stable/freqtrade/data/btanalysis.py

439 lines
18 KiB
Python
Raw Normal View History

2019-03-07 20:20:32 +00:00
"""
Helpers when analyzing backtest data
"""
2019-06-16 08:41:05 +00:00
import logging
2019-03-07 20:20:32 +00:00
from pathlib import Path
from typing import Any, Dict, List, Optional, Tuple, Union
2019-03-07 20:20:32 +00:00
import numpy as np
import pandas as pd
2020-06-28 07:27:19 +00:00
from freqtrade.constants import LAST_BT_RESULT_FN
2019-03-07 20:20:32 +00:00
from freqtrade.misc import json_load
from freqtrade.persistence import LocalTrade, Trade, init_db
2019-06-16 08:41:05 +00:00
2020-09-28 17:39:41 +00:00
2019-06-16 08:41:05 +00:00
logger = logging.getLogger(__name__)
2019-03-07 20:20:32 +00:00
# Old format - maybe remove?
BT_DATA_COLUMNS_OLD = ["pair", "profit_percent", "open_date", "close_date", "index",
"trade_duration", "open_rate", "close_rate", "open_at_end", "sell_reason"]
2021-08-16 12:16:24 +00:00
# Mid-term format, created by BacktestResult Named Tuple
BT_DATA_COLUMNS_MID = ['pair', 'profit_percent', 'open_date', 'close_date', 'trade_duration',
'open_rate', 'close_rate', 'open_at_end', 'sell_reason', 'fee_open',
'fee_close', 'amount', 'profit_abs', 'profit_ratio']
# Newest format
BT_DATA_COLUMNS = ['pair', 'stake_amount', 'amount', 'open_date', 'close_date',
2021-01-24 08:56:27 +00:00
'open_rate', 'close_rate',
'fee_open', 'fee_close', 'trade_duration',
'profit_ratio', 'profit_abs', 'sell_reason',
'initial_stop_loss_abs', 'initial_stop_loss_ratio', 'stop_loss_abs',
2021-07-21 13:05:35 +00:00
'stop_loss_ratio', 'min_rate', 'max_rate', 'is_open', 'buy_tag']
2019-03-07 20:20:32 +00:00
2020-09-27 14:50:22 +00:00
def get_latest_optimize_filename(directory: Union[Path, str], variant: str) -> str:
2020-06-26 07:34:18 +00:00
"""
Get latest backtest export based on '.last_result.json'.
:param directory: Directory to search for last result
2020-09-27 14:50:22 +00:00
:param variant: 'backtest' or 'hyperopt' - the method to return
2020-06-26 07:34:18 +00:00
:return: string containing the filename of the latest backtest result
:raises: ValueError in the following cases:
* Directory does not exist
* `directory/.last_result.json` does not exist
* `directory/.last_result.json` has the wrong content
"""
if isinstance(directory, str):
directory = Path(directory)
if not directory.is_dir():
2020-06-27 04:46:54 +00:00
raise ValueError(f"Directory '{directory}' does not exist.")
2020-06-28 07:27:19 +00:00
filename = directory / LAST_BT_RESULT_FN
2020-06-26 07:34:18 +00:00
if not filename.is_file():
raise ValueError(
f"Directory '{directory}' does not seem to contain backtest statistics yet.")
2020-06-26 07:34:18 +00:00
with filename.open() as file:
data = json_load(file)
2020-09-27 14:50:22 +00:00
if f'latest_{variant}' not in data:
2020-06-28 07:27:19 +00:00
raise ValueError(f"Invalid '{LAST_BT_RESULT_FN}' format.")
2020-06-26 07:34:18 +00:00
2020-09-27 14:50:22 +00:00
return data[f'latest_{variant}']
def get_latest_backtest_filename(directory: Union[Path, str]) -> str:
"""
Get latest backtest export based on '.last_result.json'.
:param directory: Directory to search for last result
:return: string containing the filename of the latest backtest result
:raises: ValueError in the following cases:
* Directory does not exist
* `directory/.last_result.json` does not exist
* `directory/.last_result.json` has the wrong content
"""
return get_latest_optimize_filename(directory, 'backtest')
def get_latest_hyperopt_filename(directory: Union[Path, str]) -> str:
"""
Get latest hyperopt export based on '.last_result.json'.
