def testSelectSignalByDateAndInnerCode(self):
signalData = SourceDataDao.load_signal_data('')
signalData = SourceDataDao.select_signal_by_date_and_inner_code(signalData, '2004-01-31', '000049.SZ')
# self.assertEqual(self.sum(1, 2), 3)
print('signalData:'+str(len(signalData)))
#print(signalData)
self.assertNotEqual(len(signalData), 0)
def createFakeSignalData(addr):
signalData = SourceDataDao.load_signal_data(addr)
startDate = signalData.index[0][0]
endDate = signalData.index[-1][0]
startYear = DateUtil.datetime2_year_str(startDate)
endYear = DateUtil.datetime2_year_str(endDate)
FORMAT = "%d-%d-%d"
# c = time.strftime('%Y%m%d', time.strptime(y, '%Y-%m-%d'))
for year in range(int(startYear), int(endYear) + 1, 1):
for month in range(1, 13):
d = calendar.monthrange(year, month)
exportStartDate = FORMAT % (year, month, 1)
exportEndDate = FORMAT % (year, month, d[1])
# exportStartDate = str(year)+'-'+addZero(month)+'-'+'01'
# exportEndDate = str(year) + '-' + addZero(month) + '-' + getMonthEndDay(month)
subSignalData = SourceDataDao.select_by_date(
signalData, exportStartDate, exportEndDate)
print('exportStartDate:' + exportStartDate)
print('exportEndDate:' + exportEndDate)
print('subSignalData:' + str(len(subSignalData)))
if (len(subSignalData) > 0):
newSubSignalData = subSignalData.copy()
newSubSignalData['Mom2'] = subSignalData['Mom']
newSubSignalData['Mom3'] = subSignalData['Mom']
SourceDataDao.export_to_hdfstore(
newSubSignalData,
StockConst.ROOT + StockConst.FAKE_SIGNAL_DATA_H5)
def printSingalDataByDate(date,addr):
signalData = SourceDataDao.load_signal_data(addr)
t1 = pd.Timestamp(date)
#t2 = pd.Timestamp(endDate)
#signalData = signalData.loc[t1:t2, :]
signalData = signalData[signalData.index.get_level_values(0) == t1]
signalData = signalData.sort_values(by=[StockConst.MOM], ascending=False)
print(signalData)
def addBuySellFlgToSignalData2(signalDataAddr, dailyQuoteAddr):
signalData = SourceDataDao.load_signal_data(signalDataAddr)
dailyQuote = SourceDataDao.load_new_daily_quote(dailyQuoteAddr)
groupedSignalData = signalData.groupby(level=StockConst.TRADINGDAY,
as_index=False).apply(
SelectUtil.top, 5,
StockConst.MOM, False)
groupedSignalData[StockConst.BUY_FLG] = dailyQuote[StockConst.BUY_FLG]
groupedSignalData[StockConst.SELL_FLG] = dailyQuote[StockConst.SELL_FLG]
# groupedSignalData.insert(1, StockConst.buyFlg, Series())
# groupedSignalData.insert(2, StockConst.sellFlg, Series())
# for index, row in groupedSignalData.iterrows():
# # print('tradingDate:'+str(index[0])+',innerCode:'+str(index[2]))
# groupedSignalData.ix[index, [StockConst.buyFlg]] = dailyQuote.ix[(index[0], index[2]), [StockConst.buyFlg]]
# groupedSignalData.ix[index, [StockConst.sellFlg]] = dailyQuote.ix[(index[0], index[2]), [StockConst.sellFlg]]
groupedSignalData.to_csv(StockConst.ROOT + '\export\groupedSignalData.csv')
def testSelect_stock_func1(self):
signalData = SourceDataDao.load_signal_data('')
techList = [StockConst.MOM]
sliceIdx = None
numOfDailySignal = 5
sliceTotalNum = None
sliceDict = {
'sliceIdx': sliceIdx,
'numOfDailySignal': numOfDailySignal,
'sliceTotalNum': sliceTotalNum
}
groupedSignalData = BackTestMainNew.select_stock_func1(
signalData, techList, sliceDict)
currSignalData = groupedSignalData.xs('2005-01-05').xs('2005-01-05')
entity = currSignalData.ix['000950.SZ']
print(entity)
def test10(addr):
signalData = SourceDataDao.load_signal_data(addr)
groupedSignalData = signalData.groupby(level=StockConst.TRADINGDAY, as_index=False).apply(SelectUtil.top, 5, StockConst.MOM, False)
