on startup, failing to start if an invalid setting for your exchange has been selected.
Supported price types are gonna differs between each exchanges. Please check with your exchange on which price types it supports.
Stoploss on exchange on futures markets can trigger on different price types.
The naming for these prices in exchange terminology often varies, but is usually something around "last" (or "contract price" ), "mark" and "index".
Acceptable values for this setting are `"last"`, `"mark"` and `"index"` - which freqtrade will transfer automatically to the corresponding API type, and place the [stoploss on exchange](#stoploss_on_exchange-and-stoploss_on_exchange_limit_ratio) order correspondingly.
### force_exit
`force_exit` is an optional value, which defaults to the same value as `exit` and is used when sending a `/forceexit` command from Telegram or from the Rest API.
@@ -87,7 +99,7 @@ At this stage the bot contains the following stoploss support modes:
2. Trailing stop loss.
3. Trailing stop loss, custom positive loss.
4. Trailing stop loss only once the trade has reached a certain offset.
`enter_tag` is limited to 100 characters, remaining data will be truncated.
!!! Warning
There is only one `enter_tag` column, which is used for both long and short trades.
As a consequence, this column must be treated as "last write wins" (it's just a dataframe column after all).
In fancy situations, where multiple signals collide (or if signals are deactivated again based on different conditions), this can lead to odd results with the wrong tag applied to an entry signal.
These results are a consequence of the strategy overwriting prior tags - where the last tag will "stick" and will be the one freqtrade will use.
## Exit tag
Similar to [Buy Tagging](#buy-tag), you can also specify a sell tag.
@@ -159,6 +159,7 @@ The stoploss price can only ever move upwards - if the stoploss value returned f
The method must return a stoploss value (float / number) as a percentage of the current price.
E.g. If the `current_rate` is 200 USD, then returning `0.02` will set the stoploss price 2% lower, at 196 USD.
During backtesting, `current_rate` (and `current_profit`) are provided against the candle's high (or low for short trades) - while the resulting stoploss is evaluated against the candle's low (or high for short trades).
The absolute value of the return value is used (the sign is ignored), so returning `0.05` or `-0.05` have the same result, a stoploss 5% below the current price.
@@ -643,7 +644,7 @@ This callback is **not** called when there is an open order (either buy or sell)
Additional Buys are ignored once you have reached the maximum amount of extra buys that you have set on `max_entry_position_adjustment`, but the callback is called anyway looking for partial exits.
Position adjustments will always be applied in the direction of the trade, so a positive value will always increase your position (negative values will decrease your position), no matter if it's a long or short trade. Modifications to leverage are not possible.
Position adjustments will always be applied in the direction of the trade, so a positive value will always increase your position (negative values will decrease your position), no matter if it's a long or short trade. Modifications to leverage are not possible, and the stake-amount is assumed to be before applying leverage.
!!! Note "About stake size"
Using fixed stake size means it will be the amount used for the first order, just like without position adjustment.
@@ -658,6 +659,7 @@ Position adjustments will always be applied in the direction of the trade, so a
!!! Warning "Backtesting"
During backtesting this callback is called for each candle in `timeframe` or `timeframe_detail`, so run-time performance will be affected.
This can also cause deviating results between live and backtesting, since backtesting can adjust the trade only once per candle, whereas live could adjust the trade multiple times per candle.
``` python
from freqtrade.persistence import Trade
@@ -772,7 +774,7 @@ class DigDeeperStrategy(IStrategy):
* Sell 150@14\$ -> Avg price: 10\$, total realized profit 950\$, 40% <- *This will be the last "Exit" message*
The total profit for this trade was 950$ on a 3350$ investment (`100@8$ + 100@9$ + 150@11$`). As such - the final relative profit is 28.35% (`950 / 3350`).
@@ -826,7 +828,7 @@ class AwesomeStrategy(IStrategy):
"""
# Limit orders to use and follow SMA200 as price target for the first 10 minutes since entry trigger for BTC/USDT pair.
if pair == 'BTC/USDT' and entry_tag == 'long_sma200' and side == 'long' and (current_time - timedelta(minutes=10) > trade.open_date_utc:
if pair == 'BTC/USDT' and entry_tag == 'long_sma200' and side == 'long' and (current_time - timedelta(minutes=10)) > trade.open_date_utc:
# just cancel the order if it has been filled more than half of the amount
@@ -166,7 +166,7 @@ Additional technical libraries can be installed as necessary, or custom indicato
Most indicators have an instable startup period, in which they are either not available (NaN), or the calculation is incorrect. This can lead to inconsistencies, since Freqtrade does not know how long this instable period should be.