:param directory: Directory to search for last result
:return: string containing the filename of the latest hyperopt result
:raises: ValueError in the following cases:
* Directory does not exist
* `directory/.last_result.json` does not exist
* `directory/.last_result.json` has the wrong content
"""
try:
return get_latest_optimize_filename(directory, 'hyperopt')
except ValueError:
# Return default (legacy) pickle filename
return 'hyperopt_results.pickle'
def get_latest_hyperopt_file(directory: Union[Path, str], predef_filename: str = None) -> Path:
2020-09-27 14:50:22 +00:00
"""
Get latest hyperopt export based on '.last_result.json'.
:param directory: Directory to search for last result
:return: string containing the filename of the latest hyperopt result
:raises: ValueError in the following cases:
* Directory does not exist
* `directory/.last_result.json` does not exist
* `directory/.last_result.json` has the wrong content
"""
2020-09-27 17:40:55 +00:00
if isinstance(directory, str):
directory = Path(directory)
if predef_filename:
return directory / predef_filename
2020-09-27 14:50:22 +00:00
return directory / get_latest_hyperopt_filename(directory)
2020-06-26 07:34:18 +00:00
2020-06-26 05:46:59 +00:00
def load_backtest_stats(filename: Union[Path, str]) -> Dict[str, Any]:
"""
Load backtest statistics file.
:param filename: pathlib.Path object, or string pointing to the file.
:return: a dictionary containing the resulting file.
"""
if isinstance(filename, str):
filename = Path(filename)
if filename.is_dir():
2020-06-28 07:51:49 +00:00
filename = filename / get_latest_backtest_filename(filename)
2020-06-26 05:46:59 +00:00
if not filename.is_file():
raise ValueError(f"File {filename} does not exist.")
logger.info(f"Loading backtest result from {filename}")
2020-06-26 05:46:59 +00:00
with filename.open() as file:
data = json_load(file)
return data
2020-06-27 07:56:37 +00:00
def load_backtest_data(filename: Union[Path, str], strategy: Optional[str] = None) -> pd.DataFrame:
2019-03-07 20:20:32 +00:00
"""
Load backtest data file.
:param filename: pathlib.Path object, or string pointing to a file or directory
2020-06-27 07:56:37 +00:00
:param strategy: Strategy to load - mainly relevant for multi-strategy backtests
Can also serve as protection to load the correct result.
2019-06-23 20:10:37 +00:00
:return: a dataframe with the analysis results
:raise: ValueError if loading goes wrong.
2019-03-07 20:20:32 +00:00
"""
data = load_backtest_stats(filename)
if not isinstance(data, list):
2020-06-27 07:56:37 +00:00
# new, nested format
if 'strategy' not in data:
2020-06-27 07:56:37 +00:00
raise ValueError("Unknown dataformat.")
if not strategy:
if len(data['strategy']) == 1:
strategy = list(data['strategy'].keys())[0]
else:
raise ValueError("Detected backtest result with more than one strategy. "
"Please specify a strategy.")
if strategy not in data['strategy']:
raise ValueError(f"Strategy {strategy} not available in the backtest result.")
data = data['strategy'][strategy]['trades']
df = pd.DataFrame(data)
if not df.empty:
df['open_date'] = pd.to_datetime(df['open_date'],
utc=True,
infer_datetime_format=True
)
df['close_date'] = pd.to_datetime(df['close_date'],
utc=True,
infer_datetime_format=True
)
else:
# old format - only with lists.
df = pd.DataFrame(data, columns=BT_DATA_COLUMNS_OLD)
if not df.empty:
df['open_date'] = pd.to_datetime(df['open_date'],
unit='s',
utc=True,
infer_datetime_format=True
)
df['close_date'] = pd.to_datetime(df['close_date'],
unit='s',
utc=True,
infer_datetime_format=True
)
# Create compatibility with new format
df['profit_abs'] = df['close_rate'] - df['open_rate']
if not df.empty:
if 'profit_ratio' not in df.columns:
df['profit_ratio'] = df['profit_percent']
df = df.sort_values("open_date").reset_index(drop=True)
2019-03-07 20:20:32 +00:00
return df
def analyze_trade_parallelism(results: pd.DataFrame, timeframe: str) -> pd.DataFrame:
2019-03-07 20:20:32 +00:00
"""
Find overlapping trades by expanding each trade once per period it was open
and then counting overlaps.