print(groupedSignalData)
def exportSignal():
signalData = SourceDataDao.load_signal_data(StockConst.ROOT + '/Mom_5_0302.h5')
# signalData.to_csv(StockConst.root + '\export2\signalData.csv')
# signalData.to_excel(StockConst.root + '\export2\signalData.xls', sheet_name='Sheet1')
signalData.to_json(StockConst.ROOT + '\export2\signalData.json')
signalData.to_hdf(StockConst.ROOT + '\export2\signalData.h5')
def printGroupedSingalData(addr):
signalData = SourceDataDao.load_signal_data(addr)
groupedSignalData = signalData.groupby(level=StockConst.TRADINGDAY).apply(SelectUtil.top, 5, StockConst.MOM, False)
groupedSignalData.to_csv(StockConst.ROOT + '\export\groupedSignalData.csv')
def main(select_stock_1day_func,trade_1day_func,techList,signalDataAddr = '', dailyQuoteAddr = '', startDate = '2004-01-01', endDate = '2005-12-31'):
start = time.clock()
signalData = SourceDataDao.load_signal_data(signalDataAddr) #ByDate
#dailyQuote = SourceDataDao.loadDailyQuote()
dailyQuote = SourceDataDao.load_new_daily_quote(dailyQuoteAddr) #ByDate
start_1 = time.clock()
signalData = signalData[(signalData.index.get_level_values(0) >= startDate) & (signalData.index.get_level_values(0) <= endDate) ]
end_1 = time.clock()
print("signalData花费时间：%f s" % (end_1 - start_1))
# print(startDate)
# print(endDate)
dateList = SourceDataDao.select_date_from_signal(signalData, startDate, endDate)
# print(dateList)
# 统计结果
stockStatDailyDf = pd.DataFrame(index=pd.date_range(startDate, endDate),
columns=['netValue', 'changePCT', 'buyCnt', 'sellCnt', 'prevHoldCnt',
'currHoldCnt', 'cannotSellCnt', 'cannotBuyCnt',
'usableCach', 'mv'])
# 每日持仓
stockHoldDailyList = []
# 交易结果
stockTradeList = []
start_trade = time.clock()
# 交易
t1d = process1day(startDate, endDate, dailyQuote, signalData, dateList, stockStatDailyDf, stockHoldDailyList, stockTradeList)
res = signalData.groupby(level=StockConst.TRADINGDAY).apply(t1d.deal, techList, select_stock_1day_func, trade_1day_func)
end_trade = time.clock()
print("交易花费时间：%f s" % (end_trade - start_trade))
result = ''
start_stat = time.clock()
# 信号数据
#groupedSelectStock.to_csv(StockConst.root + '\export\groupedSelectStock.csv')
# 每日交易
# print('每日交易:')
# print(stockTradeList)
stockTradeDailyDf = pd.DataFrame(stockTradeList)
stockTradeDailyDf.sort_values(by=['tradingDate'], ascending=True)
# stockTradeDailyDf.to_csv(StockConst.root + '\export\stockTradeDaily.csv')
# 每日持仓
#print('每日持仓:')
stockHoldDailyDf = pd.DataFrame(stockHoldDailyList)
stockHoldDailyDf.sort_values(by=['tradingDate'],ascending=True)
# stockHoldDailyDf.to_csv(StockConst.root + '\export\stockHoldDaily.csv')
#每日统计(收益，净值，买入数，卖出数，持有数)
stockStatDailyDf = stockStatDailyDf.dropna(how='all')
result += '--' * 70 + '\n'
result += '每日统计:' + '\n'
result += stockStatDailyDf.to_string() + '\n'
# stockStatDailyDf.to_csv(StockConst.root + '\export\stockStatDaily.csv')
# 每年统计
result += '--' * 70 + '\n'
yearDf = StockYearService.main(stockStatDailyDf)
result += '每年统计:' + '\n'
result += yearDf.to_string() + '\n'
# 最大回撤
result += '--' * 70 + '\n'
maxdrop = StockMaxDropNewService.get_max_drop(stockStatDailyDf)
result += '最大回撤:' + '\n'
result += SetUtil.dict_to_string(maxdrop) + '\n'
#每年的最大回撤
maxdropDf = StockMaxDropNewService.get_max_drop_for_each_year(stockStatDailyDf)
maxdropDf.sort_values(by=["year"])
result += '每年的最大回撤:' + '\n'
result += maxdropDf.to_string() + '\n'
#夏普比率
result += '--' * 70 + '\n'
