stable/tests/data/test_btanalysis.py

334 lines
13 KiB
Python
Raw Normal View History

from pathlib import Path
2019-06-16 08:57:21 +00:00
from unittest.mock import MagicMock
2019-03-16 16:50:57 +00:00
import pytest
2019-06-29 15:19:42 +00:00
from arrow import Arrow
from pandas import DataFrame, DateOffset, Timestamp, to_datetime
from freqtrade.configuration import TimeRange
2020-06-28 07:27:19 +00:00
from freqtrade.constants import LAST_BT_RESULT_FN
from freqtrade.data.btanalysis import (BT_DATA_COLUMNS, BT_DATA_COLUMNS_MID, BT_DATA_COLUMNS_OLD,
2021-02-14 18:44:13 +00:00
analyze_trade_parallelism, calculate_csum,
calculate_market_change, calculate_max_drawdown,
combine_dataframes_with_mean, create_cum_profit,
extract_trades_of_period, get_latest_backtest_filename,
get_latest_hyperopt_file, load_backtest_data, load_trades,
load_trades_from_db)
2019-09-07 18:34:25 +00:00
from freqtrade.data.history import load_data, load_pair_history
from tests.conftest import create_mock_trades
2020-09-10 05:40:19 +00:00
from tests.conftest_trades import MOCK_TRADE_COUNT
2019-03-16 16:50:57 +00:00
2020-06-27 04:46:54 +00:00
def test_get_latest_backtest_filename(testdatadir, mocker):
with pytest.raises(ValueError, match=r"Directory .* does not exist\."):
get_latest_backtest_filename(testdatadir / 'does_not_exist')
with pytest.raises(ValueError,
match=r"Directory .* does not seem to contain .*"):
get_latest_backtest_filename(testdatadir.parent)
res = get_latest_backtest_filename(testdatadir)
assert res == 'backtest-result_new.json'
2020-06-27 13:59:22 +00:00
res = get_latest_backtest_filename(str(testdatadir))
assert res == 'backtest-result_new.json'
2020-06-27 04:46:54 +00:00
2020-06-27 13:59:22 +00:00
mocker.patch("freqtrade.data.btanalysis.json_load", return_value={})
2020-06-27 04:46:54 +00:00
with pytest.raises(ValueError, match=r"Invalid '.last_result.json' format."):
get_latest_backtest_filename(testdatadir)
def test_get_latest_hyperopt_file(testdatadir, mocker):
res = get_latest_hyperopt_file(testdatadir / 'does_not_exist', 'testfile.pickle')
assert res == testdatadir / 'does_not_exist/testfile.pickle'
res = get_latest_hyperopt_file(testdatadir.parent)
assert res == testdatadir.parent / "hyperopt_results.pickle"
2020-09-27 22:35:19 +00:00
res = get_latest_hyperopt_file(str(testdatadir.parent))
assert res == testdatadir.parent / "hyperopt_results.pickle"
2020-06-27 08:06:59 +00:00
def test_load_backtest_data_old_format(testdatadir):
2019-03-16 16:50:57 +00:00
2019-09-07 18:56:03 +00:00
filename = testdatadir / "backtest-result_test.json"
2019-03-16 16:50:57 +00:00
bt_data = load_backtest_data(filename)
assert isinstance(bt_data, DataFrame)
assert list(bt_data.columns) == BT_DATA_COLUMNS_OLD + ['profit_abs', 'profit_ratio']
2019-03-16 16:50:57 +00:00
assert len(bt_data) == 179
# Test loading from string (must yield same result)
bt_data2 = load_backtest_data(str(filename))
assert bt_data.equals(bt_data2)
with pytest.raises(ValueError, match=r"File .* does not exist\."):
load_backtest_data(str("filename") + "nofile")
2019-06-16 08:57:21 +00:00
2020-06-27 08:06:59 +00:00
def test_load_backtest_data_new_format(testdatadir):
filename = testdatadir / "backtest-result_new.json"
bt_data = load_backtest_data(filename)
assert isinstance(bt_data, DataFrame)
assert set(bt_data.columns) == set(BT_DATA_COLUMNS_MID)
2020-06-27 08:06:59 +00:00
assert len(bt_data) == 179
# Test loading from string (must yield same result)
bt_data2 = load_backtest_data(str(filename))
assert bt_data.equals(bt_data2)
