def recordResampleTickTransactionDataFromSqlServer(self,code,startDate,endDate):
#获取股票日线数据，筛选出非停牌的日期
daily=KLineDataProcess('daily')
dailyData=daily.getDataByDate(code,startDate,endDate)
if dailyData.empty==True:
logger.warning(f'There no daily data of {code} from {startDate} to {endDate}!')
return
dailyData=dailyData[dailyData['status']!='停牌']
tradedays=list(dailyData['date'])
for date in tradedays:
path=os.path.join(self.filePath,code.replace('.','_'))
file=os.path.join(path,date+'.h5')
HDF5Utility.pathCreate(path)
exists=os.path.isfile(file)
if exists==False:
logger.info(f'get TickTransaction data of {code} in {date} from source!')
mydata=self.__getResampleTickTransactionDataFromSqlServer(code,date)
if mydata.empty==False:
with pd.HDFStore(file,'a',complib='blosc:zstd',append=True,complevel=9) as store:
store.append('data',mydata,append=False,format="table",data_columns=mydata.columns)
else:
logger.warning(f'there is no data of {code} in {date} from source!')
else:
logger.info(f'TickTransaction data of {code} in {date} is exists!')
pass
pass
def getTickDataAndFactorsByDateFromLocalFile(self,
code,
date,
factors=TICKFACTORSUSED):
myfactor = factorBase()
mydata = pd.DataFrame()
for item in factors:
factor = item['factor']
data = myfactor.getDataFromLocalFile(code, date, factor)
if mydata.shape[0] == 0: #如果还没有取出来数据
mydata = data.copy()
pass
elif data.shape[0] != 0:
mydata = pd.merge(mydata,
data,
how='left',
left_index=True,
right_index=True)
pass
tick = TickDataProcess()
tickData = tick.getDataByDateFromLocalFile(code, date)
mydata = pd.merge(mydata,
tickData,
how='left',
left_index=True,
right_index=True)
if mydata.shape[0] == 0:
return mydata
#dailyFactor=['closeStd','index','marketValue','industry']
dailyRepo = dailyFactorsProcess()
dailyData = dailyRepo.getSingleStockDailyFactors(code, date, date)
for col in dailyData.columns:
if col not in ['date', 'code', 'return']:
mydata[col] = dailyData[col].iloc[0]
dailyKLineRepo = KLineDataProcess('daily')
dailyKLineData = dailyKLineRepo.getDataByDate(code, date, date)
mydata['preClose'] = dailyKLineData['preClose'].iloc[0]
mydata['increaseToday'] = mydata['midPrice'] / mydata['preClose'] - 1
ceiling = mydata[(mydata['B1'] == 0) | (mydata['S1'] == 0)]
if ceiling.shape[0] > 0:
ceilingTime = ceiling['tick'].iloc[0]
mydata = mydata[mydata['tick'] < ceilingTime]
pass
return mydata
def recordResampleTickShotDataToInfluxdbFromSqlServer(
self, code, startDate, endDate):
#获取股票日线数据，筛选出非停牌的日期
daily = KLineDataProcess('daily')
dailyData = daily.getDataByDate(code, startDate, endDate)
if dailyData.empty == True:
#logger.warning(f'There no daily data of {code} from {startDate} to {endDate}!')
return
dailyData = dailyData[dailyData['status'] != '停牌']
tradedays = list(dailyData['date'])
for date in tradedays:
#logger.info(f'get tickshot data of {code} in {date} from source!')