To account for this, the strategy can be assigned the `startup_candle_count` attribute.
This should be set to the maximum number of candles that the strategy requires to calculate stable indicators.
This should be set to the maximum number of candles that the strategy requires to calculate stable indicators. In the case where a user includes higher timeframes with informative pairs, the `startup_candle_count` does not necessarily change. The value is the maximum period (in candles) that any of the informatives timeframes need to compute stable indicators.
In this example strategy, this should be set to 100 (`startup_candle_count = 100`), since the longest needed history is 100 candles.
Edit the method `populate_exit_trend()` into your strategy file to update your exit strategy.
Please note that the exit-signal is only used if `use_exit_signal` is set to true in the configuration.
The exit-signal is only used for exits if `use_exit_signal` is set to true in the configuration.
`use_exit_signal` will not influence [signal collision rules](#colliding-signals) - which will still apply and can prevent entries.
It's important to always return the dataframe without removing/modifying the columns `"open", "high", "low", "close", "volume"`, otherwise these fields would contain something unexpected.
@@ -362,9 +363,9 @@ class AwesomeStrategy(IStrategy):
timeframe = "1d"
timeframe_mins = timeframe_to_minutes(timeframe)
minimal_roi = {
"0": 0.05, # 5% for the first 3 candles
str(timeframe_mins * 3)): 0.02, # 2% after 3 candles
str(timeframe_mins * 6)): 0.01, # 1% After 6 candles
"0": 0.05, # 5% for the first 3 candles
str(timeframe_mins * 3): 0.02, # 2% after 3 candles
str(timeframe_mins * 6): 0.01, # 1% After 6 candles
}
```
@@ -445,15 +446,17 @@ A full sample can be found [in the DataProvider section](#complete-data-provider
??? Note "Alternative candle types"
Informative_pairs can also provide a 3rd tuple element defining the candle type explicitly.
Availability of alternative candle-types will depend on the trading-mode and the exchange. Details about this can be found in the exchange documentation.
Availability of alternative candle-types will depend on the trading-mode and the exchange.
In general, spot pairs cannot be used in futures markets, and futures candles can't be used as informative pairs for spot bots.
Details about this may vary, if they do, this can be found in the exchange documentation.
``` python
def informative_pairs(self):
return [
("ETH/USDT", "5m", ""), # Uses default candletype, depends on trading_mode
("ETH/USDT", "5m", "spot"), # Forces usage of spot candles
Be careful when using dataprovider in backtesting. `historic_ohlcv()` (and `get_pair_dataframe()`
for the backtesting runmode) provides the full time-range in one go,
so please be aware of it and make sure to not "look into the future" to avoid surprises when running in dry/live mode.
In backtesting, `dp.get_pair_dataframe()` behavior differs depending on where it's called.
Within `populate_*()` methods, `dp.get_pair_dataframe()` returns the full timerange. Please make sure to not "look into the future" to avoid surprises when running in dry/live mode.
Within [callbacks](strategy-callbacks.md), you'll get the full timerange up to the current (simulated) candle.
### *get_analyzed_dataframe(pair, timeframe)*
@@ -669,13 +672,13 @@ It can also be used in specific callbacks to get the signal that caused the acti
Debugging a strategy can be time-consuming. Freqtrade offers helper functions to visualize raw data.
The following assumes you work with SampleStrategy, data for 5m timeframe from Binance and have downloaded them into the data directory in the default location.
Please follow the [documentation](https://www.freqtrade.io/en/stable/data-download/) for more details.
