adelorenzo/stable
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge branch 'freqtrade:develop' into strategy_utils

Browse Source
This commit is contained in:
hippocritical
2023-03-10 09:18:44 +01:00
committed by GitHub
parent 5f8202e1b5 5b2a291109
commit 5a467eb969
17 changed files with 98 additions and 58 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
freqtrade/constants.py
View File

@@ -588,6 +588,7 @@ CONF_SCHEMA = {
"rl_config": {
"type": "object",
"properties": {
"drop_ohlc_from_features": {"type": "boolean", "default": False},
"train_cycles": {"type": "integer"},
"max_trade_duration_candles": {"type": "integer"},
"add_state_info": {"type": "boolean", "default": False},

1
freqtrade/exchange/exchange.py
View File

@@ -69,6 +69,7 @@ class Exchange:
# Check https://github.com/ccxt/ccxt/issues/10767 for removal of ohlcv_volume_currency
"ohlcv_volume_currency": "base", # "base" or "quote"
"tickers_have_quoteVolume": True,
"tickers_have_bid_ask": True, # bid / ask empty for fetch_tickers
"tickers_have_price": True,
"trades_pagination": "time", # Possible are "time" or "id"
"trades_pagination_arg": "since",

1
freqtrade/exchange/gate.py
View File

@@ -32,6 +32,7 @@ class Gate(Exchange):
_ft_has_futures: Dict = {
"needs_trading_fees": True,
"tickers_have_bid_ask": False,
"fee_cost_in_contracts": False, # Set explicitly to false for clarity
"order_props_in_contracts": ['amount', 'filled', 'remaining'],
"stop_price_type_field": "price_type",

29
freqtrade/freqai/RL/BaseReinforcementLearningModel.py
View File

@@ -114,6 +114,7 @@ class BaseReinforcementLearningModel(IFreqaiModel):
# normalize all data based on train_dataset only
prices_train, prices_test = self.build_ohlc_price_dataframes(dk.data_dictionary, pair, dk)
data_dictionary = dk.normalize_data(data_dictionary)
# data cleaning/analysis
@@ -148,12 +149,8 @@ class BaseReinforcementLearningModel(IFreqaiModel):
env_info = self.pack_env_dict(dk.pair)
self.train_env = self.MyRLEnv(df=train_df,
prices=prices_train,
**env_info)
self.eval_env = Monitor(self.MyRLEnv(df=test_df,
prices=prices_test,
**env_info))
self.train_env = self.MyRLEnv(df=train_df, prices=prices_train, **env_info)
self.eval_env = Monitor(self.MyRLEnv(df=test_df, prices=prices_test, **env_info))
self.eval_callback = EvalCallback(self.eval_env, deterministic=True,
render=False, eval_freq=len(train_df),
best_model_save_path=str(dk.data_path))
@@ -238,6 +235,9 @@ class BaseReinforcementLearningModel(IFreqaiModel):
filtered_dataframe, _ = dk.filter_features(
unfiltered_df, dk.training_features_list, training_filter=False
)
filtered_dataframe = self.drop_ohlc_from_df(filtered_dataframe, dk)
filtered_dataframe = dk.normalize_data_from_metadata(filtered_dataframe)
dk.data_dictionary["prediction_features"] = filtered_dataframe
@@ -285,7 +285,6 @@ class BaseReinforcementLearningModel(IFreqaiModel):
train_df = data_dictionary["train_features"]
test_df = data_dictionary["test_features"]
# %-raw_volume_gen_shift-2_ETH/USDT_1h
# price data for model training and evaluation
tf = self.config['timeframe']
rename_dict = {'%-raw_open': 'open', '%-raw_low': 'low',
@@ -318,8 +317,24 @@ class BaseReinforcementLearningModel(IFreqaiModel):
prices_test.rename(columns=rename_dict, inplace=True)
prices_test.reset_index(drop=True)
train_df = self.drop_ohlc_from_df(train_df, dk)
test_df = self.drop_ohlc_from_df(test_df, dk)
return prices_train, prices_test
def drop_ohlc_from_df(self, df: DataFrame, dk: FreqaiDataKitchen):
"""
Given a dataframe, drop the ohlc data
"""
drop_list = ['%-raw_open', '%-raw_low', '%-raw_high', '%-raw_close']
if self.rl_config["drop_ohlc_from_features"]:
df.drop(drop_list, axis=1, inplace=True)
feature_list = dk.training_features_list
dk.training_features_list = [e for e in feature_list if e not in drop_list]
return df
def load_model_from_disk(self, dk: FreqaiDataKitchen) -> Any:
"""
Can be used by user if they are trying to limit_ram_usage *and*

7
freqtrade/plugins/pairlist/SpreadFilter.py
View File

@@ -5,6 +5,7 @@ import logging
from typing import Any, Dict, Optional
from freqtrade.constants import Config
from freqtrade.exceptions import OperationalException
from freqtrade.exchange.types import Ticker
from freqtrade.plugins.pairlist.IPairList import IPairList
@@ -22,6 +23,12 @@ class SpreadFilter(IPairList):
self._max_spread_ratio = pairlistconfig.get('max_spread_ratio', 0.005)
self._enabled = self._max_spread_ratio != 0
if not self._exchange.get_option('tickers_have_bid_ask'):
raise OperationalException(
f"{self.name} requires exchange to have bid/ask data for tickers, "
"which is not available for the selected exchange / trading mode."
)
@property
def needstickers(self) -> bool:
"""

20
freqtrade/strategy/strategy_helper.py
View File

@@ -86,37 +86,41 @@ def merge_informative_pair(dataframe: pd.DataFrame, informative: pd.DataFrame,
def stoploss_from_open(
open_relative_stop: float,
current_profit: float,
is_short: bool = False
is_short: bool = False,
leverage: float = 1.0
) -> float:
"""
Given the current profit, and a desired stop loss value relative to the open price,
Given the current profit, and a desired stop loss value relative to the trade entry price,
return a stop loss value that is relative to the current price, and which can be
returned from `custom_stoploss`.
The requested stop can be positive for a stop above the open price, or negative for
a stop below the open price. The return value is always >= 0.
`open_relative_stop` will be considered as adjusted for leverage if leverage is provided..
Returns 0 if the resulting stop price would be above/below (longs/shorts) the current price
:param open_relative_stop: Desired stop loss percentage relative to open price
:param open_relative_stop: Desired stop loss percentage, relative to the open price,
adjusted for leverage
:param current_profit: The current profit percentage
:param is_short: When true, perform the calculation for short instead of long
:param leverage: Leverage to use for the calculation
:return: Stop loss value relative to current price
"""
# formula is undefined for current_profit -1 (longs) or 1 (shorts), return maximum value
if (current_profit == -1 and not is_short) or (is_short and current_profit == 1):
_current_profit = current_profit / leverage
if (_current_profit == -1 and not is_short) or (is_short and _current_profit == 1):
return 1
if is_short is True:
stoploss = -1 + ((1 - open_relative_stop) / (1 - current_profit))
stoploss = -1 + ((1 - open_relative_stop / leverage) / (1 - _current_profit))
else:
stoploss = 1 - ((1 + open_relative_stop) / (1 + current_profit))
stoploss = 1 - ((1 + open_relative_stop / leverage) / (1 + _current_profit))
# negative stoploss values indicate the requested stop price is higher/lower
# (long/short) than the current price
return max(stoploss, 0.0)
return max(stoploss * leverage, 0.0)
def stoploss_from_absolute(stop_rate: float, current_rate: float, is_short: bool = False) -> float: