freqtrade_origin/freqtrade/optimize/hyperopt_tools.py

503 lines
22 KiB
Python
Raw Normal View History

import io
import logging
2021-05-29 14:49:28 +00:00
from copy import deepcopy
from datetime import datetime, timezone
from pathlib import Path
from typing import Any, Dict, Iterator, List, Optional, Tuple
import numpy as np
2021-09-19 23:44:12 +00:00
import pandas as pd
import rapidjson
import tabulate
from colorama import Fore, Style
from pandas import isna, json_normalize
from freqtrade.constants import FTHYPT_FILEVERSION, USERPATH_STRATEGIES
from freqtrade.exceptions import OperationalException
from freqtrade.misc import deep_merge_dicts, round_coin_value, round_dict, safe_value_fallback2
from freqtrade.optimize.hyperopt_epoch_filters import hyperopt_filter_epochs
logger = logging.getLogger(__name__)
NON_OPT_PARAM_APPENDIX = " # value loaded from strategy"
def hyperopt_serializer(x):
if isinstance(x, np.integer):
return int(x)
2021-07-02 18:52:25 +00:00
if isinstance(x, np.bool_):
return bool(x)
return str(x)
class HyperoptTools():
2021-05-29 14:49:28 +00:00
@staticmethod
2021-05-29 14:56:36 +00:00
def get_strategy_filename(config: Dict, strategy_name: str) -> Optional[Path]:
2021-05-29 14:49:28 +00:00
"""
Get Strategy-location (filename) from strategy_name
"""
from freqtrade.resolvers.strategy_resolver import StrategyResolver
directory = Path(config.get('strategy_path', config['user_data_dir'] / USERPATH_STRATEGIES))
strategy_objs = StrategyResolver.search_all_objects(directory, False)
2021-05-29 14:56:36 +00:00
strategies = [s for s in strategy_objs if s['name'] == strategy_name]
if strategies:
strategy = strategies[0]
2021-05-29 14:49:28 +00:00
return Path(strategy['location'])
2021-05-29 14:56:36 +00:00
return None
2021-05-29 14:49:28 +00:00
@staticmethod
def export_params(params, strategy_name: str, filename: Path):
"""
Generate files
"""
final_params = deepcopy(params['params_not_optimized'])
final_params = deep_merge_dicts(params['params_details'], final_params)
final_params = {
'strategy_name': strategy_name,
'params': final_params,
'ft_stratparam_v': 1,
'export_time': datetime.now(timezone.utc),
2021-05-29 14:49:28 +00:00
}
logger.info(f"Dumping parameters to {filename}")
rapidjson.dump(final_params, filename.open('w'), indent=2,
default=hyperopt_serializer,
number_mode=rapidjson.NM_NATIVE | rapidjson.NM_NAN
)
2021-05-29 14:49:28 +00:00
@staticmethod
2021-06-30 04:33:40 +00:00
def try_export_params(config: Dict[str, Any], strategy_name: str, params: Dict):
if params.get(FTHYPT_FILEVERSION, 1) >= 2 and not config.get('disableparamexport', False):
# Export parameters ...
fn = HyperoptTools.get_strategy_filename(config, strategy_name)
if fn:
2021-06-30 04:33:40 +00:00
HyperoptTools.export_params(params, strategy_name, fn.with_suffix('.json'))
else:
logger.warning("Strategy not found, not exporting parameter file.")
2021-05-01 14:36:35 +00:00
@staticmethod
def has_space(config: Dict[str, Any], space: str) -> bool:
"""
Tell if the space value is contained in the configuration
"""
2021-08-03 05:14:31 +00:00
# 'trailing' and 'protection spaces are not included in the 'default' set of spaces
if space in ('trailing', 'protection'):
2021-05-01 14:36:35 +00:00
return any(s in config['spaces'] for s in [space, 'all'])
else:
return any(s in config['spaces'] for s in [space, 'all', 'default'])
@staticmethod
def _read_results(results_file: Path, batch_size: int = 10) -> Iterator[List[Any]]:
"""
Stream hyperopt results from file
"""
import rapidjson
logger.info(f"Reading epochs from '{results_file}'")
with results_file.open('r') as f:
data = []
for line in f:
data += [rapidjson.loads(line)]
if len(data) >= batch_size:
yield data
data = []
yield data
@staticmethod
def _test_hyperopt_results_exist(results_file) -> bool:
if results_file.is_file() and results_file.stat().st_size > 0:
if results_file.suffix == '.pickle':
2021-08-09 05:03:13 +00:00
raise OperationalException(
"Legacy hyperopt results are no longer supported."