## Setup
### Change Working directory to repository root
```python
importos
frompathlibimportPath
# Change directory
# Modify this cell to insure that the output shows the correct path.
# Define all paths relative to the project root shown in the cell output
project_root="somedir/freqtrade"
i=0
try:
os.chdirdir(project_root)
assertPath('LICENSE').is_file()
except:
whilei<4and(notPath('LICENSE').is_file()):
os.chdir(Path(Path.cwd(),'../'))
i+=1
project_root=Path.cwd()
print(Path.cwd())
```
### Configure Freqtrade environment
```python
fromfreqtrade.configurationimportConfiguration
# Customize these according to your needs.
@@ -15,14 +40,14 @@ from freqtrade.configuration import Configuration
# Initialize empty configuration object
config=Configuration.from_files([])
# Optionally (recommended), use existing configuration file
*`ignore_buying_expired_candle_after` -> moved to root level instead of "ask_strategy/exit_pricing"
* Terminology changes
* Sell reasons changed to reflect the new naming of "exit" instead of sells. Be careful in your strategy if you're using `exit_reason` checks and eventually update your strategy.
*`sell_signal` -> `exit_signal`
*`custom_sell` -> `custom_exit`
*`force_sell` -> `force_exit`
*`emergency_sell` -> `emergency_exit`
* Order pricing
*`bid_strategy` -> `entry_pricing`
*`ask_strategy` -> `exit_pricing`
*`ask_last_balance` -> `price_last_balance`
*`bid_last_balance` -> `price_last_balance`
* Webhook terminology changed from "sell" to "exit", and from "buy" to entry
*`webhookbuy` -> `webhookentry`
*`webhookbuyfill` -> `webhookentryfill`
*`webhookbuycancel` -> `webhookentrycancel`
*`webhooksell` -> `webhookexit`
*`webhooksellfill` -> `webhookexitfill`
*`webhooksellcancel` -> `webhookexitcancel`
*`webhookbuy` -> `entry`
*`webhookbuyfill` -> `entry_fill`
*`webhookbuycancel` -> `entry_cancel`
*`webhooksell` -> `exit`
*`webhooksellfill` -> `exit_fill`
*`webhooksellcancel` -> `exit_cancel`
* Telegram notification settings
*`buy` -> `entry`
*`buy_fill` -> `entry_fill`
@@ -332,8 +338,8 @@ After:
``` python hl_lines="2 3"
order_time_in_force: Dict = {
"entry": "gtc",
"exit": "gtc",
"entry": "GTC",
"exit": "GTC",
}
```
@@ -443,6 +449,7 @@ Please refer to the [pricing documentation](configuration.md#prices-used-for-ord
"use_order_book": true,
"order_book_top": 1,
"bid_last_balance": 0.0
"ignore_buying_expired_candle_after": 120
}
}
```
@@ -466,6 +473,258 @@ after:
"use_order_book": true,
"order_book_top": 1,
"price_last_balance": 0.0
}
},
"ignore_buying_expired_candle_after": 120
}
```
## FreqAI strategy
The `populate_any_indicators()` method has been split into `feature_engineering_expand_all()`, `feature_engineering_expand_basic()`, `feature_engineering_standard()` and`set_freqai_targets()`.
For each new function, the pair (and timeframe where necessary) will be automatically added to the column.
As such, the definition of features becomes much simpler with the new logic.
For a full explanation of each method, please go to the corresponding [freqAI documentation page](freqai-feature-engineering.md#defining-the-features)
`reload` allows you to disable reload-buttons on selected messages.
## Create a custom keyboard (command shortcut buttons)
@@ -159,26 +162,33 @@ official commands. You can ask at any moment for help with `/help`.
| Command | Description |
|----------|-------------|
| **System commands**
| `/start` | Starts the trader
| `/stop` | Stops the trader
| `/stopbuy | /stopentry` | Stops the trader from opening new trades. Gracefully closes open trades according to their rules.
| `/reload_config` | Reloads the configuration file
| `/show_config` | Shows part of the current configuration with relevant settings to operation
| `/logs [limit]` | Show last log messages.
| `/help` | Show help message
| `/version` | Show version
| **Status** |
| `/status` | Lists all open trades
| `/status <trade_id>` | Lists one or more specific trade. Separate multiple <trade_id> with a blank space.
| `/status table` | List all open trades in a table format. Pending buy orders are marked with an asterisk (*) Pending sell orders are marked with a double asterisk (**)
| `/trades [limit]` | List all recently closed trades in a table format.