sharpRatio = SharpRatioNewService.get_sharp_ratio(stockStatDailyDf)
result += '夏普比率:' + '\n'
result += str(sharpRatio) + '\n'
#每年的夏普比率
sharpRatioDf = SharpRatioNewService.get_sharp_ratio_for_each_year(stockStatDailyDf)
sharpRatioDf.sort_values(by=["year"])
result += '每年的夏普比率:' + '\n'
result += sharpRatioDf.to_string() + '\n'
end_stat = time.clock()
print("统计花费时间：%f s" % (end_stat - start_stat))
# result += '回测完成'
end = time.clock()
print("main花费时间：%f s" % (end - start))
# 净值曲线图
stockStatDailyDf['netValue'].plot()
plt.show()
resultDic={'stockStatDailyDf':stockStatDailyDf,'result':result}
print(result)
return resultDic
def testLoadSignalData(self):
signalData = SourceDataDao.load_signal_data('')
# self.assertEqual(self.sum(1, 2), 3)
print('signalData:' + str(len(signalData)))
self.assertNotEqual(len(signalData), 0)
def main(select_stock_func,
trade_func,
techList,
signalDataAddr='',
dailyQuoteAddr='',
startDate='1/1/2004',
endDate='12/31/2005'):
start = time.clock()
signalData = SourceDataDao.load_signal_data(signalDataAddr)
#dailyQuote = SourceDataDao.loadDailyQuote()
dailyQuote = SourceDataDao.load_new_daily_quote(dailyQuoteAddr)
#参数
#每天信号数
#numOfDailySignal = 5
#初始资金
initialMV = 1000000 #1000000
#开始结束日期
#startDate = '1/8/2001'
#startDate = '1/1/2004'
#endDate = '1/1/2017'
#eendDate = '5/18/2001'
#endDate = '12/31/2005'
#endDate = '12/31/2002'
#endDate = '1/9/2001'
#选股结果（外部传入）
#select top 5 group by TradingDay order by Mom desc
groupedSignalData = select_stock_func(signalData, techList)
#当前持仓 key:innerCode value:
currHoldSet = {}
#资金情况
capitalEntity = CapitalEntity.CapitalEntity(initialMV)
lastMV = initialMV
#初始化每日统计表
stockStatDailyDf = pd.DataFrame(index=pd.date_range(startDate, endDate),
columns=[
'netValue', 'changePCT', 'buyCnt',
'sellCnt', 'prevHoldCnt',
'currHoldCnt', 'cannotSellCnt',
'cannotBuyCnt', 'usableCach', 'mv'
])
#初始化每日持仓表
stockHoldDailyList = []
stockTradeList = []
#从信号表中取得唯一性日期
dateList = SourceDataDao.select_date_from_signal(signalData, startDate,
endDate)
# print(dateList)
start_loop = time.clock()
for index, item in enumerate(dateList):
#for date in dateList:
#print('index:'+str(index)+' len:'+str(len(dateList)))
start_inloop0 = time.clock()
#信号日
signalDay = dateList[index]
# 非最后一天
if index < (len(dateList) - 1):
#交易日
tradingDay = dateList[index + 1]
tradingDayStr = DateUtil.datetime2_str(tradingDay)
#print('dateStr:'+dateStr)
# 最后一天
else:
break
#print(dateStr)
currSignalData = groupedSignalData.ix[signalDay]
# print(currSignalData)
if StockConst.IS_DEBUG:
print("currSignalData:" + str(len(currSignalData)))
#print(currSignalData)
end_inloop0 = time.clock()
print("inloop0花费时间：%f s" % (end_inloop0 - start_inloop0))
#交易方法（外部传入）
start_inloop1 = time.clock()
dict = trade_func(currSignalData, currHoldSet, dailyQuote, tradingDay,
signalData, capitalEntity)
end_inloop1 = time.clock()
print("trade_func花费时间：%f s" % (end_inloop1 - start_inloop1))
actualBuyList = dict['actualBuyList']
actualSellList = dict['actualSellList']
prevHoldList = dict['prevHoldList']
cannotBuyList = dict['cannotBuyList']
cannotSellList = dict['cannotSellList']
stockTradeListDaily = dict['stockTradeList']
if len(stockTradeListDaily) > 0:
stockTradeList.extend(stockTradeListDaily)
# print(currHoldSet)
#3.计算当日市值
start_inloop2 = time.clock()
currMV = calculateDailyMV(currHoldSet, capitalEntity, dailyQuote,
tradingDay, stockHoldDailyList, initialMV)
end_inloop2 = time.clock()
print("calculateDailyMV花费时间：%f s" % (end_inloop2 - start_inloop2))
start_inloop3 = time.clock()
