self.detailDf = single_trade_details(self.df,
poChgCol=poChgCol,
originSignal=originSignal,
at_once=at_once,
mode=mode)
self.evalSer = overall_eval(self.df, self.detailDf)
self.yearDf = year_month_analysis(self.detailDf)
self.plot_equity_curve(period=period, contrast=contrast, labels=labels)
#################### Single_Proportion用于日线策略的回测例子 #############################################################
# 这个是single_proportion_backtest的使用例子，以及在图上显示回测结果,参数是at_once=True,mode='longshort'
if __name__ == '__main__':
import talib as ta
from tools.data import fetch
df = fetch.index_one('000001.SH', endTime='2010-12-31') # 短一点，避免卡
# 计算DEA指标
df['DIFF'], df['DEA'], df['MACD'] = ta.MACD(df['CLOSE'])
df.dropna(axis=0, inplace=True)
df['MACD'] = df['MACD'] * 2
from tools.backtest import signal
# 生成原始信号和买卖标记
signal.up_cross(df, fastline='DIFF', slowline='DEA') # 多了一列signal
signal.signal2position_single(df,
signalCol='signal',
at_once=True,
mode='longshort') # 仓位：即期，多空，止损，保证金
# 单个品种的回测
btest = Single_Proportion(df,
# 第一个划分牛熊的想法
if __name__ == '__main__':
from tools.data import fetch
import datetime as dt
import pandas as pd
from tools.tinytools import pandas_related, group_method
df = fetch.index_one('000001.SH', startTime='2000-01-01')
df['6_month_peak'] = df['HIGH'].rolling(130).max()
df['1_month_peak'] = df['HIGH'].rolling(22).max()
df['3_month_bottom'] = df['LOW'].rolling(65).min()
df['12_month_bottom'] = df['LOW'].rolling(250).min()
df['24_month_bottom'] = df['LOW'].rolling(510).min()
df['3_month_std'] = df['CLOSE'].rolling(65).std()
df['bb_tag'] = None
pandas_related.cal_intraperiod_return(df)
# 分牛熊
for index in df.index[1:]:
preIndex = pandas_related.previous_index(df, index)
# if index[0] == dt.datetime(2009, 4, 24):
# pass
if df.at[index, 'monthly_return'] > 0.1 \
and df.at[index, 'CLOSE'] / df.at[index, '3_month_bottom'] > 1.4 \
and df.at[preIndex, 'bb_tag'] is None:
df.at[index, 'bb_tag'] = '牛市'
_base = df.at[index, '3_month_bottom']
if __name__ == '__main__':
import pandas as pd
from tools.data import fetch
from tools.backtest import timing, signal
dfzz500 = fetch.index_one('000905.SH',startTime='2007-01-15')
dfsz50 = fetch.index_one('000016.SH',startTime='2007-01-15')
df = pd.DataFrame()
df['target'] = dfsz50.reset_index(level=1)['PCTCHANGE'] - dfzz500.reset_index(level=1)['PCTCHANGE']
df['sz50'] = dfsz50.reset_index(level=1)['CLOSE']
df['zz500'] = dfzz500.reset_index(level=1)['CLOSE']
df['return_sz50'] = dfsz50.reset_index(level=1)['PCTCHANGE'] / 100
df['return_zz500'] = dfzz500.reset_index(level=1)['PCTCHANGE'] / 100
signal.up_cross(df, 'target')
signal.signal2position_multi(df, signalCol='signal',
mode='longshort', )
timing.multi_proportion_backtest(df,['sz50','zz500'],[1,-1],[0.5,0.5],tCost=0.0002,)
detailDf = timing.multi_trade_details(df,originSignal='signal',pxColList=['sz50','zz500'],longshortList=[1,-1],proportionList=[0.5,0.5])
evalSer = timing.overall_eval(df,detailDf)
yearDf = timing.year_month_analysis(detailDf)
f"self.window.{minIndex}.patches.remove(self.allDf.at['{minIndex}', 'musk'])"
)
self.allDf.at[minIndex, 'musk'] = None
self.window.fig.canvas.draw_idle()
if __name__ == '__main__':
# intradayWindow = Intraday_General()
dfDay = fetch.stock_one('601318',
startTime='2014-04-28',
endTime='2018-08-10')
df1min = fetch.n_min('601318', 1)
intraday = Intraday_General(dfDay=dfDay, df1min=df1min)
dftechMin = fetch.n_min('000001.SH',
startTime='2014-04-28',
endTime='2018-08-10')
dftechDay = fetch.index_one('000001.SH',
startTime='2014-04-28',
endTime='2018-08-10')
intraday.attach_min_data(dfminTech=dftechMin,
axName='tech1',
className='Intraday_Close',
precloseSrcDf=dftechDay)
intraday.attach_day_data(dfdayTech=dftechDay,
axName='tech2',
className='Candle')
intraday.musk(if_day=False, if_min=False)
df['poChg'] = df['poChg_real']
if df.at[df.loc[df['poChg'] != 0.0].index[0], 'poChg'] \
* df.at[df.loc[df['poChg'] != 0.0].index[-1], 'poChg'] < 0:
pass
else:
df.at[df.loc[df['poChg'] != 0.0].index[-1], 'poChg'] = 0
df.drop('poChg_real', axis=1, inplace=True)
#################### 产生信号函数的使用范例 ##############################################################################
if __name__ == '__main__':
# 这个是up_cross的使用例子
import talib as ta
from tools.data import fetch
df = fetch.index_one('000001.SH', endTime='2010-12-31') # 短一点，避免卡
# 计算DEA指标
df['DIFF'], df['DEA'], df['MACD'] = ta.MACD(df['CLOSE'])
df.dropna(axis=0, inplace=True)
df['MACD'] = df['MACD'] * 2
# 生成买卖标记
up_cross(df, fastline='DIFF', slowline='DEA') # 多了一列signal
del df
if __name__ == '__main__':
# 这个是long_short_signal的使用例子
import talib as ta
from tools.data import fetch
df = fetch.index_one('000001.SH', endTime='2010-12-31') # 短一点，避免卡
groupCopy.dropna(inplace=True)
return groupCopy
def groupFunc_find_updown_ratio(group):
return pd.DataFrame([
len(group.loc[group['确认点'] == 1].index),
len(group.loc[group['确认点'] == -1].index)
],
index=['up', 'down']).T
# 下面是日线级别的行业趋势度的计算
if __name__ == '__main__':
from tools.data import fetch
dfAllA = fetch.index_one('881001.WI')
# 取中信二级行业来代替原报告中申万二级行业
dfCiticslvl2 = fetch.index_industry_list('all', citicslevel=2)
dfCiticslvl2.dropna(inplace=True)
dfCiticslvl2.drop(['S_INFO_NAME', 'S_INFO_INDUSTRYCODE', 'level'],
axis=1,
inplace=True)
# 对中信2级行业中每个行业计算DEA分段新逻辑的买卖确认点
dfCiticslvl2 = dfCiticslvl2.groupby(
level=1).apply(groupfunc_find_subindustry_buysell)
dfCiticslvl2.reset_index(level=0, inplace=True, drop=True)
dfCiticslvl2.sort_index(level=[0, 1], inplace=True)
# 由于数据原因，前面的日期处于状态内的行业数只有少数几个，后面有些行业处于状态数也比较小，一个简单的标准是取大于40的日期。
_groupCountTemp = dfCiticslvl2.groupby(level=0)['CLOSE'].count()