stable/freqtrade/optimize/backtesting.py

608 lines
26 KiB
Python

# pragma pylint: disable=missing-docstring, W0212, too-many-arguments
"""
This module contains the backtesting logic
"""
import logging
import operator
from argparse import Namespace
from datetime import datetime, timedelta
from typing import Any, Dict, List, NamedTuple, Optional, Tuple
import arrow
from pandas import DataFrame, to_datetime
from tabulate import tabulate
import freqtrade.optimize as optimize
from freqtrade import DependencyException, constants
from freqtrade.arguments import Arguments
from freqtrade.configuration import Configuration
from freqtrade.exchange import Exchange
from freqtrade.misc import file_dump_json
from freqtrade.optimize.backslapping import Backslapping
from freqtrade.persistence import Trade
from freqtrade.strategy.interface import SellType
from freqtrade.strategy.resolver import IStrategy, StrategyResolver
from collections import OrderedDict
import timeit
from time import sleep
import pdb
logger = logging.getLogger(__name__)
class BacktestResult(NamedTuple):
"""
NamedTuple Defining BacktestResults inputs.
"""
pair: str
profit_percent: float
profit_abs: float
open_time: datetime
close_time: datetime
open_index: int
close_index: int
trade_duration: float
open_at_end: bool
open_rate: float
close_rate: float
sell_reason: SellType
class Backtesting(object):
"""
Backtesting class, this class contains all the logic to run a backtest
To run a backtest:
backtesting = Backtesting(config)
backtesting.start()
"""
def __init__(self, config: Dict[str, Any]) -> None:
self.config = config
self.strategy: IStrategy = StrategyResolver(self.config).strategy
self.ticker_interval = self.strategy.ticker_interval
self.tickerdata_to_dataframe = self.strategy.tickerdata_to_dataframe
self.advise_buy = self.strategy.advise_buy
self.advise_sell = self.strategy.advise_sell
# Reset keys for backtesting
self.config['exchange']['key'] = ''
self.config['exchange']['secret'] = ''
self.config['exchange']['password'] = ''
self.config['exchange']['uid'] = ''
self.config['dry_run'] = True
self.exchange = Exchange(self.config)
self.fee = self.exchange.get_fee()
self.stop_loss_value = self.strategy.stoploss
#### backslap config
'''
Numpy arrays are used for 100x speed up
We requires setting Int values for
buy stop triggers and stop calculated on
# buy 0 - open 1 - close 2 - sell 3 - high 4 - low 5 - stop 6
'''
self.np_buy: int = 0
self.np_open: int = 1
self.np_close: int = 2
self.np_sell: int = 3
self.np_high: int = 4
self.np_low: int = 5
self.np_stop: int = 6
self.np_bto: int = self.np_close # buys_triggered_on - should be close
self.np_bco: int = self.np_open # buys calculated on - open of the next candle.
self.np_sto: int = self.np_low # stops_triggered_on - Should be low, FT uses close
self.np_sco: int = self.np_stop # stops_calculated_on - Should be stop, FT uses close
# self.np_sto: int = self.np_close # stops_triggered_on - Should be low, FT uses close
# self.np_sco: int = self.np_close # stops_calculated_on - Should be stop, FT uses close
if 'backslap' in config:
self.use_backslap = config['backslap'] # Enable backslap - if false Orginal code is executed.
