stable/freqtrade/optimize/backtesting.py

340 lines
13 KiB
Python
Raw Normal View History

# pragma pylint: disable=missing-docstring, W0212, too-many-arguments
2017-11-14 21:15:24 +00:00
"""
This module contains the backtesting logic
"""
2018-03-25 19:37:14 +00:00
import logging
2018-03-29 18:16:25 +00:00
import operator
2018-03-17 21:43:36 +00:00
from argparse import Namespace
2018-06-10 11:15:25 +00:00
from typing import Dict, Tuple, Any, List, Optional, NamedTuple
2018-03-17 21:44:47 +00:00
import arrow
2018-03-29 18:16:25 +00:00
from pandas import DataFrame
from tabulate import tabulate
2017-09-28 21:26:28 +00:00
2018-01-10 07:51:36 +00:00
import freqtrade.optimize as optimize
from freqtrade import exchange
from freqtrade.analyze import Analyze
2018-03-17 21:44:47 +00:00
from freqtrade.arguments import Arguments
from freqtrade.configuration import Configuration
from freqtrade.misc import file_dump_json
2017-09-28 21:26:28 +00:00
from freqtrade.persistence import Trade
2018-03-25 19:37:14 +00:00
logger = logging.getLogger(__name__)
2018-06-10 11:15:25 +00:00
class BacktestResult(NamedTuple):
"""
NamedTuple Defining BacktestResults inputs.
"""
pair: str
profit_percent: float
profit_abs: float
open_time: float
close_time: float
trade_duration: float
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.analyze = Analyze(self.config)
self.ticker_interval = self.analyze.strategy.ticker_interval
self.tickerdata_to_dataframe = self.analyze.tickerdata_to_dataframe
self.populate_buy_trend = self.analyze.populate_buy_trend
self.populate_sell_trend = self.analyze.populate_sell_trend
# Reset keys for backtesting
2018-03-24 18:45:23 +00:00
self.config['exchange']['key'] = ''
self.config['exchange']['secret'] = ''
self.config['exchange']['password'] = ''
self.config['exchange']['uid'] = ''
self.config['dry_run'] = True
2018-03-24 18:45:23 +00:00
exchange.init(self.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
"""
2018-03-29 18:16:25 +00:00
timeframe = [
(arrow.get(min(frame.date)), arrow.get(max(frame.date)))
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
"""
2018-06-02 11:43:51 +00:00
stake_currency = str(self.config.get('stake_currency'))
floatfmt = ('s', 'd', '.2f', '.8f', '.1f')
tabular_data = []
headers = ['pair', 'buy count', 'avg profit %',
'total profit ' + stake_currency, 'avg duration', 'profit', 'loss']
for pair in data:
2018-06-10 11:15:25 +00:00
result = results[results.pair == pair]
tabular_data.append([
pair,
len(result.index),
result.profit_percent.mean() * 100.0,
2018-06-10 11:15:25 +00:00
result.profit_abs.sum(),
result.trade_duration.mean(),
len(result[result.profit_abs > 0]),
len(result[result.profit_abs < 0])
])
# Append Total
tabular_data.append([
'TOTAL',
len(results.index),
results.profit_percent.mean() * 100.0,
2018-06-10 11:15:25 +00:00
results.profit_abs.sum(),
results.trade_duration.mean(),
len(results[results.profit_abs > 0]),
len(results[results.profit_abs < 0])
])
return tabulate(tabular_data, headers=headers, floatfmt=floatfmt, tablefmt="pipe")
2018-03-17 21:43:36 +00:00
def _get_sell_trade_entry(
self, pair: str, buy_row: DataFrame,
2018-06-10 11:15:25 +00:00
partial_ticker: List, trade_count_lock: Dict, args: Dict) -> Optional[BacktestResult]:
2018-03-17 21:43:36 +00:00
stake_amount = args['stake_amount']
max_open_trades = args.get('max_open_trades', 0)
fee = exchange.get_fee()
trade = Trade(
open_rate=buy_row.close,
open_date=buy_row.date,
stake_amount=stake_amount,
amount=stake_amount / buy_row.open,
fee_open=fee,
fee_close=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
if self.analyze.should_sell(trade, sell_row.close, sell_row.date, buy_signal,
sell_row.sell):
2018-06-10 11:15:25 +00:00
return BacktestResult(pair=pair,
profit_percent=trade.calc_profit_percent(rate=sell_row.close),
profit_abs=trade.calc_profit(rate=sell_row.close),
open_time=buy_row.date,
close_time=sell_row.date,
trade_duration=(sell_row.date - buy_row.date).seconds // 60
)
if partial_ticker:
# no sell condition found - trade stil open at end of backtest period
sell_row = partial_ticker[-1]
2018-06-10 11:15:25 +00:00
btr = BacktestResult(pair=pair,
profit_percent=trade.calc_profit_percent(rate=sell_row.close),
profit_abs=trade.calc_profit(rate=sell_row.close),
open_time=buy_row.date,
close_time=sell_row.date,
trade_duration=(sell_row.date - buy_row.date).seconds // 60
)
logger.info('Force_selling still open trade %s with %s perc - %s', btr.pair,