mydata = self.__getResampleTickShotDataFromSqlServer(code, date)
if mydata.shape[0] > 0:
mydata = mydata.fillna(0)
InfluxdbUtility.saveDataFrameDataToInfluxdb(
mydata, INFLUXDBTICKTICKDATABASE, code, {})
pass
def getStockData(self):
myindex=IndexComponentDataProcess(True)
index500=myindex.getCSI500DataByDate(20190404,20190404)
index300=myindex.getHS300DataByDate(20190404,20190404)
index50=myindex.getSSE50DataByDate(20190404,20190404)
stockCodes=list(pd.concat([index500,index300,index50],ignore_index=True)['code'].drop_duplicates())
myMinute=KLineDataProcess('minute',True)
myDaily=KLineDataProcess('daily',True)
num=0
for code in stockCodes:
print(datetime.datetime.now())
num=num+1
myMinute.getDataByDate(code,20070101,20190327)
myDaily.getDataByDate(code,20070101,20190327)
print("{0}({1} of 800) complete!".format(code,num))
########################################################################
def reverse_singleCode(self, code, startDate, endDate):
days = list(TradedayDataProcess().getTradedays(startDate, endDate))
#factors=['closeStd','index','marketValue','industry']
dailyRepo = dailyFactorsProcess()
dailyFactor = dailyRepo.getSingleStockDailyFactors(
code, startDate, endDate)
dailyKLine = KLineDataProcess('daily')
dailyData = dailyKLine.getDataByDate(code, startDate, endDate)
if dailyData.empty == True:
logger.error(
f'there is no data of {code} from {startDate} to {endDate}')
return pd.DataFrame()
tick = TickDataProcess()
tickfactor = tickFactorsProcess()
mydata = []
position = 0
profit = 0
myStatusList = []
myTradeList = []
myStatus = {}
myTrade = {}
positionYesterday = 0
select1 = []
select2 = []
selectall = []
for today in days:
#logger.info(f'{code} in {today} start!')
todayInfo = dailyFactor[dailyFactor['date'] == today]
if todayInfo.empty == True:
logger.error(
f'there is no factor data of {code} in date {today}')
continue
pass
todayKLine = dailyData[dailyData['date'] == today]
if todayKLine.empty == True:
logger.error(
f'there is no KLine data of {code} in date {today}')
continue
pass
myStatus['date'] = today
myStatus['closeStd20'] = todayInfo['closeStd20'].iloc[0]
myStatus['weight50'] = todayInfo['weight50'].iloc[0]
myStatus['weight300'] = todayInfo['weight300'].iloc[0]
myStatus['weight500'] = todayInfo['weight500'].iloc[0]
myStatus['ts_closeStd20'] = todayInfo['closeStd20'].iloc[0]
myStatus['adjFactor'] = todayKLine['adjFactor'].iloc[0]
myStatus['preClose'] = todayKLine['preClose'].iloc[0]
positionNow = positionYesterday
if (todayInfo.empty
== False) & (todayKLine['status'].iloc[0] != '停牌'):
#tickData=tick.getDataByDateFromLocalFile(code,today)
tickData = tick.getTickShotDataFromInfluxdbServer(code, today)
#factors=tickfactor.getDataByDateFromLocalFile(code,today,'buySellForce')
factors = tickfactor.getFactorsUsedByDateFromLocalFile(
code, today)
all = pd.merge(tickData,
factors,
how='left',
left_index=True,
right_index=True)
all['closeStd20'] = todayInfo['closeStd20'].iloc[0]
all['ts_closeStd20'] = todayInfo['ts_closeStd20'].iloc[0]
all['adjFactor'] = todayKLine['adjFactor'].iloc[0]
all['preClose'] = todayKLine['preClose'].iloc[0]
all['increaseToday'] = all['midPrice'] / all['preClose'] - 1
all['midIncreasePrevious5m'] = all['midPrice'] / all[
'midPrice'].shift(60) - 1
#select0=all[all['increaseToday']>2*all['closeStd20']]
selectall.append(all)
select0 = all[all['midIncreasePrevious5m'] > 0.5 *
all['closeStd20']]
select1.append(select0)
select0 = all[all['midIncreasePrevious5m'] < -0.5 *
all['closeStd20']]
select2.append(select0)
pass
else:
logger.warning(f'There is no data of {code} in {today}')
mycolumns = [
'midIncreaseNext1m', 'midIncreaseNext5m', 'ts_buyForceIncrease',
'ts_sellForceIncrease', 'ts_buySellVolumeRatio5',
'midIncreasePrevious3m', 'differenceHighLow',
'ts_differenceHighLow', 'differenceMidVwap',
'ts_differenceMidVwap', 'midStd60', 'ts_midStd60'
]
select1 = pd.concat(select1)
select2 = pd.concat(select2)
selectall = pd.concat(selectall)
selectall = selectall[mycolumns]
s = select1[(select1['ts_sellForceIncrease'] > 0.6)
& (select1['ts_buySellVolumeRatio5'] > 0.6) &
(select1['increaseToday'] > select1['closeStd20'])]
print(s['midIncreaseNext5m'].mean())
ss = select2[(select2['ts_buySellVolumeRatio5'] > 0.6)
& (select2['increaseToday'] < -select2['closeStd20'])]
print(ss['midIncreaseNext5m'].mean())
return mydata
########################################################################
class stockReverseMovement(object):
"""股票异动,专注股票大涨之后的回调"""
#----------------------------------------------------------------------
def __init__(self):
self.__myMinute = KLineDataProcess('minute', True)
self.__myDaily = KLineDataProcess('daily', True)
self.__localFileStr = LocalFileAddress + "\\intermediateResult\\ceilingFeature.h5"
self.__allMinute = pd.DataFrame()
pass
#----------------------------------------------------------------------
def __getStockList(self):
myindex = IndexComponentDataProcess()
index500 = myindex.getCSI500DataByDate(20190404, 20190404)
index300 = myindex.getHS300DataByDate(20190404, 20190404)
index50 = myindex.getSSE50DataByDate(20190404, 20190404)
stockCodes = list(
pd.concat([index500, index300, index50],
ignore_index=True)['code'].drop_duplicates())
return stockCodes
#----------------------------------------------------------------------
def __dataPrepared(self):
mylist = self.__getStockList()
self.__minuteData = {}
self.__dailyData = {}
num = 0
for code in mylist:
num = num + 1
print("{0}({1} of 50) start!".format(code, num))
print(datetime.datetime.now())
self.__dataSelectOneByOne(code)
d = self.__dailyData[code]
m = self.__minuteData[code]
d['ceiling'] = 0
d['ceilingYesterday'] = 0
d['ceilingYesterday2'] = 0
d['ceilingIn5Days'] = 0
d.loc[(d['close'] == round(d['preClose'] * 1.1, 2)), 'ceiling'] = 1
d.loc[(d['ceiling'].shift(1) == 1), 'ceilingYesterday'] = 1
d.loc[((d['ceiling'].shift(1) == 1) &
(d['ceiling'].shift(2) == 1)), 'ceilingYesterday2'] = 1
d['ceilingIn5Days'] = d['ceilingYesterday'].rolling(5).sum()
#print(d[d['ceilingYesterday']>0][['date','close','preClose','ceilingIn5Days']])
m.loc[(m['date'] == m['date'].shift(5)),
'increase5m'] = (m['open'] / m['open'].shift(5) - 1)
m.loc[(m['date'] == m['date'].shift(1)),
'increase1m'] = (m['open'] / m['open'].shift(1) - 1)
d = d.set_index('date')
dailyInfo = d.loc[m['date'], [
'preClose', 'ceilingYesterday', 'ceilingYesterday2',
'ceilingIn5Days'
]]
dailyInfo.index = m.index
m[[
'yesterdayClose', 'ceilingYesterday', 'ceilingYesterday2',
'ceilingIn5Days'
]] = dailyInfo
m['increaseInDay'] = (m['open'] / m['yesterdayClose'] - 1)
m['ceiling'] = 0
m['return5m'] = 0
m['return10m'] = 0
m.loc[(m['low'] == round(m['yesterdayClose'] * 1.1, 2)),
'ceiling'] = 1
m.loc[(m['open'].shift(-5) / m['open'] - 1) > 0.01, 'return5m'] = 1
m.loc[(m['open'].shift(-5) / m['open'] - 1) < -0.01,
'return5m'] = -1
m.loc[(m['open'].shift(-10) / m['open'] - 1) > 0.01,
'return10m'] = 1
m.loc[(m['open'].shift(-10) / m['open'] - 1) < -0.01,
'return10m'] = -1
m['ceilingInNext5m'] = m['ceiling'].shift(-5).rolling(5).max()
m['maxLossInNext5m'] = round(
(m['low'].shift(-5).rolling(5).min() - m['open']) / m['open'] -
1, 2)
m['ceilingInNext10m'] = m['ceiling'].shift(-10).rolling(10).max()
m['maxLossInNext10m'] = round(
(m['low'].shift(-10).rolling(10).min() - m['open']) / m['open']
- 1, 2)
m[m['minute'] > '1450']['ceilingInNext5m', 'maxLossInNext5m',
'ceilingInNext10m',
'maxLossInNext10m'] = None
mselect = m[(m['increaseInDay'] > 0.07)
& (m['increaseInDay'] < 0.08)]
mselect = mselect.dropna(axis=0, how='any')
self.__allMinute = self.__allMinute.append(mselect)
pass
pass
#----------------------------------------------------------------------
def __dataSelectOneByOne(self, code):
m = self.__myMinute.getDataByDate(code, self.startDate, self.endDate)
d = self.__myDaily.getDataByDate(code, self.startDate, self.endDate)
days = list(d[d['high'] > d['preClose'] * 1.05]['date'])
m = m.loc[m['date'].isin(days), :]
self.__dailyData[code] = d
self.__minuteData[code] = m
pass
#----------------------------------------------------------------------
def reverse(self, startDate, endDate):
self.startDate = startDate
self.endDate = endDate
self.tradeDays = TradedayDataProcess.getTradedays(startDate, endDate)
self.__dataPrepared()
store = pd.HDFStore(self.__localFileStr, 'a')
store.append('ceiling',
self.__allMinute,
append=False,
format="table",
data_columns=[
'code', 'date', 'minute', 'open', 'high', 'low',
'close', 'volume', 'amount', 'increase5m',
'increase1m', 'yesterdayClose', 'ceilingYesterday',
'ceilingYesterday2', 'ceilingIn5Days',
'increaseInDay', 'ceiling', 'ceilingInNext5m',
'ceilingInNext10m', 'maxLossInNext5m',
'maxLossInNext10m'
])
store.close()
def __init__(self):
self.__myMinute = KLineDataProcess('minute', True)
self.__myDaily = KLineDataProcess('daily', True)
self.__localFileStr = LocalFileAddress + "\\intermediateResult\\ceilingFeature.h5"
self.__allMinute = pd.DataFrame()
pass
def reverse_singleCode(self,
code,
startDate,
endDate,
parameters=[300, 100000000, 1.5]):
myindex = parameters[0]
totalCash = parameters[1]
std1 = parameters[2]
days = list(TradedayDataProcess().getTradedays(startDate, endDate))
#factors=['closeStd','index','marketValue','industry']
dailyRepo = dailyFactorsProcess()
dailyFactor = dailyRepo.getSingleStockDailyFactors(
code, startDate, endDate)
dailyKLine = KLineDataProcess('daily')
dailyData = dailyKLine.getDataByDate(code, startDate, endDate)
if dailyData.empty == True:
logger.error(
f'there is no data of {code} from {startDate} to {endDate}')
return pd.DataFrame()
tick = TickDataProcess()
mydata = []
position = 0
profit = 0
myStatusList = []
myTradeList = []
myStatus = {}
myTrade = {}
positionYesterday = 0
for today in days:
#logger.info(f'{code} in {today} start!')
todayInfo = dailyFactor[dailyFactor['date'] == today]
if todayInfo.empty == True:
logger.error(
f'there is no factor data of {code} in date {today}')
continue
pass
todayKLine = dailyData[dailyData['date'] == today]
if todayKLine.empty == True:
logger.error(
f'there is no KLine data of {code} in date {today}')
continue
pass
myStatus['date'] = today
myStatus['closeStd20'] = todayInfo['closeStd20'].iloc[0]
myStatus['weight50'] = todayInfo['weight50'].iloc[0]
myStatus['weight300'] = todayInfo['weight300'].iloc[0]
myStatus['weight500'] = todayInfo['weight500'].iloc[0]
myStatus['ts_closeStd20'] = todayInfo['closeStd20'].iloc[0]
myStatus['adjFactor'] = todayKLine['adjFactor'].iloc[0]
myStatus['preClose'] = todayKLine['preClose'].iloc[0]
positionNow = positionYesterday
if (todayInfo.empty
== False) & (todayKLine['status'].iloc[0] != '停牌'):
if myindex == 300:
maxPosition = myStatus[
'weight300'] * totalCash * 0.01 / myStatus['preClose']
elif myindex == 500:
maxPosition = myStatus[
'weight300'] * totalCash * 0.01 / myStatus['preClose']
elif myindex == 50:
maxPosition = myStatus[