2019-03-07 20:20:32 +00:00
:param results: Results Dataframe - can be loaded
:param timeframe: Timeframe used for backtest
:return: dataframe with open-counts per time-period in timeframe
2019-03-07 20:20:32 +00:00
"""
from freqtrade.exchange import timeframe_to_minutes
timeframe_min = timeframe_to_minutes(timeframe)
dates = [pd.Series(pd.date_range(row[1]['open_date'], row[1]['close_date'],
freq=f"{timeframe_min}min"))
for row in results[['open_date', 'close_date']].iterrows()]
2019-03-07 20:20:32 +00:00
deltas = [len(x) for x in dates]
dates = pd.Series(pd.concat(dates).values, name='date')
df2 = pd.DataFrame(np.repeat(results.values, deltas, axis=0), columns=results.columns)
df2 = pd.concat([dates, df2], axis=1)
df2 = df2.set_index('date')
df_final = df2.resample(f"{timeframe_min}min")[['pair']].count()
df_final = df_final.rename({'pair': 'open_trades'}, axis=1)
return df_final
def evaluate_result_multi(results: pd.DataFrame, timeframe: str,
max_open_trades: int) -> pd.DataFrame:
"""
Find overlapping trades by expanding each trade once per period it was open
and then counting overlaps
:param results: Results Dataframe - can be loaded
:param timeframe: Frequency used for the backtest
:param max_open_trades: parameter max_open_trades used during backtest run
:return: dataframe with open-counts per time-period in freq
"""
df_final = analyze_trade_parallelism(results, timeframe)
return df_final[df_final['open_trades'] > max_open_trades]
2019-06-16 08:41:05 +00:00
def trade_list_to_dataframe(trades: List[LocalTrade]) -> pd.DataFrame:
"""
Convert list of Trade objects to pandas Dataframe
:param trades: List of trade objects
:return: Dataframe with BT_DATA_COLUMNS
"""
df = pd.DataFrame.from_records([t.to_json() for t in trades], columns=BT_DATA_COLUMNS)
if len(df) > 0:
df.loc[:, 'close_date'] = pd.to_datetime(df['close_date'], utc=True)
df.loc[:, 'open_date'] = pd.to_datetime(df['open_date'], utc=True)
2021-01-24 08:56:27 +00:00
df.loc[:, 'close_rate'] = df['close_rate'].astype('float64')
return df
2020-06-27 07:56:37 +00:00
def load_trades_from_db(db_url: str, strategy: Optional[str] = None) -> pd.DataFrame:
2019-06-16 08:41:05 +00:00
"""
2019-06-22 13:45:20 +00:00
Load trades from a DB (using dburl)
2019-06-16 08:41:05 +00:00
:param db_url: Sqlite url (default format sqlite:///tradesv3.dry-run.sqlite)
2020-06-27 07:56:37 +00:00
:param strategy: Strategy to load - mainly relevant for multi-strategy backtests
Can also serve as protection to load the correct result.