"Please rerun hyperopt or use an older version to load this file."
)
return True
else:
# No file found.
return False
@staticmethod
def load_filtered_results(results_file: Path, config: Dict[str, Any]) -> Tuple[List, int]:
filteroptions = {
'only_best': config.get('hyperopt_list_best', False),
'only_profitable': config.get('hyperopt_list_profitable', False),
'filter_min_trades': config.get('hyperopt_list_min_trades', 0),
'filter_max_trades': config.get('hyperopt_list_max_trades', 0),
'filter_min_avg_time': config.get('hyperopt_list_min_avg_time', None),
'filter_max_avg_time': config.get('hyperopt_list_max_avg_time', None),
'filter_min_avg_profit': config.get('hyperopt_list_min_avg_profit', None),
'filter_max_avg_profit': config.get('hyperopt_list_max_avg_profit', None),
'filter_min_total_profit': config.get('hyperopt_list_min_total_profit', None),
'filter_max_total_profit': config.get('hyperopt_list_max_total_profit', None),
'filter_min_objective': config.get('hyperopt_list_min_objective', None),
'filter_max_objective': config.get('hyperopt_list_max_objective', None),
}
if not HyperoptTools._test_hyperopt_results_exist(results_file):
# No file found.
return [], 0
epochs = []
total_epochs = 0
for epochs_tmp in HyperoptTools._read_results(results_file):
if total_epochs == 0 and epochs_tmp[0].get('is_best') is None:
raise OperationalException(
"The file with HyperoptTools results is incompatible with this version "
"of Freqtrade and cannot be loaded.")
total_epochs += len(epochs_tmp)
epochs += hyperopt_filter_epochs(epochs_tmp, filteroptions, log=False)
logger.info(f"Loaded {total_epochs} previous evaluations from disk.")
# Final filter run ...
epochs = hyperopt_filter_epochs(epochs, filteroptions, log=True)
return epochs, total_epochs
@staticmethod
2021-06-13 09:24:24 +00:00
def show_epoch_details(results, total_epochs: int, print_json: bool,
no_header: bool = False, header_str: str = None) -> None:
"""
Display details of the hyperopt result
"""
params = results.get('params_details', {})
non_optimized = results.get('params_not_optimized', {})
# Default header string
if header_str is None:
header_str = "Best result"
if not no_header:
explanation_str = HyperoptTools._format_explanation_string(results, total_epochs)
print(f"\n{header_str}:\n\n{explanation_str}\n")
if print_json:
result_dict: Dict = {}
2021-08-03 05:14:31 +00:00
for s in ['buy', 'sell', 'protection', 'roi', 'stoploss', 'trailing']:
HyperoptTools._params_update_for_json(result_dict, params, non_optimized, s)
print(rapidjson.dumps(result_dict, default=str, number_mode=rapidjson.NM_NATIVE))
else:
HyperoptTools._params_pretty_print(params, 'buy', "Buy hyperspace params:",
non_optimized)
HyperoptTools._params_pretty_print(params, 'sell', "Sell hyperspace params:",
non_optimized)
2021-08-03 05:14:31 +00:00
HyperoptTools._params_pretty_print(params, 'protection',
"Protection hyperspace params:", non_optimized)
HyperoptTools._params_pretty_print(params, 'roi', "ROI table:", non_optimized)
HyperoptTools._params_pretty_print(params, 'stoploss', "Stoploss:", non_optimized)
HyperoptTools._params_pretty_print(params, 'trailing', "Trailing stop:", non_optimized)
@staticmethod
def _params_update_for_json(result_dict, params, non_optimized, space: str) -> None:
if (space in params) or (space in non_optimized):
space_params = HyperoptTools._space_params(params, space)
space_non_optimized = HyperoptTools._space_params(non_optimized, space)
all_space_params = space_params
# Merge non optimized params if there are any
if len(space_non_optimized) > 0:
all_space_params = {**space_params, **space_non_optimized}
if space in ['buy', 'sell']:
result_dict.setdefault('params', {}).update(all_space_params)
elif space == 'roi':