| `/delete <trade_id>` | Delete a specific trade from the Database. Tries to close open orders. Requires manual handling of this trade on the exchange.
| `/count` | Displays number of trades used and available
| `/locks` | Show currently locked pairs.
| `/unlock <pairorlock_id>` | Remove the lock for this pair (or for this lock id).
| `/profit [<n>]` | Display a summary of your profit/loss from close trades and some stats about your performance, over the last n days (all trades by default)
| **Modify Trade states** |
| `/forceexit <trade_id> | /fx <tradeid>` | Instantly exits the given trade (Ignoring `minimum_roi`).
| `/forceexit all | /fx all` | Instantly exits all open trades (Ignoring `minimum_roi`).
| `/fx` | alias for `/forceexit`
| `/forcelong <pair> [rate]` | Instantly buys the given pair. Rate is optional and only applies to limit orders. (`force_entry_enable` must be set to True)
| `/forceshort <pair> [rate]` | Instantly shorts the given pair. Rate is optional and only applies to limit orders. This will only work on non-spot markets. (`force_entry_enable` must be set to True)
| `/delete <trade_id>` | Delete a specific trade from the Database. Tries to close open orders. Requires manual handling of this trade on the exchange.
| `/cancel_open_order <trade_id> | /coo <trade_id>` | Cancel an open order for a trade.
| **Metrics** |
| `/profit [<n>]` | Display a summary of your profit/loss from close trades and some stats about your performance, over the last n days (all trades by default)
| `/performance` | Show performance of each finished trade grouped by pair
| `/balance` | Show account balance per currency
| `/daily <n>` | Shows profit or loss per day, over the last n days (n defaults to 7)
@@ -190,8 +200,7 @@ official commands. You can ask at any moment for help with `/help`.
| `/whitelist [sorted] [baseonly]` | Show the current whitelist. Optionally display in alphabetical order and/or with just the base currency of each pairing.
| `/blacklist [pair]` | Show the current blacklist, or adds a pair to the blacklist.
| `/edge` | Show validated pairs by Edge if it is enabled.
A position freqtrade enters is stored in a `Trade` object - which is persisted to the database.
It's a core concept of freqtrade - and something you'll come across in many sections of the documentation, which will most likely point you to this location.
It will be passed to the strategy in many [strategy callbacks](strategy-callbacks.md). The object passed to the strategy cannot be modified directly. Indirect modifications may occur based on callback results.
## Trade - Available attributes
The following attributes / properties are available for each individual trade - and can be used with `trade.<property>` (e.g. `trade.pair`).
| Attribute | DataType | Description |
|------------|-------------|-------------|
`pair`| string | Pair of this trade
`is_open`| boolean | Is the trade currently open, or has it been concluded
`open_rate`| float | Rate this trade was entered at (Avg. entry rate in case of trade-adjustments)
`close_rate`| float | Close rate - only set when is_open = False
`stake_amount`| float | Amount in Stake (or Quote) currency.
`amount`| float | Amount in Asset / Base currency that is currently owned.
`open_date`| datetime | Timestamp when trade was opened **use `open_date_utc` instead**
`open_date_utc`| datetime | Timestamp when trade was opened - in UTC
`close_date`| datetime | Timestamp when trade was closed **use `close_date_utc` instead**
`close_date_utc`| datetime | Timestamp when trade was closed - in UTC
`close_profit`| float | Relative profit at the time of trade closure. `0.01` == 1%
`close_profit_abs`| float | Absolute profit (in stake currency) at the time of trade closure.
`leverage` | float | Leverage used for this trade - defaults to 1.0 in spot markets.
`enter_tag`| string | Tag provided on entry via the `enter_tag` column in the dataframe
`is_short` | boolean | True for short trades, False otherwise
`orders` | Order[] | List of order objects attached to this trade (includes both filled and cancelled orders)
`date_last_filled_utc` | datetime | Time of the last filled order
`entry_side` | "buy" / "sell" | Order Side the trade was entered
`exit_side` | "buy" / "sell" | Order Side that will result in a trade exit / position reduction.
`trade_direction` | "long" / "short" | Trade direction in text - long or short.