2019-06-23 20:10:37 +00:00
:return: Dataframe containing Trades
2019-06-16 08:41:05 +00:00
"""
2020-10-16 05:39:12 +00:00
init_db(db_url, clean_open_orders=False)
2020-06-27 07:56:37 +00:00
filters = []
if strategy:
2020-08-18 13:20:37 +00:00
filters.append(Trade.strategy == strategy)
trades = trade_list_to_dataframe(Trade.get_trades(filters).all())
2019-06-16 08:41:05 +00:00
return trades
2020-03-18 10:42:42 +00:00
def load_trades(source: str, db_url: str, exportfilename: Path,
2020-06-27 07:56:37 +00:00
no_trades: bool = False, strategy: Optional[str] = None) -> pd.DataFrame:
"""
2020-08-26 18:52:09 +00:00
Based on configuration option 'trade_source':
* loads data from DB (using `db_url`)
2019-07-03 04:26:39 +00:00
* loads data from backtestfile (using `exportfilename`)
:param source: "DB" or "file" - specify source to load from
:param db_url: sqlalchemy formatted url to a database
:param exportfilename: Json file generated by backtesting
2020-03-15 20:20:32 +00:00
:param no_trades: Skip using trades, only return backtesting data columns
:return: DataFrame containing trades
"""
2020-03-15 20:20:32 +00:00
if no_trades:
2020-03-14 21:15:03 +00:00
df = pd.DataFrame(columns=BT_DATA_COLUMNS)
return df
2019-08-22 18:17:36 +00:00
if source == "DB":
return load_trades_from_db(db_url)
elif source == "file":
2020-06-27 07:56:37 +00:00
return load_backtest_data(exportfilename, strategy)
def extract_trades_of_period(dataframe: pd.DataFrame, trades: pd.DataFrame,
date_index=False) -> pd.DataFrame:
"""
Compare trades and backtested pair DataFrames to get trades performed on backtested period
:return: the DataFrame of a trades of period
"""
if date_index:
trades_start = dataframe.index[0]
trades_stop = dataframe.index[-1]
else:
trades_start = dataframe.iloc[0]['date']
trades_stop = dataframe.iloc[-1]['date']
trades = trades.loc[(trades['open_date'] >= trades_start) &
(trades['close_date'] <= trades_stop)]
return trades
2019-06-29 14:57:04 +00:00
def calculate_market_change(data: Dict[str, pd.DataFrame], column: str = "close") -> float:
"""
Calculate market change based on "column".
Calculation is done by taking the first non-null and the last non-null element of each column
and calculating the pctchange as "(last - first) / first".
Then the results per pair are combined as mean.
:param data: Dict of Dataframes, dict key should be pair.
:param column: Column in the original dataframes to use
:return:
"""
tmp_means = []
for pair, df in data.items():
start = df[column].dropna().iloc[0]
end = df[column].dropna().iloc[-1]
tmp_means.append((end - start) / start)
2021-01-31 10:21:23 +00:00
return float(np.mean(tmp_means))
def combine_dataframes_with_mean(data: Dict[str, pd.DataFrame],
column: str = "close") -> pd.DataFrame:
2019-06-30 08:04:43 +00:00
"""
Combine multiple dataframes "column"
:param data: Dict of Dataframes, dict key should be pair.
2019-06-30 08:04:43 +00:00
:param column: Column in the original dataframes to use
:return: DataFrame with the column renamed to the dict key, and a column
named mean, containing the mean of all pairs.
:raise: ValueError if no data is provided.
2019-06-30 08:04:43 +00:00
"""
df_comb = pd.concat([data[pair].set_index('date').rename(
{column: pair}, axis=1)[pair] for pair in data], axis=1)
2019-06-30 08:04:43 +00:00
df_comb['mean'] = df_comb.mean(axis=1)
return df_comb
def create_cum_profit(df: pd.DataFrame, trades: pd.DataFrame, col_name: str,
timeframe: str) -> pd.DataFrame:
2019-06-29 14:57:04 +00:00
"""
Adds a column `col_name` with the cumulative profit for the given trades array.
2019-06-29 15:19:42 +00:00
:param df: DataFrame with date index
2021-04-25 08:10:09 +00:00
:param trades: DataFrame containing trades (requires columns close_date and profit_abs)
:param col_name: Column name that will be assigned the results
:param timeframe: Timeframe used during the operations
2019-06-29 15:19:42 +00:00
:return: Returns df with one additional column, col_name, containing the cumulative profit.
:raise: ValueError if trade-dataframe was found empty.
2019-06-29 14:57:04 +00:00
"""
if len(trades) == 0:
raise ValueError("Trade dataframe empty.")