# Convert keys in min_roi dict to strings because
# rapidjson cannot dump dicts with integer keys...
result_dict['minimal_roi'] = {str(k): v for k, v in all_space_params.items()}
else: # 'stoploss', 'trailing'
result_dict.update(all_space_params)
@staticmethod
def _params_pretty_print(params, space: str, header: str, non_optimized={}) -> None:
if space in params or space in non_optimized:
space_params = HyperoptTools._space_params(params, space, 5)
no_params = HyperoptTools._space_params(non_optimized, space, 5)
2021-06-15 18:33:35 +00:00
appendix = ''
if not space_params and not no_params:
# No parameters - don't print
return
if not space_params:
# Not optimized parameters - append string
2021-06-15 18:33:35 +00:00
appendix = NON_OPT_PARAM_APPENDIX
result = f"\n# {header}\n"
if space == "stoploss":
stoploss = safe_value_fallback2(space_params, no_params, space, space)
2021-06-15 18:33:35 +00:00
result += (f"stoploss = {stoploss}{appendix}")
elif space == "roi":
2021-06-15 18:33:35 +00:00
result = result[:-1] + f'{appendix}\n'
minimal_roi_result = rapidjson.dumps({
2021-08-06 22:19:36 +00:00
str(k): v for k, v in (space_params or no_params).items()
}, default=str, indent=4, number_mode=rapidjson.NM_NATIVE)
result += f"minimal_roi = {minimal_roi_result}"
elif space == "trailing":
for k, v in (space_params or no_params).items():
2021-06-15 18:33:35 +00:00
result += f"{k} = {v}{appendix}\n"
else:
# Buy / sell parameters
2021-06-15 18:33:35 +00:00
result += f"{space}_params = {HyperoptTools._pprint_dict(space_params, no_params)}"
result = result.replace("\n", "\n ")
print(result)
@staticmethod
def _space_params(params, space: str, r: int = None) -> Dict:
d = params.get(space)
if d:
# Round floats to `r` digits after the decimal point if requested
return round_dict(d, r) if r else d
return {}
@staticmethod
2021-06-15 18:33:35 +00:00
def _pprint_dict(params, non_optimized, indent: int = 4):
"""
Pretty-print hyperopt results (based on 2 dicts - with add. comment)
"""
p = params.copy()
p.update(non_optimized)
result = '{\n'
for k, param in p.items():
2021-05-07 18:23:11 +00:00
result += " " * indent + f'"{k}": '
result += f'"{param}",' if isinstance(param, str) else f'{param},'
if k in non_optimized:
result += NON_OPT_PARAM_APPENDIX
result += "\n"
result += '}'
return result
@staticmethod
def is_best_loss(results, current_best_loss: float) -> bool:
return bool(results['loss'] < current_best_loss)
@staticmethod
def format_results_explanation_string(results_metrics: Dict, stake_currency: str) -> str:
"""
Return the formatted results explanation in a string
"""
return (f"{results_metrics['total_trades']:6d} trades. "
f"{results_metrics['wins']}/{results_metrics['draws']}"
f"/{results_metrics['losses']} Wins/Draws/Losses. "
f"Avg profit {results_metrics['profit_mean'] * 100: 6.2f}%. "
f"Median profit {results_metrics['profit_median'] * 100: 6.2f}%. "
f"Total profit {results_metrics['profit_total_abs']: 11.8f} {stake_currency} "
f"({results_metrics['profit_total'] * 100: 7.2f}%). "
f"Avg duration {results_metrics['holding_avg']} min."