2020-06-28 07:51:49 +00:00
# Test loading from folder (must yield same result)
bt_data3 = load_backtest_data(testdatadir)
assert bt_data.equals(bt_data3)
2020-06-27 08:06:59 +00:00
with pytest.raises(ValueError, match=r"File .* does not exist\."):
load_backtest_data(str("filename") + "nofile")
2020-06-27 13:59:22 +00:00
with pytest.raises(ValueError, match=r"Unknown dataformat."):
2020-06-28 07:27:19 +00:00
load_backtest_data(testdatadir / LAST_BT_RESULT_FN)
2020-06-27 13:59:22 +00:00
def test_load_backtest_data_multi(testdatadir):
filename = testdatadir / "backtest-result_multistrat.json"
for strategy in ('DefaultStrategy', 'TestStrategy'):
bt_data = load_backtest_data(filename, strategy=strategy)
assert isinstance(bt_data, DataFrame)
assert set(bt_data.columns) == set(BT_DATA_COLUMNS_MID)
2020-06-27 13:59:22 +00:00
assert len(bt_data) == 179
# Test loading from string (must yield same result)
bt_data2 = load_backtest_data(str(filename), strategy=strategy)
assert bt_data.equals(bt_data2)
with pytest.raises(ValueError, match=r"Strategy XYZ not available in the backtest result\."):
load_backtest_data(filename, strategy='XYZ')
with pytest.raises(ValueError, match=r"Detected backtest result with more than one strategy.*"):
load_backtest_data(filename)
2020-06-27 08:06:59 +00:00
2019-06-16 08:57:21 +00:00
@pytest.mark.usefixtures("init_persistence")
def test_load_trades_from_db(default_conf, fee, mocker):
2019-06-16 08:57:21 +00:00
create_mock_trades(fee)
# remove init so it does not init again
2020-10-16 05:39:12 +00:00
init_mock = mocker.patch('freqtrade.data.btanalysis.init_db', MagicMock())
2019-06-16 08:57:21 +00:00
trades = load_trades_from_db(db_url=default_conf['db_url'])
2019-06-16 08:57:21 +00:00
assert init_mock.call_count == 1
2020-09-10 05:40:19 +00:00
assert len(trades) == MOCK_TRADE_COUNT
2019-06-16 08:57:21 +00:00
assert isinstance(trades, DataFrame)
assert "pair" in trades.columns
2020-06-26 07:21:28 +00:00
assert "open_date" in trades.columns
assert "profit_ratio" in trades.columns
for col in BT_DATA_COLUMNS:
if col not in ['index', 'open_at_end']:
assert col in trades.columns
2020-06-27 07:59:23 +00:00
trades = load_trades_from_db(db_url=default_conf['db_url'], strategy='DefaultStrategy')
2021-01-01 18:38:28 +00:00
assert len(trades) == 4
2020-06-27 07:59:23 +00:00
trades = load_trades_from_db(db_url=default_conf['db_url'], strategy='NoneStrategy')
assert len(trades) == 0
2019-09-07 18:56:03 +00:00
def test_extract_trades_of_period(testdatadir):
pair = "UNITTEST/BTC"
# 2018-11-14 06:07:00
timerange = TimeRange('date', None, 1510639620, 0)
data = load_pair_history(pair=pair, timeframe='1m',
2019-09-07 18:56:03 +00:00
datadir=testdatadir, timerange=timerange)
trades = DataFrame(
{'pair': [pair, pair, pair, pair],
'profit_ratio': [0.0, 0.1, -0.2, -0.5],
'profit_abs': [0.0, 1, -2, -5],
2020-06-26 07:21:28 +00:00
'open_date': to_datetime([Arrow(2017, 11, 13, 15, 40, 0).datetime,
Arrow(2017, 11, 14, 9, 41, 0).datetime,
Arrow(2017, 11, 14, 14, 20, 0).datetime,
Arrow(2017, 11, 15, 3, 40, 0).datetime,
], utc=True
),
2020-06-26 07:21:28 +00:00
'close_date': to_datetime([Arrow(2017, 11, 13, 16, 40, 0).datetime,
Arrow(2017, 11, 14, 10, 41, 0).datetime,
Arrow(2017, 11, 14, 15, 25, 0).datetime,
Arrow(2017, 11, 15, 3, 55, 0).datetime,
], utc=True)
})
trades1 = extract_trades_of_period(data, trades)
# First and last trade are dropped as they are out of range
assert len(trades1) == 2
2020-06-26 07:21:28 +00:00