from freqtrade.exchange import timeframe_to_minutes
2019-11-02 19:34:39 +00:00
timeframe_minutes = timeframe_to_minutes(timeframe)
# Resample to timeframe to make sure trades match candles
_trades_sum = trades.resample(f'{timeframe_minutes}min', on='close_date'
2021-04-25 08:10:09 +00:00
)[['profit_abs']].sum()
df.loc[:, col_name] = _trades_sum['profit_abs'].cumsum()
2019-06-29 14:57:04 +00:00
# Set first value to 0
df.loc[df.iloc[0].name, col_name] = 0
# FFill to get continuous
df[col_name] = df[col_name].ffill()
return df
2020-03-03 06:13:11 +00:00
def _calc_drawdown_series(profit_results: pd.DataFrame, *, date_col: str, value_col: str
) -> pd.DataFrame:
max_drawdown_df = pd.DataFrame()
max_drawdown_df['cumulative'] = profit_results[value_col].cumsum()
max_drawdown_df['high_value'] = max_drawdown_df['cumulative'].cummax()
max_drawdown_df['drawdown'] = max_drawdown_df['cumulative'] - max_drawdown_df['high_value']
max_drawdown_df['date'] = profit_results.loc[:, date_col]
return max_drawdown_df
def calculate_underwater(trades: pd.DataFrame, *, date_col: str = 'close_date',
value_col: str = 'profit_ratio'
):
"""
Calculate max drawdown and the corresponding close dates
:param trades: DataFrame containing trades (requires columns close_date and profit_ratio)
:param date_col: Column in DataFrame to use for dates (defaults to 'close_date')
:param value_col: Column in DataFrame to use for values (defaults to 'profit_ratio')
:return: Tuple (float, highdate, lowdate, highvalue, lowvalue) with absolute max drawdown,
high and low time and high and low value.
:raise: ValueError if trade-dataframe was found empty.
"""
if len(trades) == 0:
raise ValueError("Trade dataframe empty.")
profit_results = trades.sort_values(date_col).reset_index(drop=True)
max_drawdown_df = _calc_drawdown_series(profit_results, date_col=date_col, value_col=value_col)
return max_drawdown_df
def calculate_max_drawdown(trades: pd.DataFrame, *, date_col: str = 'close_date',
value_col: str = 'profit_ratio'
2021-02-14 18:30:17 +00:00
) -> Tuple[float, pd.Timestamp, pd.Timestamp, float, float]:
2020-03-03 06:13:11 +00:00
"""
Calculate max drawdown and the corresponding close dates
:param trades: DataFrame containing trades (requires columns close_date and profit_ratio)
:param date_col: Column in DataFrame to use for dates (defaults to 'close_date')
:param value_col: Column in DataFrame to use for values (defaults to 'profit_ratio')
2021-02-14 18:30:17 +00:00
:return: Tuple (float, highdate, lowdate, highvalue, lowvalue) with absolute max drawdown,
high and low time and high and low value.
2020-03-03 06:13:11 +00:00
:raise: ValueError if trade-dataframe was found empty.
"""
if len(trades) == 0:
2020-03-03 06:20:41 +00:00
raise ValueError("Trade dataframe empty.")
profit_results = trades.sort_values(date_col).reset_index(drop=True)
max_drawdown_df = _calc_drawdown_series(profit_results, date_col=date_col, value_col=value_col)
idxmin = max_drawdown_df['drawdown'].idxmin()
2020-04-05 12:43:01 +00:00
if idxmin == 0:
raise ValueError("No losing trade, therefore no drawdown.")
high_date = profit_results.loc[max_drawdown_df.iloc[:idxmin]['high_value'].idxmax(), date_col]
low_date = profit_results.loc[idxmin, date_col]
2021-02-14 18:30:17 +00:00
high_val = max_drawdown_df.loc[max_drawdown_df.iloc[:idxmin]
['high_value'].idxmax(), 'cumulative']
low_val = max_drawdown_df.loc[idxmin, 'cumulative']
return abs(min(max_drawdown_df['drawdown'])), high_date, low_date, high_val, low_val
def calculate_csum(trades: pd.DataFrame, starting_balance: float = 0) -> Tuple[float, float]:
"""
Calculate min/max cumsum of trades, to show if the wallet/stake amount ratio is sane
:param trades: DataFrame containing trades (requires columns close_date and profit_percent)
:param starting_balance: Add starting balance to results, to show the wallets high / low points
:return: Tuple (float, float) with cumsum of profit_abs
:raise: ValueError if trade-dataframe was found empty.
"""
if len(trades) == 0:
raise ValueError("Trade dataframe empty.")
csum_df = pd.DataFrame()
csum_df['sum'] = trades['profit_abs'].cumsum()
csum_min = csum_df['sum'].min() + starting_balance
csum_max = csum_df['sum'].max() + starting_balance
return csum_min, csum_max