'weight50'] * totalCash * 0.01 / myStatus['preClose']
else:
maxPosition = myStatus['totalCash'] * 0.001 / myStatus[
'preClose']
maxPosition = round(maxPosition, -2)
if maxPosition == 0:
continue
tickData = tick.getDataByDateFromLocalFile(code, today)
#['code' ,'date','tick' ,'lastPrice','S1','S2','S3','S4','S5','S6','S7','S8','S9','S10','B1','B2','B3','B4','B5','B6','B7','B8','B9','B10','SV1','SV2','SV3','SV4','SV5','SV6','SV7','SV8','SV9','SV10','BV1','BV2','BV3','BV4','BV5','BV6','BV7','BV8','BV9','BV10','volume' ,'amount','volumeIncrease','amountIncrease']
tickList = tickData.as_matrix()
for i in range(0, len(tickList) - 60):
now = tickList[i]
midPrice = (now[4] + now[14]) / 2
lastPrice = now[3]
tickShot = now[4:43]
upCeiling = False
downCeiling = False
if now[24] == 0:
upCeiling = True
pass
if now[34] == 0:
downCeiling = True
pass
mytime = datetime.datetime.strptime(
now[1] + now[2], '%Y%m%d%H%M%S%f')
increaseToday = tickList[i][3] / myStatus['preClose'] - 1
if (i >= 100) & (tickList[i - 100][3] > 0):
increase5m = tickList[i][3] / tickList[i - 100][3] - 1
else:
increase5m = np.nan
if ((positionNow == 0) & (positionYesterday == 0) &
(i <= 4500) & (i >= 100) &
(increaseToday > std1 * myStatus['closeStd20']) &
(maxPosition > 0) & (downCeiling == False)):
#开空头
[price, deltaPosition,
amount] = TradeUtility.sellByTickShotData(
tickShot, maxPosition, 0.001)
positionNow = -deltaPosition
myTrade['date'] = today
myTrade['opentime'] = mytime
myTrade['position'] = positionNow
myTrade['open'] = price
myTrade['openAdj'] = myStatus['adjFactor']
myTrade['increase5m'] = increase5m
myTrade['increaseToday'] = increaseToday
openIndex = i
maxPosition = maxPosition - deltaPosition
pass
elif ((positionNow == 0) & (positionYesterday == 0) &
(i <= 4500) & (i >= 100) &
(increaseToday < -std1 * myStatus['closeStd20']) &
(maxPosition > 0) & (upCeiling == False)):
#开多头
[price, deltaPosition,
amount] = TradeUtility.buyByTickShotData(
tickShot, maxPosition, 0.001)
positionNow = deltaPosition
myTrade['date'] = today
myTrade['opentime'] = mytime
myTrade['position'] = positionNow
myTrade['open'] = price
myTrade['openAdj'] = myStatus['adjFactor']
myTrade['increase5m'] = increase5m
myTrade['increaseToday'] = increaseToday
maxPosition = maxPosition - deltaPosition
openIndex = i
pass
elif (positionNow > 0):
profit = (lastPrice * myStatus['adjFactor'] -
myTrade['open'] * myTrade['openAdj']) / (
myTrade['open'] * myTrade['openAdj'])
if (((profit < 0) & (i >= openIndex + 400)) |
(i >= 4700) | (i >= openIndex + 1200) |
(positionYesterday != 0)):
if downCeiling == False: #未跌停
#平多头,记录一笔交易
[price, deltaPosition,
amount] = TradeUtility.sellByTickShotData(
tickShot, positionNow, 0.1)
myTrade['closetime'] = mytime
myTrade['close'] = price
myTrade['closeAdj'] = myStatus['adjFactor']
positionNow = 0
positionYesterday = 0
openIndex = 0
tradeCopy = copy.deepcopy(myTrade)
myTradeList.append(tradeCopy)
myTrade = {}
pass
elif (positionNow < 0):
profit = (-lastPrice * myStatus['adjFactor'] +
myTrade['open'] * myTrade['openAdj']) / (
myTrade['open'] * myTrade['openAdj'])
if (((profit < 0) & (i >= openIndex + 400)) |
(i >= 4700) | (i >= openIndex + 1200) |
(positionYesterday != 0)):
if upCeiling == False:
#平空头，记录一笔交易
[price, deltaPosition,
amount] = TradeUtility.buyByTickShotData(
tickShot, -positionNow, 0.1)
myTrade['closetime'] = mytime
myTrade['close'] = price
myTrade['closeAdj'] = myStatus['adjFactor']
positionNow = 0
positionYesterday = 0
openIndex = 0
tradeCopy = copy.deepcopy(myTrade)
myTradeList.append(tradeCopy)
myTrade = {}
pass
if positionNow != 0:
logger.info(
f'{code} of {today} can not close the position!')