)
@staticmethod
def _format_explanation_string(results, total_epochs) -> str:
return (("*" if results['is_initial_point'] else " ") +
f"{results['current_epoch']:5d}/{total_epochs}: " +
f"{results['results_explanation']} " +
f"Objective: {results['loss']:.5f}")
@staticmethod
2021-09-19 23:44:12 +00:00
def prepare_trials_columns(trials: pd.DataFrame, legacy_mode: bool,
has_drawdown: bool) -> pd.DataFrame:
trials['Best'] = ''
2021-06-09 15:03:24 +00:00
if 'results_metrics.winsdrawslosses' not in trials.columns:
# Ensure compatibility with older versions of hyperopt results
trials['results_metrics.winsdrawslosses'] = 'N/A'
2021-06-09 15:03:24 +00:00
if not has_drawdown:
# Ensure compatibility with older versions of hyperopt results
trials['results_metrics.max_drawdown_abs'] = None
trials['results_metrics.max_drawdown'] = None
2021-06-09 15:03:24 +00:00
if not legacy_mode:
# New mode, using backtest result for metrics
trials['results_metrics.winsdrawslosses'] = trials.apply(
lambda x: f"{x['results_metrics.wins']} {x['results_metrics.draws']:>4} "
f"{x['results_metrics.losses']:>4}", axis=1)
trials = trials[['Best', 'current_epoch', 'results_metrics.total_trades',
'results_metrics.winsdrawslosses',
'results_metrics.profit_mean', 'results_metrics.profit_total_abs',
'results_metrics.profit_total', 'results_metrics.holding_avg',
2021-06-07 21:15:26 +00:00
'results_metrics.max_drawdown', 'results_metrics.max_drawdown_abs',
'loss', 'is_initial_point', 'is_best']]
else:
# Legacy mode
trials = trials[['Best', 'current_epoch', 'results_metrics.trade_count',
'results_metrics.winsdrawslosses', 'results_metrics.avg_profit',
'results_metrics.total_profit', 'results_metrics.profit',
'results_metrics.duration', 'results_metrics.max_drawdown',
'results_metrics.max_drawdown_abs', 'loss', 'is_initial_point',
'is_best']]
trials.columns = ['Best', 'Epoch', 'Trades', ' Win Draw Loss', 'Avg profit',
'Total profit', 'Profit', 'Avg duration', 'Max Drawdown',
'max_drawdown_abs', 'Objective', 'is_initial_point', 'is_best']
2021-06-09 15:03:24 +00:00
return trials
@staticmethod
def get_result_table(config: dict, results: list, total_epochs: int, highlight_best: bool,
print_colorized: bool, remove_header: int) -> str:
"""
Log result table
"""
if not results:
return ''
tabulate.PRESERVE_WHITESPACE = True
trials = json_normalize(results, max_level=1)
legacy_mode = 'results_metrics.total_trades' not in trials
has_drawdown = 'results_metrics.max_drawdown_abs' in trials.columns
trials = HyperoptTools.prepare_trials_columns(trials, legacy_mode, has_drawdown)
trials['is_profit'] = False
trials.loc[trials['is_initial_point'], 'Best'] = '* '
trials.loc[trials['is_best'], 'Best'] = 'Best'
trials.loc[trials['is_initial_point'] & trials['is_best'], 'Best'] = '* Best'
trials.loc[trials['Total profit'] > 0, 'is_profit'] = True
trials['Trades'] = trials['Trades'].astype(str)
2021-05-01 11:32:53 +00:00
perc_multi = 1 if legacy_mode else 100
trials['Epoch'] = trials['Epoch'].apply(
lambda x: '{}/{}'.format(str(x).rjust(len(str(total_epochs)), ' '), total_epochs)
)
trials['Avg profit'] = trials['Avg profit'].apply(
2021-05-01 11:32:53 +00:00
lambda x: f'{x * perc_multi:,.2f}%'.rjust(7, ' ') if not isna(x) else "--".rjust(7, ' ')
)
trials['Avg duration'] = trials['Avg duration'].apply(
lambda x: f'{x:,.1f} m'.rjust(7, ' ') if isinstance(x, float) else f"{x}"
if not isna(x) else "--".rjust(7, ' ')
)
trials['Objective'] = trials['Objective'].apply(
lambda x: f'{x:,.5f}'.rjust(8, ' ') if x != 100000 else "N/A".rjust(8, ' ')
)
stake_currency = config['stake_currency']
2021-06-07 21:15:26 +00:00
if has_drawdown:
trials['Max Drawdown'] = trials.apply(
lambda x: '{} {}'.format(
round_coin_value(x['max_drawdown_abs'], stake_currency),
'({:,.2f}%)'.format(x['Max Drawdown'] * perc_multi).rjust(10, ' ')