else:
self.use_backslap = False
logger.info("using backslap: {}".format(self.use_backslap))
self.debug = False # Main debug enable, very print heavy, enable 2 loops recommended
self.debug_timing = False # Stages within Backslap
self.debug_2loops = False # Limit each pair to two loops, useful when debugging
self.debug_vector = False # Debug vector calcs
self.debug_timing_main_loop = False # print overall timing per pair - works in Backtest and Backslap
self.backslap_show_trades = False # prints trades in addition to summary report
self.backslap_save_trades = True # saves trades as a pretty table to backslap.txt
self.stop_stops: int = 9999 # stop back testing any pair with this many stops, set to 999999 to not hit
self.backslap = Backslapping(config)
@staticmethod
def get_timeframe(data: Dict[str, DataFrame]) -> Tuple[arrow.Arrow, arrow.Arrow]:
"""
Get the maximum timeframe for the given backtest data
:param data: dictionary with preprocessed backtesting data
:return: tuple containing min_date, max_date
"""
timeframe = [
(arrow.get(frame['date'].min()), arrow.get(frame['date'].max()))
for frame in data.values()
]
return min(timeframe, key=operator.itemgetter(0))[0], \
max(timeframe, key=operator.itemgetter(1))[1]
def _generate_text_table(self, data: Dict[str, Dict], results: DataFrame) -> str:
"""
Generates and returns a text table for the given backtest data and the results dataframe
:return: pretty printed table with tabulate as str
"""
stake_currency = str(self.config.get('stake_currency'))
floatfmt = ('s', 'd', '.2f', '.2f', '.8f', 'd', '.1f', '.1f')
tabular_data = []
# headers = ['pair', 'buy count', 'avg profit %', 'cum profit %',
# 'total profit ' + stake_currency, 'avg duration', 'profit', 'loss', 'total loss ab', 'total profit ab', 'Risk Reward Ratio', 'Win Rate']
headers = ['pair', 'buy count', 'avg profit %', 'cum profit %',
'total profit ' + stake_currency, 'avg duration', 'profit', 'loss', 'RRR', 'Win Rate %', 'Required RR']
for pair in data:
result = results[results.pair == pair]
win_rate = (len(result[result.profit_abs > 0]) / len(result.index)) if (len(result.index) > 0) else None
tabular_data.append([
pair,
len(result.index),
result.profit_percent.mean() * 100.0,
result.profit_percent.sum() * 100.0,
result.profit_abs.sum(),
str(timedelta(
minutes=round(result.trade_duration.mean()))) if not result.empty else '0:00',
len(result[result.profit_abs > 0]),
len(result[result.profit_abs < 0]),
# result[result.profit_abs < 0]['profit_abs'].sum(),
# result[result.profit_abs > 0]['profit_abs'].sum(),
abs(1 / ((result[result.profit_abs < 0]['profit_abs'].sum() / len(result[result.profit_abs < 0])) / (result[result.profit_abs > 0]['profit_abs'].sum() / len(result[result.profit_abs > 0])))),
win_rate * 100 if win_rate else "nan",
((1 / win_rate) - 1) if win_rate else "nan"
])
# Append Total
tabular_data.append([
'TOTAL',
len(results.index),
results.profit_percent.mean() * 100.0,
results.profit_percent.sum() * 100.0,
results.profit_abs.sum(),
str(timedelta(
minutes=round(results.trade_duration.mean()))) if not results.empty else '0:00',
len(results[results.profit_abs > 0]),
len(results[results.profit_abs < 0])
])
return tabulate(tabular_data, headers=headers, floatfmt=floatfmt, tablefmt="pipe")
def _generate_text_table_edge_positioning(self, data: Dict[str, Dict], results: DataFrame) -> str:
"""
This is a temporary version of edge positioning calculation.
The function will be eventually moved to a plugin called Edge in order to calculate necessary WR, RRR and
other indictaors related to money management periodically (each X minutes) and keep it in a storage.
The calulation will be done per pair and per strategy.
"""
tabular_data = []
headers = ['Number of trades', 'RRR', 'Win Rate %', 'Required RR']