positionYesterday = positionNow
pass
else:
logger.warning(f'There is no data of {code} in {today}')
m = pd.DataFrame(data=myTradeList,
columns=[
'date', 'opentime', 'position', 'open', 'openAdj',
'increase5m', 'increaseToday', 'closetime',
'close', 'closeAdj'
])
m['profit'] = m['position'] * (m['close'] * m['closeAdj'] -
m['open'] * m['openAdj']) / m['openAdj']
m['fee'] = m['position'].abs() * m[['open', 'close'
]].max(axis=1) * 0.0012
m['netProfit'] = m['profit'] - m['fee']
m['code'] = code
mycolumns = [
'date', 'closeStd20', 'ts_closeStd20', 'industry', 'industryName'
]
dailyFactor = dailyFactor[mycolumns]
mydata = pd.merge(m,
dailyFactor,
how='left',
left_on='date',
right_on='date')
#mydata=mydata[((mydata['increase5m']>mydata['closeStd20']) & (mydata['position']<0)) |((mydata['increase5m']<-mydata['closeStd20']) & (mydata['position']>0))]
'''
print(mydata)
print(mydata['profit'].sum())
print(mydata['fee'].sum())
print(mydata['netProfit'])
'''
return mydata
########################################################################
def dataPrepared(self, stockCodes, startDate, endDate, recordFilePath):
warnings.filterwarnings('ignore')
mytradedays = TradedayDataProcess.getTradedays(startDate, endDate)
mylist = stockCodes
myMinute = KLineDataProcess('minute', False)
num = 0
store = pd.HDFStore(self.__factorsAddress, 'a')
allDailyData = store.select(
self.__key,
where=['date>="%s" and date<="%s"' % (startDate, endDate)])
store.close()
store = pd.HDFStore(recordFilePath, 'a')
oldKeys = store.keys()
for code in mylist:
num = num + 1
print("{0}({1} of {2}) start!".format(str(code), num, len(mylist)))
#print(datetime.datetime.now())
if ('/' + code) in oldKeys:
continue
pass
m = myMinute.getDataByDate(code, startDate, endDate)
if len(m) == 0:
continue
m = m[m['date'].isin(mytradedays)]
m['vwap'] = m['amount'] / m['volume']
m.loc[(m['date'] == m['date'].shift(5)),
'increase5m'] = (m['open'] / m['open'].shift(5) - 1)
m.loc[(m['date'] == m['date'].shift(1)),
'increase1m'] = (m['open'] / m['open'].shift(1) - 1)
d = allDailyData.xs(code, level='code')
dailyInfo = d.loc[m['date']]
deleteColumns = [
'code', 'date', 'open', 'high', 'low', 'close', 'volume',
'amount', 'change', 'pctChange', 'vwap'
]
mycolumns = [
col for col in dailyInfo.columns if col not in deleteColumns
]
dailyInfo = dailyInfo[mycolumns]
dailyInfo.rename(columns={'preClose': 'yesterdayClose'},
inplace=True)
dailyInfo.index = m.index
m = pd.concat([m, dailyInfo], axis=1)
m['increaseInDay'] = (m['open'] / m['yesterdayClose'] - 1)
m = m[m['status'] != 'N']
m['canBuy'] = 0
m['canSell'] = 0
m['canBuyPrice'] = None
m['canSellPrice'] = None
m.loc[((m['open'] < round(1.097 * m['yesterdayClose'], 2)) &
(m['open'] > 0)), 'canBuy'] = 1
m.loc[((m['open'] > round(0.903 * m['yesterdayClose'], 2)) &
(m['open'] > 0)), 'canSell'] = 1
m.loc[m['canBuy'] == 1, 'canBuyPrice'] = m.loc[m['canBuy'] == 1,
'open']
m.loc[m['canBuy'] == 1, 'canBuyPriceAdj'] = m.loc[m['canBuy'] == 1,
'adjFactor']
m['canBuyPrice'] = m['canBuyPrice'].fillna(method='bfill')
m['canBuyPriceAdj'] = m['canBuyPriceAdj'].fillna(method='bfill')
m.loc[m['canSell'] == 1, 'canSellPrice'] = m.loc[m['canSell'] == 1,
'open']
m.loc[m['canSell'] == 1,
'canSellPriceAdj'] = m.loc[m['canSell'] == 1, 'adjFactor']
m['canSellPrice'] = m['canSellPrice'].fillna(method='bfill')
m['canSellPriceAdj'] = m['canSellPriceAdj'].fillna(method='bfill')
m['timeStamp'] = m['date'] + m['tick']
#日内分钟信息
#成交量在前20分钟的分位数
m['ts_rank_volume'] = m['volume'].shift(1).rolling(