assert trades1.iloc[0].open_date == Arrow(2017, 11, 14, 9, 41, 0).datetime
assert trades1.iloc[0].close_date == Arrow(2017, 11, 14, 10, 41, 0).datetime
assert trades1.iloc[-1].open_date == Arrow(2017, 11, 14, 14, 20, 0).datetime
assert trades1.iloc[-1].close_date == Arrow(2017, 11, 14, 15, 25, 0).datetime
2019-06-29 15:19:42 +00:00
def test_analyze_trade_parallelism(default_conf, mocker, testdatadir):
filename = testdatadir / "backtest-result_test.json"
bt_data = load_backtest_data(filename)
res = analyze_trade_parallelism(bt_data, "5m")
assert isinstance(res, DataFrame)
assert 'open_trades' in res.columns
assert res['open_trades'].max() == 3
assert res['open_trades'].min() == 0
def test_load_trades(default_conf, mocker):
db_mock = mocker.patch("freqtrade.data.btanalysis.load_trades_from_db", MagicMock())
bt_mock = mocker.patch("freqtrade.data.btanalysis.load_backtest_data", MagicMock())
2019-08-22 18:17:36 +00:00
load_trades("DB",
db_url=default_conf.get('db_url'),
exportfilename=default_conf.get('exportfilename'),
2020-06-27 07:59:23 +00:00
no_trades=False,
strategy="DefaultStrategy",
2019-08-22 18:17:36 +00:00
)
assert db_mock.call_count == 1
assert bt_mock.call_count == 0
db_mock.reset_mock()
bt_mock.reset_mock()
default_conf['exportfilename'] = Path("testfile.json")
2019-08-22 18:17:36 +00:00
load_trades("file",
db_url=default_conf.get('db_url'),
2020-03-14 21:15:03 +00:00
exportfilename=default_conf.get('exportfilename'),
)
assert db_mock.call_count == 0
assert bt_mock.call_count == 1
2020-03-14 23:09:08 +00:00
db_mock.reset_mock()
bt_mock.reset_mock()
default_conf['exportfilename'] = "testfile.json"
load_trades("file",
db_url=default_conf.get('db_url'),
exportfilename=default_conf.get('exportfilename'),
2020-03-15 20:20:32 +00:00
no_trades=True
2020-03-14 23:09:08 +00:00
)
assert db_mock.call_count == 0
assert bt_mock.call_count == 0
def test_calculate_market_change(testdatadir):
pairs = ["ETH/BTC", "ADA/BTC"]
data = load_data(datadir=testdatadir, pairs=pairs, timeframe='5m')
result = calculate_market_change(data)
assert isinstance(result, float)
assert pytest.approx(result) == 0.00955514
def test_combine_dataframes_with_mean(testdatadir):
2019-10-02 08:59:45 +00:00
pairs = ["ETH/BTC", "ADA/BTC"]
data = load_data(datadir=testdatadir, pairs=pairs, timeframe='5m')
df = combine_dataframes_with_mean(data)
2019-06-30 08:04:43 +00:00
assert isinstance(df, DataFrame)
assert "ETH/BTC" in df.columns
2019-10-02 08:59:45 +00:00
assert "ADA/BTC" in df.columns
2019-06-30 08:04:43 +00:00
assert "mean" in df.columns
2019-09-07 18:56:03 +00:00
def test_create_cum_profit(testdatadir):
filename = testdatadir / "backtest-result_test.json"
2019-06-29 15:19:42 +00:00
bt_data = load_backtest_data(filename)
timerange = TimeRange.parse_timerange("20180110-20180112")
2019-06-29 15:19:42 +00:00
df = load_pair_history(pair="TRX/BTC", timeframe='5m',
2019-09-07 18:56:03 +00:00
datadir=testdatadir, timerange=timerange)
2019-06-29 15:19:42 +00:00
cum_profits = create_cum_profit(df.set_index('date'),
bt_data[bt_data["pair"] == 'TRX/BTC'],
"cum_profits", timeframe="5m")
2019-06-29 15:19:42 +00:00
assert "cum_profits" in cum_profits.columns
assert cum_profits.iloc[0]['cum_profits'] == 0
assert cum_profits.iloc[-1]['cum_profits'] == 0.0798005
2019-10-28 13:30:01 +00:00
def test_create_cum_profit1(testdatadir):
filename = testdatadir / "backtest-result_test.json"
bt_data = load_backtest_data(filename)
# Move close-time to "off" the candle, to make sure the logic still works
2020-06-26 07:21:28 +00:00