).rjust(25 + len(stake_currency))
if x['Max Drawdown'] != 0.0 else '--'.rjust(25 + len(stake_currency)),
axis=1
)
else:
trials = trials.drop(columns=['Max Drawdown'])
trials = trials.drop(columns=['max_drawdown_abs'])
trials['Profit'] = trials.apply(
lambda x: '{} {}'.format(
round_coin_value(x['Total profit'], stake_currency),
2021-05-01 11:32:53 +00:00
'({:,.2f}%)'.format(x['Profit'] * perc_multi).rjust(10, ' ')
).rjust(25+len(stake_currency))
if x['Total profit'] != 0.0 else '--'.rjust(25+len(stake_currency)),
axis=1
)
trials = trials.drop(columns=['Total profit'])
if print_colorized:
for i in range(len(trials)):
if trials.loc[i]['is_profit']:
for j in range(len(trials.loc[i])-3):
trials.iat[i, j] = "{}{}{}".format(Fore.GREEN,
str(trials.loc[i][j]), Fore.RESET)
if trials.loc[i]['is_best'] and highlight_best:
for j in range(len(trials.loc[i])-3):
trials.iat[i, j] = "{}{}{}".format(Style.BRIGHT,
str(trials.loc[i][j]), Style.RESET_ALL)
trials = trials.drop(columns=['is_initial_point', 'is_best', 'is_profit'])
if remove_header > 0:
table = tabulate.tabulate(
trials.to_dict(orient='list'), tablefmt='orgtbl',
headers='keys', stralign="right"
)
table = table.split("\n", remove_header)[remove_header]
elif remove_header < 0:
table = tabulate.tabulate(
trials.to_dict(orient='list'), tablefmt='psql',
headers='keys', stralign="right"
)
table = "\n".join(table.split("\n")[0:remove_header])
else:
table = tabulate.tabulate(
trials.to_dict(orient='list'), tablefmt='psql',
headers='keys', stralign="right"
)
return table
@staticmethod
2021-09-19 23:44:12 +00:00
def export_csv_file(config: dict, results: list, csv_file: str) -> None:
"""
Log result to csv-file
"""
if not results:
return
# Verification for overwrite
if Path(csv_file).is_file():
logger.error(f"CSV file already exists: {csv_file}")
return
try:
io.open(csv_file, 'w+').close()
except IOError:
logger.error(f"Failed to create CSV file: {csv_file}")
return
trials = json_normalize(results, max_level=1)
trials['Best'] = ''
trials['Stake currency'] = config['stake_currency']
2021-08-08 09:02:54 +00:00
base_metrics = ['Best', 'current_epoch', 'results_metrics.total_trades',
'results_metrics.profit_mean', 'results_metrics.profit_median',
'results_metrics.profit_total',
'Stake currency',
'results_metrics.profit_total_abs', 'results_metrics.holding_avg',
'loss', 'is_initial_point', 'is_best']
perc_multi = 100
param_metrics = [("params_dict."+param) for param in results[0]['params_dict'].keys()]
trials = trials[base_metrics + param_metrics]
base_columns = ['Best', 'Epoch', 'Trades', 'Avg profit', 'Median profit', 'Total profit',
'Stake currency', 'Profit', 'Avg duration', 'Objective',
'is_initial_point', 'is_best']
param_columns = list(results[0]['params_dict'].keys())
trials.columns = base_columns + param_columns
trials['is_profit'] = False
trials.loc[trials['is_initial_point'], 'Best'] = '*'
trials.loc[trials['is_best'], 'Best'] = 'Best'
trials.loc[trials['is_initial_point'] & trials['is_best'], 'Best'] = '* Best'
trials.loc[trials['Total profit'] > 0, 'is_profit'] = True
trials['Epoch'] = trials['Epoch'].astype(str)
trials['Trades'] = trials['Trades'].astype(str)
trials['Median profit'] = trials['Median profit'] * perc_multi
trials['Total profit'] = trials['Total profit'].apply(
lambda x: f'{x:,.8f}' if x != 0.0 else ""
)
trials['Profit'] = trials['Profit'].apply(
lambda x: f'{x:,.2f}' if not isna(x) else ""
)
trials['Avg profit'] = trials['Avg profit'].apply(
lambda x: f'{x * perc_multi:,.2f}%' if not isna(x) else ""
)
trials['Objective'] = trials['Objective'].apply(
lambda x: f'{x:,.5f}' if x != 100000 else ""
)
trials = trials.drop(columns=['is_initial_point', 'is_best', 'is_profit'])
trials.to_csv(csv_file, index=False, header=True, mode='w', encoding='UTF-8')
logger.info(f"CSV file created: {csv_file}")