`nr_of_successful_entries` | int | Number of successful (filled) entry orders
`nr_of_successful_exits` | int | Number of successful (filled) exit orders
## Class methods
The following are class methods - which return generic information, and usually result in an explicit query against the database.
They can be used as `Trade.<method>` - e.g. `open_trades = Trade.get_open_trade_count()`
!!! Warning "Backtesting/hyperopt"
Most methods will work in both backtesting / hyperopt and live/dry modes.
During backtesting, it's limited to usage in [strategy callbacks](strategy-callbacks.md). Usage in `populate_*()` methods is not supported and will result in wrong results.
### get_trades_proxy
When your strategy needs some information on existing (open or close) trades - it's best to use `Trade.get_trades_proxy()`.
`get_trades_proxy()` supports the following keyword arguments. All arguments are optional - calling `get_trades_proxy()` without arguments will return a list of all trades in the database.
* `pair` e.g. `pair='ETH/USDT'`
* `is_open` e.g. `is_open=False`
* `open_date` e.g. `open_date=current_date - timedelta(days=2)`
* `close_date` e.g. `close_date=current_date - timedelta(days=5)`
### get_open_trade_count
Get the number of currently open trades
``` python
from freqtrade.persistence import Trade
# ...
open_trades = Trade.get_open_trade_count()
```
### get_total_closed_profit
Retrieve the total profit the bot has generated so far.
Aggregates `close_profit_abs` for all closed trades.
``` python
from freqtrade.persistence import Trade
# ...
profit = Trade.get_total_closed_profit()
```
### total_open_trades_stakes
Retrieve the total stake_amount that's currently in trades.
``` python
from freqtrade.persistence import Trade
# ...
profit = Trade.total_open_trades_stakes()
```
### get_overall_performance
Retrieve the overall performance - similar to the `/performance` telegram command.
``` python
from freqtrade.persistence import Trade
# ...
if self.config['runmode'].value in ('live', 'dry_run'):
performance = Trade.get_overall_performance()
```
Sample return value: ETH/BTC had 5 trades, with a total profit of 1.5% (ratio of 0.015).
``` json
{"pair": "ETH/BTC", "profit": 0.015, "count": 5}
```
## Order Object
An `Order` object represents an order on the exchange (or a simulated order in dry-run mode).
An `Order` object will always be tied to it's corresponding [`Trade`](#trade-object), and only really makes sense in the context of a trade.
### Order - Available attributes
an Order object is typically attached to a trade.
Most properties here can be None as they are dependant on the exchange response.
| Attribute | DataType | Description |
|------------|-------------|-------------|
`trade` | Trade | Trade object this order is attached to
`ft_pair` | string | Pair this order is for
`ft_is_open` | boolean | is the order filled?
`order_type` | string | Order type as defined on the exchange - usually market, limit or stoploss
`status` | string | Status as defined by ccxt. Usually open, closed, expired or canceled
`side` | string | Buy or Sell
`price` | float | Price the order was placed at
`average` | float | Average price the order filled at
`amount` | float | Amount in base currency
`filled` | float | Filled amount (in base currency)
`remaining` | float | Remaining amount
`cost` | float | Cost of the order - usually average * filled
`order_date` | datetime | Order creation date **use `order_date_utc` instead**
`order_date_utc` | datetime | Order creation date (in UTC)
`order_fill_date` | datetime | Order fill date **use `order_fill_utc` instead**
`order_fill_date_utc` | datetime | Order fill date
@@ -6,14 +6,14 @@ To update your freqtrade installation, please use one of the below methods, corr
Breaking changes / changed behavior will be documented in the changelog that is posted alongside every release.
For the develop branch, please follow PR's to avoid being surprised by changes.
## docker-compose
## docker
!!! Note "Legacy installations using the `master` image"
We're switching from master to stable for the release Images - please adjust your docker-file and replace `freqtradeorg/freqtrade:master` with `freqtradeorg/freqtrade:stable`
``` bash
docker-compose pull
docker-compose up -d
dockercompose pull
dockercompose up -d
```
## Installation via setup script
@@ -37,3 +37,12 @@ pip install -e .
# Ensure freqUI is at the latest version
freqtrade install-ui
```
### Problems updating
Update-problems usually come missing dependencies (you didn't follow the above instructions) - or from updated dependencies, which fail to install (for example TA-lib).