###
# The algorithm should be:
# 1) Removing outliers from dataframe. i.e. all profit_percent which are outside (mean -+ (2 * (standard deviation))).
# 2) Removing pairs with less than X trades (X defined in config).
# 3) Calculating RRR and WR.
# 4) Removing pairs for which WR and RRR are not in an acceptable range (e.x. WR > 95%).
# 5) Sorting the result based on the delta between required RR and RRR.
# Here we assume initial data in order to calculate position size.
# these values will be replaced by exchange info or config
for pair in data:
result = results[results.pair == pair]
# WinRate is calculated as follows: (Number of profitable trades) / (Total Trades)
win_rate = (len(result[result.profit_abs > 0]) / len(result.index)) if (len(result.index) > 0) else None
# Risk Reward Ratio is calculated as follows: 1 / ((total loss on losing trades / number of losing trades) / (total gain on profitable trades / number of winning trades))
risk_reward_ratio = abs(1 / ((result[result.profit_abs < 0]['profit_abs'].sum() / len(result[result.profit_abs < 0])) / (result[result.profit_abs > 0]['profit_abs'].sum() / len(result[result.profit_abs > 0]))))
# Required Reward Ratio is (1 / WinRate) - 1
required_risk_reward = ((1 / win_rate) - 1) if win_rate else None
#pdb.set_trace()
tabular_data.append([
pair,
len(result.index),
risk_reward_ratio,
win_rate * 100 if win_rate else "nan",
required_risk_reward
])
# for pair in data:
# result = results[results.pair == pair]
# win_rate = (len(result[result.profit_abs > 0]) / len(result.index)) if (len(result.index) > 0) else None
# tabular_data.append([
# pair,
# #len(result.index),
# #result.profit_percent.mean() * 100.0,
# #result.profit_percent.sum() * 100.0,
# #result.profit_abs.sum(),
# str(timedelta(
# minutes=round(result.trade_duration.mean()))) if not result.empty else '0:00',
# len(result[result.profit_abs > 0]),
# len(result[result.profit_abs < 0]),
# # result[result.profit_abs < 0]['profit_abs'].sum(),
# # result[result.profit_abs > 0]['profit_abs'].sum(),
# abs(1 / ((result[result.profit_abs < 0]['profit_abs'].sum() / len(result[result.profit_abs < 0])) / (result[result.profit_abs > 0]['profit_abs'].sum() / len(result[result.profit_abs > 0])))),
# win_rate * 100 if win_rate else "nan",
# ((1 / win_rate) - 1) if win_rate else "nan"
# ])
#return tabulate(tabular_data, headers=headers, floatfmt=floatfmt, tablefmt="pipe")
return tabulate(tabular_data, headers=headers, tablefmt="pipe")
def _generate_text_table_sell_reason(self, data: Dict[str, Dict], results: DataFrame) -> str:
"""
Generate small table outlining Backtest results
"""
tabular_data = []
headers = ['Sell Reason', 'Count']
for reason, count in results['sell_reason'].value_counts().iteritems():
tabular_data.append([reason.value, count])
return tabulate(tabular_data, headers=headers, tablefmt="pipe")
def _store_backtest_result(self, recordfilename: Optional[str], results: DataFrame) -> None:
records = [(t.pair, t.profit_percent, t.open_time.timestamp(),
t.close_time.timestamp(), t.open_index - 1, t.trade_duration,
t.open_rate, t.close_rate, t.open_at_end, t.sell_reason.value)
for index, t in results.iterrows()]
if records:
logger.info('Dumping backtest results to %s', recordfilename)
file_dump_json(recordfilename, records)
def _get_sell_trade_entry(
self, pair: str, buy_row: DataFrame,
partial_ticker: List, trade_count_lock: Dict, args: Dict) -> Optional[BacktestResult]:
stake_amount = args['stake_amount']
max_open_trades = args.get('max_open_trades', 0)
trade = Trade(
open_rate=buy_row.open,
open_date=buy_row.date,
stake_amount=stake_amount,
amount=stake_amount / buy_row.open,
fee_open=self.fee,
fee_close=self.fee
)
# calculate win/lose forwards from buy point
for sell_row in partial_ticker:
if max_open_trades > 0:
# Increase trade_count_lock for every iteration
trade_count_lock[sell_row.date] = trade_count_lock.get(sell_row.date, 0) + 1
buy_signal = sell_row.buy
sell = self.strategy.should_sell(trade, sell_row.open, sell_row.date, buy_signal,
sell_row.sell)
if sell.sell_flag:
return BacktestResult(pair=pair,
profit_percent=trade.calc_profit_percent(rate=sell_row.open),
profit_abs=trade.calc_profit(rate=sell_row.open),
open_time=buy_row.date,
close_time=sell_row.date,
trade_duration=int((
sell_row.date - buy_row.date).total_seconds() // 60),
open_index=buy_row.Index,
close_index=sell_row.Index,
open_at_end=False,
open_rate=buy_row.open,
close_rate=sell_row.open,
sell_reason=sell.sell_type
)
if partial_ticker:
# no sell condition found - trade stil open at end of backtest period
sell_row = partial_ticker[-1]
btr = BacktestResult(pair=pair,
profit_percent=trade.calc_profit_percent(rate=sell_row.open),
profit_abs=trade.calc_profit(rate=sell_row.open),
open_time=buy_row.date,
close_time=sell_row.date,
trade_duration=int((
sell_row.date - buy_row.date).total_seconds() // 60),
open_index=buy_row.Index,
close_index=sell_row.Index,
open_at_end=True,
open_rate=buy_row.open,
close_rate=sell_row.open,
sell_reason=SellType.FORCE_SELL
)
logger.debug('Force_selling still open trade %s with %s perc - %s', btr.pair,
btr.profit_percent, btr.profit_abs)
return btr
return None
def s(self):
st = timeit.default_timer()
return st
def f(self, st):
return (timeit.default_timer() - st)
def backtest(self, args: Dict) -> DataFrame:
"""
Implements backtesting functionality
NOTE: This method is used by Hyperopt at each iteration. Please keep it optimized.