bt_data.loc[:, 'close_date'] = bt_data.loc[:, 'close_date'] + DateOffset(seconds=20)
2019-10-28 13:30:01 +00:00
timerange = TimeRange.parse_timerange("20180110-20180112")
df = load_pair_history(pair="TRX/BTC", timeframe='5m',
2019-10-28 13:30:01 +00:00
datadir=testdatadir, timerange=timerange)
cum_profits = create_cum_profit(df.set_index('date'),
bt_data[bt_data["pair"] == 'TRX/BTC'],
2019-10-28 13:30:01 +00:00
"cum_profits", timeframe="5m")
assert "cum_profits" in cum_profits.columns
assert cum_profits.iloc[0]['cum_profits'] == 0
assert cum_profits.iloc[-1]['cum_profits'] == 0.0798005
2020-03-03 06:20:41 +00:00
with pytest.raises(ValueError, match='Trade dataframe empty.'):
create_cum_profit(df.set_index('date'), bt_data[bt_data["pair"] == 'NOTAPAIR'],
"cum_profits", timeframe="5m")
2020-03-03 06:20:41 +00:00
def test_calculate_max_drawdown(testdatadir):
filename = testdatadir / "backtest-result_test.json"
bt_data = load_backtest_data(filename)
2021-02-14 18:30:17 +00:00
drawdown, hdate, lowdate, hval, lval = calculate_max_drawdown(bt_data)
2020-03-03 06:20:41 +00:00
assert isinstance(drawdown, float)
assert pytest.approx(drawdown) == 0.21142322
2021-02-14 18:30:17 +00:00
assert isinstance(hdate, Timestamp)
assert isinstance(lowdate, Timestamp)
assert isinstance(hval, float)
assert isinstance(lval, float)
assert hdate == Timestamp('2018-01-24 14:25:00', tz='UTC')
assert lowdate == Timestamp('2018-01-30 04:45:00', tz='UTC')
2020-03-03 06:20:41 +00:00
with pytest.raises(ValueError, match='Trade dataframe empty.'):
2021-02-14 18:30:17 +00:00
drawdown, hdate, lowdate, hval, lval = calculate_max_drawdown(DataFrame())
2021-02-14 18:44:13 +00:00
def test_calculate_csum(testdatadir):
filename = testdatadir / "backtest-result_test.json"
bt_data = load_backtest_data(filename)
csum_min, csum_max = calculate_csum(bt_data)
assert isinstance(csum_min, float)
assert isinstance(csum_max, float)
assert csum_min < 0.01
assert csum_max > 0.02
csum_min1, csum_max1 = calculate_csum(bt_data, 5)
assert csum_min1 == csum_min + 5
assert csum_max1 == csum_max + 5
2021-02-14 18:44:13 +00:00
with pytest.raises(ValueError, match='Trade dataframe empty.'):
csum_min, csum_max = calculate_csum(DataFrame())
def test_calculate_max_drawdown2():
values = [0.011580, 0.010048, 0.011340, 0.012161, 0.010416, 0.010009, 0.020024,
-0.024662, -0.022350, 0.020496, -0.029859, -0.030511, 0.010041, 0.010872,
-0.025782, 0.010400, 0.012374, 0.012467, 0.114741, 0.010303, 0.010088,
-0.033961, 0.010680, 0.010886, -0.029274, 0.011178, 0.010693, 0.010711]
dates = [Arrow(2020, 1, 1).shift(days=i) for i in range(len(values))]
2020-06-26 07:21:28 +00:00
df = DataFrame(zip(values, dates), columns=['profit', 'open_date'])
# sort by profit and reset index
df = df.sort_values('profit').reset_index(drop=True)
df1 = df.copy()
2021-02-14 18:30:17 +00:00
drawdown, hdate, ldate, hval, lval = calculate_max_drawdown(
df, date_col='open_date', value_col='profit')
# Ensure df has not been altered.
assert df.equals(df1)
assert isinstance(drawdown, float)
# High must be before low
2021-02-14 18:30:17 +00:00
assert hdate < ldate
# High value must be higher than low value
assert hval > lval
assert drawdown == 0.091755
2020-04-05 12:43:01 +00:00
2020-06-26 07:21:28 +00:00
df = DataFrame(zip(values[:5], dates[:5]), columns=['profit', 'open_date'])
2020-04-05 12:43:01 +00:00
with pytest.raises(ValueError, match='No losing trade, therefore no drawdown.'):
2020-06-26 07:21:28 +00:00
calculate_max_drawdown(df, date_col='open_date', value_col='profit')