Please refer to the corresponding installation sections (common problems linked below)
Common problems and their solutions:
* [ta-lib update on windows](windows_installation.md#2-install-ta-lib)
Use the `list-freqaimodels` subcommand to see all freqAI models available.
This subcommand is useful for finding problems in your environment with loading freqAI models: modules with models that contain errors and failed to load are printed in red (LOAD FAILED), while models with duplicate names are printed in yellow (DUPLICATE NAME).
-v, --verbose Verbose mode (-vv for more, -vvv to get all messages).
-c PATH, --config PATH
Specify configuration file (default:
@@ -615,7 +652,7 @@ Common arguments:
You can also use webserver mode via docker.
Starting a one-off container requires the configuration of the port explicitly, as ports are not exposed by default.
You can use `docker-compose run --rm -p 127.0.0.1:8080:8080 freqtrade webserver` to start a one-off container that'll be removed once you stop it. This assumes that port 8080 is still available and no other bot is running on that port.
You can use `dockercompose run --rm -p 127.0.0.1:8080:8080 freqtrade webserver` to start a one-off container that'll be removed once you stop it. This assumes that port 8080 is still available and no other bot is running on that port.
Alternatively, you can reconfigure the docker-compose file to have the command updated:
@@ -625,7 +662,7 @@ Alternatively, you can reconfigure the docker-compose file to have the command u
--config /freqtrade/user_data/config.json
```
You can now use `docker-compose up` to start the webserver.
You can now use `dockercompose up` to start the webserver.
This assumes that the configuration has a webserver enabled and configured for docker (listening port = `0.0.0.0`).
@@ -88,17 +88,30 @@ Optional parameters are available to enable automatic retries for webhook messag
"url":"https://<YOURHOOKURL>",
"retries":3,
"retry_delay":0.2,
"webhookstatus":{
"status":{
"status":"Status: {status}"
}
},
```
Custom messages can be sent to Webhook endpoints via the `self.dp.send_msg()` function from within the strategy. To enable this, set the `allow_custom_messages` option to `true`:
```json
"webhook":{
"enabled":true,
"url":"https://<YOURHOOKURL>",
"allow_custom_messages":true,
"strategy_msg":{
"status":"StrategyMessage: {msg}"
}
},
```
Different payloads can be configured for different events. Not all fields are necessary, but you should configure at least one of the dicts, otherwise the webhook will never be called.
### Webhookentry
### Entry
The fields in `webhook.webhookentry` are filled when the bot executes a long/short. Parameters are filled using string.format.
The fields in `webhook.entry` are filled when the bot executes a long/short. Parameters are filled using string.format.
Possible parameters are:
*`trade_id`
@@ -118,9 +131,9 @@ Possible parameters are:
*`current_rate`
*`enter_tag`
### Webhookentrycancel
### Entrycancel
The fields in `webhook.webhookentrycancel` are filled when the bot cancels a long/short order. Parameters are filled using string.format.
The fields in `webhook.entry_cancel` are filled when the bot cancels a long/short order. Parameters are filled using string.format.
Possible parameters are:
*`trade_id`
@@ -139,9 +152,9 @@ Possible parameters are:
*`current_rate`
*`enter_tag`
### Webhookentryfill
### Entryfill
The fields in `webhook.webhookentryfill` are filled when the bot filled a long/short order. Parameters are filled using string.format.
The fields in `webhook.entry_fill` are filled when the bot filled a long/short order. Parameters are filled using string.format.
Possible parameters are:
*`trade_id`
@@ -160,9 +173,9 @@ Possible parameters are:
*`current_rate`
*`enter_tag`
### Webhookexit
### Exit
The fields in `webhook.webhookexit` are filled when the bot exits a trade. Parameters are filled using string.format.
The fields in `webhook.exit` are filled when the bot exits a trade. Parameters are filled using string.format.
Possible parameters are:
*`trade_id`
@@ -184,9 +197,9 @@ Possible parameters are:
*`open_date`
*`close_date`
### Webhookexitfill
### Exitfill
The fields in `webhook.webhookexitfill` are filled when the bot fills a exit order (closes a Trade). Parameters are filled using string.format.