btr.profit_percent, btr.profit_abs)
return btr
return None
2018-03-17 21:43:36 +00:00
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)
realistic: do we try to simulate realistic trades? (default: True)
sell_profit_only: sell if profit only
use_sell_signal: act on sell-signal
:return: DataFrame
"""
headers = ['date', 'buy', 'open', 'close', 'sell']
processed = args['processed']
max_open_trades = args.get('max_open_trades', 0)
realistic = args.get('realistic', False)
record = args.get('record', None)
2018-06-03 12:52:03 +00:00
recordfilename = args.get('recordfn', 'backtest-result.json')
records = []
trades = []
2018-06-02 11:43:51 +00:00
trade_count_lock: Dict = {}
for pair, pair_data in processed.items():
pair_data['buy'], pair_data['sell'] = 0, 0 # cleanup from previous run
ticker_data = self.populate_sell_trend(
self.populate_buy_trend(pair_data))[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)
2018-06-10 06:58:28 +00:00
# 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 realistic:
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
2018-06-10 11:15:25 +00:00
trade_entry = self._get_sell_trade_entry(pair, row, ticker[index + 1:],
trade_count_lock, args)
2018-06-10 11:15:25 +00:00
if trade_entry:
lock_pair_until = trade_entry.close_time
trades.append(trade_entry)
if record:
# Note, need to be json.dump friendly
# record a tuple of pair, current_profit_percent,
# entry-date, duration
2018-06-10 11:15:25 +00:00
records.append((pair, trade_entry.profit_percent,
trade_entry.open_time.strftime('%s'),
trade_entry.close_time.strftime('%s'),
index, trade_entry[3]))
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
# For now export inside backtest(), maybe change so that backtest()
# returns a tuple like: (dataframe, records, logs, etc)
if record and record.find('trades') >= 0:
2018-06-03 12:52:03 +00:00
logger.info('Dumping backtest results to %s', recordfilename)
file_dump_json(recordfilename, records)
labels = ['currency', 'profit_percent', 'profit_BTC', 'duration']
2018-06-10 11:15:25 +00:00
return DataFrame.from_records(trades, columns=BacktestResult._fields)
def start(self) -> None:
"""
Run a backtesting end-to-end
:return: None
"""
data = {}
pairs = self.config['exchange']['pair_whitelist']
2018-03-25 19:37:14 +00:00
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'):
2018-03-25 19:37:14 +00:00
logger.info('Downloading data for all pairs in whitelist ...')
for pair in pairs:
data[pair] = exchange.get_ticker_history(pair, self.ticker_interval)
else:
2018-03-25 19:37:14 +00:00
logger.info('Using local backtesting data (using whitelist in given config) ...')
2018-06-02 12:07:54 +00:00
timerange = Arguments.parse_timerange(None if self.config.get(
'timerange') is None else str(self.config.get('timerange')))
2018-06-05 21:34:26 +00:00
data = optimize.load_data(
self.config['datadir'],
pairs=pairs,
ticker_interval=self.ticker_interval,
refresh_pairs=self.config.get('refresh_pairs', False),
timerange=timerange
)
# Ignore max_open_trades in backtesting, except realistic flag was passed
if self.config.get('realistic_simulation', False):
max_open_trades = self.config['max_open_trades']
else:
2018-03-25 19:37:14 +00:00
logger.info('Ignoring max_open_trades (realistic_simulation not set) ...')
max_open_trades = 0
preprocessed = self.tickerdata_to_dataframe(data)
# Print timeframe
min_date, max_date = self.get_timeframe(preprocessed)
2018-03-25 19:37:14 +00:00
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,
'realistic': self.config.get('realistic_simulation', False),
2018-06-03 12:52:03 +00:00
'record': self.config.get('export'),
'recordfn': self.config.get('exportfilename'),
}
)
2018-03-25 19:37:14 +00:00
logger.info(
'\n==================================== '
'BACKTESTING REPORT'
' ====================================\n'
'%s',
self._generate_text_table(
data,
results
)
)
2018-03-17 21:43:36 +00:00
def setup_configuration(args: Namespace) -> Dict[str, Any]:
2017-11-22 23:25:06 +00:00
"""
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'] = ''
return config
2018-03-17 21:43:36 +00:00
def start(args: Namespace) -> None:
"""
Start Backtesting script
:param args: Cli args from Arguments()
:return: None
2017-11-22 23:25:06 +00:00
"""
# Initialize configuration
config = setup_configuration(args)
2018-03-25 19:41:25 +00:00
logger.info('Starting freqtrade in Backtesting mode')
# Initialize backtesting object
backtesting = Backtesting(config)
backtesting.start()