Of course try to not have ugly code. By some accessor are sometime slower than functions.
Avoid, logging on this method
:param args: a dict containing:
stake_amount: btc amount to use for each trade
processed: a processed dictionary with format {pair, data}
max_open_trades: maximum number of concurrent trades (default: 0, disabled)
position_stacking: do we allow position stacking? (default: False)
:return: DataFrame
"""
use_backslap = self.use_backslap
debug_timing = self.debug_timing_main_loop
if use_backslap: # Use Back Slap code
return self.backslap.run(args)
else: # use Original Back test code
########################## Original BT loop
headers = ['date', 'buy', 'open', 'close', 'sell']
processed = args['processed']
max_open_trades = args.get('max_open_trades', 0)
position_stacking = args.get('position_stacking', False)
trades = []
trade_count_lock: Dict = {}
for pair, pair_data in processed.items():
if debug_timing: # Start timer
fl = self.s()
pair_data['buy'], pair_data['sell'] = 0, 0 # cleanup from previous run
ticker_data = self.advise_sell(
self.advise_buy(pair_data, {'pair': pair}), {'pair': pair})[headers].copy()
# to avoid using data from future, we buy/sell with signal from previous candle
ticker_data.loc[:, 'buy'] = ticker_data['buy'].shift(1)
ticker_data.loc[:, 'sell'] = ticker_data['sell'].shift(1)
ticker_data.drop(ticker_data.head(1).index, inplace=True)
if debug_timing: # print time taken
flt = self.f(fl)
# print("populate_buy_trend:", pair, round(flt, 10))
st = self.s()
# Convert from Pandas to list for performance reasons
# (Looping Pandas is slow.)
ticker = [x for x in ticker_data.itertuples()]
lock_pair_until = None
for index, row in enumerate(ticker):
if row.buy == 0 or row.sell == 1:
continue # skip rows where no buy signal or that would immediately sell off
if not position_stacking:
if lock_pair_until is not None and row.date <= lock_pair_until:
continue
if max_open_trades > 0:
# Check if max_open_trades has already been reached for the given date
if not trade_count_lock.get(row.date, 0) < max_open_trades:
continue
trade_count_lock[row.date] = trade_count_lock.get(row.date, 0) + 1
trade_entry = self._get_sell_trade_entry(pair, row, ticker[index + 1:],
trade_count_lock, args)
if trade_entry:
lock_pair_until = trade_entry.close_time
trades.append(trade_entry)
else:
# Set lock_pair_until to end of testing period if trade could not be closed
# This happens only if the buy-signal was with the last candle
lock_pair_until = ticker_data.iloc[-1].date
if debug_timing: # print time taken
tt = self.f(st)
print("Time to BackTest :", pair, round(tt, 10))
print("-----------------------")
return DataFrame.from_records(trades, columns=BacktestResult._fields)
####################### Original BT loop end
def start(self) -> None:
"""
Run a backtesting end-to-end
:return: None
"""
data: Dict[str, Any] = {}
pairs = self.config['exchange']['pair_whitelist']
logger.info('Using stake_currency: %s ...', self.config['stake_currency'])
logger.info('Using stake_amount: %s ...', self.config['stake_amount'])
if self.config.get('live'):
logger.info('Downloading data for all pairs in whitelist ...')
self.exchange.refresh_tickers(pairs, self.ticker_interval)
data = self.exchange.klines
else:
logger.info('Using local backtesting data (using whitelist in given config) ...')
