def setUp(self): self.sym = "EURUSD" self.aEq = self.assertEqual self.aIn = self.assertIn self.aRaise = self.assertRaises self.aItEq = self.assertItemsEqual self.bt = BackTest()
def cal_alphas(self, expr_ds):
# 计算信号
for i in [
'OpenPrice', 'ClosePrice', 'HighestPrice', 'LowestPrice',
'Volume', 'VWAP', 'Position', 'TurnOver'
]:
expr_tmp = expr_ds.values.expr.replace(i, 'self.eod.%s' % i)
alphas = eval(expr_tmp)
resample_wgts = pd.DataFrame(alphas[:-1] * 1.,
index=self.eod.dates[1:],
columns=self.eod.ticker_names)
bt = BackTest(resample_wgts,
returns=self.resample_returns,
cycle='day',
IS_OOS_ratio=None,
stat_info=False,
plot=False,
quintiles=3,
turnover=1,
cost=0.,
ticker_names=None,
output_dir=None,
test_mode=False,
signal_name=None)
bt.stat_quintiles()
bt.stat_quintiles_pnl()
bt.stat_alpha_pnl()
bt.stat_alpha_sharpe()
# 保存信号
print expr_ds.values.id, expr_ds.values.expr, 'sharpe:', bt.alpha_sharpe
def should_generate_errors_over_time_for_each_model(self):
first_start_datetime = [
'2016-07-16 09:00:00-04:00', '2016-08-16 09:00:00-04:00'
]
l1_time = [9, 10]
l1_day = [0, 0]
schedule_type = [1, 1]
start_month = [7, 8]
start_year = [2016, 2016]
user_tz = ['Brasilia', 'Brasilia']
args = {
'l1_time': l1_time,
'l1_day': l1_day,
'schedule_type': schedule_type,
'first_start_datetime': first_start_datetime,
'start_month': start_month,
'start_year': start_year,
'user_tz': user_tz
}
summaries = pd.DataFrame(args)
summaries['first_start_datetime'] = pd.to_datetime(
summaries['first_start_datetime'])
bt = BackTest(data=summaries,
model=DumbModel,
training_data_span_months=1)
error = bt.errors()
expect('2016-8' in error.index.values).to.equal(True)
expect('2016-7' not in error.index.values).to.equal(True)
expect('2016-6' not in error.index.values).to.equal(True)
from backtest import BackTest
import pandas as pd
#매일 장이 끝난 후에 이 파일을 돌려야 백테스트를 위한 분봉데이터를 수집할 수 있음.
if __name__ == "__main__":
#객체 생성
backtest = BackTest()
#통신 확인
if backtest.InitPlusCheck() == False:
exit()
# 전체 코스피, 코스닥 코드 불러오기
entire_code, entire_name = backtest.get_entire_code()
# code = backtest.get_yesterday_highest()
# # 어제 상한가 종목 및 날짜 불러오기
# date_company = pd.DataFrame([[backtest.today,code]],columns=['내일날짜','상한가다음날'])
# # 해당 분봉 데이터 업데이트
# backtest.get_minute_stock(date_company, entire_code, int(backtest.today))
# 특정 날짜 빼먹었을 때 데이터 축적
code = ['A003495']
specific_date = '20201117'
date_company = pd.DataFrame([[specific_date, code]],
columns=['내일날짜', '상한가다음날'])
backtest.get_minute_stock(date_company, entire_code, int(specific_date))
# 긴 기간을 구하고 싶을때
# change = backtest.get_change(entire_code, 100)
# date_company = backtest.date_company_extract(change)
config = ConfigParser()
config.read('./config/config-excel.conf')
def create_handlers(config, event_queue):
"""
create data_handler, strategy, portfolio, execution
"""
data_handler = FileMDEngine(config, event_queue)
strategy = ExcelStrategy(config, event_queue, data_handler)
execution = ExcelExecution(config, event_queue, data_handler)
portfolio = ExcelPortfolio(config, event_queue, data_handler,
execution)
return data_handler, strategy, portfolio, execution
test = BackTest(config, create_handlers)
test.run_backtest()
metric = PerformanceMetric(test, config)
metric.multiplier = 10
metrics = metric.calculate_performance()
df1 = metrics.get('data')
columns = [
'close', 'position', 'tradepnl_wo_commission', 'cumpnl_wo_commission'
]
df1 = df1[columns]
df2 = load_excel_result()
df2.columns = columns
if __name__ == '__main__':
from talib import SMA
import pandas as pd
from dataload import get_sample_data
from mytrader import HelloWorldTrade
from backtest import BackTest
from utils.visualize import visuialize_performance
df = get_sample_data()
trading_data = df[['Close']].rename(columns={'Close': '2380_close'})
indexes = df # we don't use any index to support the strategy in this hello-world trader
trader = HelloWorldTrade()
back_tester = BackTest(trading_data, indexes)
back_tester.evaluate(trader)
data_to_plot = trading_data.copy()
data_to_plot['13MA'] = SMA(trading_data['2380_close'], 13)
data_to_plot['34MA'] = SMA(trading_data['2380_close'], 34)
visuialize_performance(back_tester, data_to_plot[['2380_close']],
data_to_plot[['13MA']], data_to_plot[['34MA']])
print("see result in output dir!")