#4.个数
buyCnt = len(actualBuyList)
sellCnt = len(actualSellList)
prevHoldCnt = len(prevHoldList)
currHoldCnt = len(currHoldSet)
cannotSellCnt = len(cannotSellList)
cannotBuyCnt = len(cannotBuyList)
# print(currMV)
#if StockConst.isDebug:
#print("dateStr:" + dateStr + " changePCTBuy:" + str(changePCTBuy) + " changePCTSell:" + str(changePCTSell) +
#" changePCTHold:" + str(changePCTHold))
#当日净值
netValue = currMV / initialMV
#print("netValue:" + str(netValue))
#当日收益
dailyChangePCT = NumUtil.get_change_pct(lastMV, currMV, 2)
#当日可用现金
usableCach = capitalEntity.get_usable_cash()
#
stockStatDailyDf.ix[
tradingDayStr] = netValue, dailyChangePCT, buyCnt, sellCnt, prevHoldCnt, currHoldCnt, cannotSellCnt, cannotBuyCnt, usableCach, currMV
#
lastMV = currMV
end_inloop3 = time.clock()
print("inloop3花费时间：%f s" % (end_inloop3 - start_inloop3))
pass
end_loop = time.clock()
print("for循环花费时间：%f s" % (end_loop - start_loop))
result = ''
start_stat = time.clock()
# 信号数据
#groupedSignalData.to_csv(StockConst.root + '\export\groupedSignalData.csv')
# 每日交易
# print('每日交易:')
# print(stockTradeList)
stockTradeDailyDf = pd.DataFrame(stockTradeList)
stockTradeDailyDf.sort_values(by=['tradingDate'], ascending=True)
# stockTradeDailyDf.to_csv(StockConst.root + '\export\stockTradeDaily.csv')
# 每日持仓
#print('每日持仓:')
stockHoldDailyDf = pd.DataFrame(stockHoldDailyList)
stockHoldDailyDf.sort_values(by=['tradingDate'], ascending=True)
# stockHoldDailyDf.to_csv(StockConst.root + '\export\stockHoldDaily.csv')
#每日统计(收益，净值，买入数，卖出数，持有数)
stockStatDailyDf = stockStatDailyDf.dropna(how='all')
result += '--' * 70 + '\n'
result += '每日统计:' + '\n'
result += stockStatDailyDf.to_string() + '\n'
# stockStatDailyDf.to_csv(StockConst.root + '\export\stockStatDaily.csv')
# 每年统计
result += '--' * 70 + '\n'
yearDf = StockYearService.main(stockStatDailyDf)
result += '每年统计:' + '\n'
result += yearDf.to_string() + '\n'
# 最大回撤
result += '--' * 70 + '\n'
maxdrop = StockMaxDropNewService.get_max_drop(stockStatDailyDf)
result += '最大回撤:' + '\n'
result += SetUtil.dict_to_string(maxdrop) + '\n'
#每年的最大回撤
maxdropDf = StockMaxDropNewService.get_max_drop_for_each_year(
stockStatDailyDf)
maxdropDf.sort_values(by=["year"])
result += '每年的最大回撤:' + '\n'
result += maxdropDf.to_string() + '\n'
#夏普比率
result += '--' * 70 + '\n'
sharpRatio = SharpRatioNewService.get_sharp_ratio(stockStatDailyDf)
result += '夏普比率:' + '\n'
result += str(sharpRatio) + '\n'
#每年的夏普比率
sharpRatioDf = SharpRatioNewService.get_sharp_ratio_for_each_year(
stockStatDailyDf)
sharpRatioDf.sort_values(by=["year"])
result += '每年的夏普比率:' + '\n'
result += sharpRatioDf.to_string() + '\n'
end_stat = time.clock()
print("统计花费时间：%f s" % (end_stat - start_stat))
# result += '回测完成'
end = time.clock()
print("main花费时间：%f s" % (end - start))
# 净值曲线图
stockStatDailyDf['netValue'].plot()
plt.show()
resultDic = {'stockStatDailyDf': stockStatDailyDf, 'result': result}
print(result)
return resultDic
def mergeSignalWithZZ800():
signalDataDf = SourceDataDao.load_signal_data('')
indexDf = SourceDataDao.load_h5(StockConst.ROOT + StockConst.ZZ800H5)
# signalDataDf = selectSignalByDate(signalDataDf,'2005-01-01','2005-04-08')
# hs300Df = hs300Df.loc[(hs300Df['NatureDay'] == '2005-04-08')]
mergeSignalWithIndexMain(signalDataDf, indexDf, StockConst.SIGNAL_DATA_BASEDON_ZZ800)
def test_select_stock_func3(self):
signalData = SourceDataDao.load_signal_data('')
techList = [StockConst.MOM, 'Momp']
BackTestMainNew.select_stock_func3(signalData, techList)