timerange = Arguments.parse_timerange(None if self.config.get(
'timerange') is None else str(self.config.get('timerange')))
data = optimize.load_data(
self.config['datadir'],
pairs=pairs,
ticker_interval=self.ticker_interval,
refresh_pairs=self.config.get('refresh_pairs', False),
exchange=self.exchange,
timerange=timerange
)
ld_files = self.s()
if not data:
logger.critical("No data found. Terminating.")
return
# Use max_open_trades in backtesting, except --disable-max-market-positions is set
if self.config.get('use_max_market_positions', True):
max_open_trades = self.config['max_open_trades']
else:
logger.info('Ignoring max_open_trades (--disable-max-market-positions was used) ...')
max_open_trades = 0
preprocessed = self.tickerdata_to_dataframe(data)
t_t = self.f(ld_files)
print("Load from json to file to df in mem took", t_t)
# Print timeframe
min_date, max_date = self.get_timeframe(preprocessed)
logger.info(
'Measuring data from %s up to %s (%s days)..',
min_date.isoformat(),
max_date.isoformat(),
(max_date - min_date).days
)
# Execute backtest and print results
results = self.backtest(
{
'stake_amount': self.config.get('stake_amount'),
'processed': preprocessed,
'max_open_trades': max_open_trades,
'position_stacking': self.config.get('position_stacking', False),
}
)
if self.config.get('export', False):
self._store_backtest_result(self.config.get('exportfilename'), results)
if self.use_backslap:
# logger.info(
# '\n====================================================== '
# 'BackSLAP REPORT'
# ' =======================================================\n'
# '%s',
# self._generate_text_table(
# data,
# results
# )
# )
logger.info(
'\n====================================================== '
'Edge positionning REPORT'
' =======================================================\n'
'%s',
self._generate_text_table_edge_positioning(
data,
results
)
)
# optional print trades
if self.backslap_show_trades:
TradesFrame = results.filter(['open_time', 'pair', 'exit_type', 'profit_percent', 'profit_abs',
'buy_spend', 'sell_take', 'trade_duration', 'close_time'], axis=1)
def to_fwf(df, fname):
content = tabulate(df.values.tolist(), list(df.columns), floatfmt=".8f", tablefmt='psql')
print(content)
DataFrame.to_fwf = to_fwf(TradesFrame, "backslap.txt")
# optional save trades
if self.backslap_save_trades:
TradesFrame = results.filter(['open_time', 'pair', 'exit_type', 'profit_percent', 'profit_abs',
'buy_spend', 'sell_take', 'trade_duration', 'close_time'], axis=1)
def to_fwf(df, fname):
content = tabulate(df.values.tolist(), list(df.columns), floatfmt=".8f", tablefmt='psql')
open(fname, "w").write(content)
DataFrame.to_fwf = to_fwf(TradesFrame, "backslap.txt")
else:
logger.info(
'\n================================================= '
'BACKTEST REPORT'
' ==================================================\n'
'%s',
self._generate_text_table(
data,
results
)
)
if 'sell_reason' in results.columns:
logger.info(
'\n' +
' SELL READON STATS '.center(119, '=') +
'\n%s \n',
self._generate_text_table_sell_reason(data, results)
)
else:
logger.info("no sell reasons available!")
logger.info(
'\n' +
' LEFT OPEN TRADES REPORT '.center(119, '=') +
'\n%s',
self._generate_text_table(
data,
results.loc[results.open_at_end]
)
)
def setup_configuration(args: Namespace) -> Dict[str, Any]:
"""
Prepare the configuration for the backtesting
:param args: Cli args from Arguments()
:return: Configuration
"""
configuration = Configuration(args)
config = configuration.get_config()
# Ensure we do not use Exchange credentials
config['exchange']['key'] = ''
config['exchange']['secret'] = ''
config['backslap'] = args.backslap
if config['stake_amount'] == constants.UNLIMITED_STAKE_AMOUNT:
raise DependencyException('stake amount could not be "%s" for backtesting' %
constants.UNLIMITED_STAKE_AMOUNT)
return config
def start(args: Namespace) -> None:
"""
Start Backtesting script
:param args: Cli args from Arguments()
:return: None
"""
# Initialize configuration
config = setup_configuration(args)
logger.info('Starting freqtrade in Backtesting mode')
# Initialize backtesting object
backtesting = Backtesting(config)
backtesting.start()