The fields in `webhook.exit_fill` are filled when the bot fills a exit order (closes a Trade). Parameters are filled using string.format.
Possible parameters are:
*`trade_id`
@@ -209,9 +222,9 @@ Possible parameters are:
*`open_date`
*`close_date`
### Webhookexitcancel
### Exitcancel
The fields in `webhook.webhookexitcancel` are filled when the bot cancels a exit order. Parameters are filled using string.format.
The fields in `webhook.exit_cancel` are filled when the bot cancels a exit order. Parameters are filled using string.format.
Possible parameters are:
*`trade_id`
@@ -234,9 +247,9 @@ Possible parameters are:
*`open_date`
*`close_date`
### Webhookstatus
### Status
The fields in `webhook.webhookstatus` are used for regular status messages (Started / Stopped / ...). Parameters are filled using string.format.
The fields in `webhook.status` are used for regular status messages (Started / Stopped / ...). Parameters are filled using string.format.
The only possible value here is `{status}`.
@@ -280,7 +293,6 @@ You can configure this as follows:
}
```
The above represents the default (`exit_fill` and `entry_fill` are optional and will default to the above configuration) - modifications are obviously possible.
Available fields correspond to the fields for webhooks and are documented in the corresponding webhook sections.
@@ -288,3 +300,13 @@ Available fields correspond to the fields for webhooks and are documented in the
The notifications will look as follows by default.
Custom messages can be sent from a strategy to Discord endpoints via the dataprovider.send_msg() function. To enable this, set the `allow_custom_messages` option to `true`:
We **strongly** recommend that Windows users use [Docker](docker_quickstart.md) as this will work much easier and smoother (also more secure).
If that is not possible, try using the Windows Linux subsystem (WSL) - for which the Ubuntu instructions should work.
Otherwise, try the instructions below.
Otherwise, please follow the instructions below.
## Install freqtrade manually
!!! Note
Make sure to use 64bit Windows and 64bit Python to avoid problems with backtesting or hyperopt due to the memory constraints 32bit applications have under Windows.
!!! Note "64bit Python version"
Please make sure to use 64bit Windows and 64bit Python to avoid problems with backtesting or hyperopt due to the memory constraints 32bit applications have under Windows.
32bit python versions are no longer supported under Windows.
!!! Hint
Using the [Anaconda Distribution](https://www.anaconda.com/distribution/) under Windows can greatly help with installation problems. Check out the [Anaconda installation section](installation.md#Anaconda) in this document for more information.
Using the [Anaconda Distribution](https://www.anaconda.com/distribution/) under Windows can greatly help with installation problems. Check out the [Anaconda installation section](installation.md#installation-with-conda) in the documentation for more information.
Install ta-lib according to the [ta-lib documentation](https://github.com/mrjbq7/ta-lib#windows).
As compiling from source on windows has heavy dependencies (requires a partial visual studio installation), there is also a repository of unofficial pre-compiled windows Wheels [here](https://www.lfd.uci.edu/~gohlke/pythonlibs/#ta-lib), which need to be downloaded and installed using `pip install TA_Lib-0.4.24-cp38-cp38-win_amd64.whl` (make sure to use the version matching your python version).
As compiling from source on windows has heavy dependencies (requires a partial visual studio installation), there is also a repository of unofficial pre-compiled windows Wheels [here](https://www.lfd.uci.edu/~gohlke/pythonlibs/#ta-lib), which need to be downloaded and installed using `pip install TA_Lib-0.4.25-cp38-cp38-win_amd64.whl` (make sure to use the version matching your python version).
Freqtrade provides these dependencies for the latest 3 Python versions (3.8, 3.9 and 3.10) and for 64bit Windows.
Other versions must be downloaded from the above link.
@@ -34,7 +35,7 @@ python -m venv .env
.env\Scripts\activate.ps1
# optionally install ta-lib from wheel
# Eventually adjust the below filename to match the downloaded wheel
:param prefix: Prefix to consider (usually FREQTRADE__)
:return: Nested dict based on available and relevant variables.
"""
no_convert=['CHAT_ID']
no_convert=['CHAT_ID','PASSWORD']
relevant_vars:Dict[str,Any]={}
forenv_var,valinsorted(env_dict.items()):
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