20, min_periods=15).apply(
(lambda x: pd.Series(x).rank().iloc[-1] / len(x)),
raw=True)
#每分钟收益
m['minuteReturn'] = (m['close'] -
m['close'].shift(1)) / m['close'].shift(1)
#收益标准差
m['minuteStd20'] = m['minuteReturn'].shift(1).rolling(
20, min_periods=17).std()
#收益标准差分位数
m['ts_rank_minuteStd20'] = m['minuteStd20'].rolling(
20, min_periods=15).apply(
(lambda x: pd.Series(x).rank().iloc[-1] / len(x)),
raw=True)
mselect = m.set_index(['timeStamp', 'code'])
store.append(code, mselect, append=False, format="table")
pass
store.close()
pass
def updateSingleStockMinuteKLines(self, code, startDate, endDate):
stock = KLineDataProcess('minute', True)
stock.getDataByDate(code, startDate, endDate)
pass
def getStockDailyFeatureData(self, stockCodes, startDate, endDate):
myDaily = KLineDataProcess('daily')
myDailyDerivative = KLineDataProcess('dailyDerivative')
myindex = IndexComponentDataProcess()
exists = os.path.isfile(self.localFileStr)
if exists == True:
store = pd.HDFStore(self.localFileStr, 'a')
allData = store.select(self.key)
allData.reset_index(inplace=True, drop=False)
existsCodes = list(allData['code'].drop_duplicates())
store.close()
else:
allData = pd.DataFrame()
existsCodes = []
num = 0
for code in stockCodes:
num = num + 1
print("{0}({1} of {2}) start!".format(code, num, len(stockCodes)))
if code in existsCodes:
stockNow = allData[allData['code'] == code]
latestDate = stockNow['date'].max()
if endDate <= latestDate:
continue
pass
startNow = max(startDate,
TradedayDataProcess.getNextTradeday(latestDate))
if startNow > endDate:
continue
pass
mydata = myDaily.getDataByDate(code, startNow, endDate)
mydata.set_index('date', drop=True, inplace=True)
myindustry = IndustryClassification.getIndustryByCode(
code, startNow, endDate)
mydata['industry'] = myindustry['industry']
mydata['industryName'] = myindustry['name']
myIndexBelongs50 = myindex.getStockBelongs(code, SSE50, startNow,
endDate)
myIndexBelongs300 = myindex.getStockBelongs(
code, HS300, startNow, endDate)
myIndexBelongs500 = myindex.getStockBelongs(
code, CSI500, startNow, endDate)
mydata['is50'] = myIndexBelongs50['exists']
mydata['is300'] = myIndexBelongs300['exists']
mydata['is500'] = myIndexBelongs500['exists']
mydataDerivative = myDailyDerivative.getDataByDate(
code, startNow, endDate)
mydataDerivative.set_index('date', inplace=True)
mydata['freeShares'] = mydataDerivative['freeShares']
mydata['freeMarketValue'] = mydataDerivative['freeMarketValue']
mydata['return'] = (mydata['close'] -
mydata['preClose']) / mydata['preClose']
#昨日计算出的标准差给今日用，防止用到未来数据
mydata['closeStd20'] = mydata['return'].shift(1).rolling(
20, min_periods=17).std()
mydata['ts_rank_closeStd20'] = mydata['closeStd20'].rolling(
50, min_periods=20).apply(
(lambda x: pd.Series(x).rank().iloc[-1] / len(x)),
raw=True)
mydata.reset_index(inplace=True)
allData = allData.append(mydata)
pass
allData = allData.set_index(['date', 'code'])
store = pd.HDFStore(self.localFileStr, 'a')
store.append(self.key, allData, append=True, format="table")
store.close()
unstack = allData.unstack()
rankMv = unstack['freeMarketValue'].rank(axis=1)
mvMax = rankMv.max(axis=1)
rankMv = rankMv.iloc[:, :].div(mvMax, axis=0)
#做rankMarketValue的操作
allData['rankMarketValue'] = rankMv.stack()
store = pd.HDFStore(self.localFileStrWithRank, 'a')
store.append(self.keyWithRank, allData, append=False, format="table")
store.close()
def updateStockDailyFactors(self, codeList, factorList):
for code in codeList:
logger.info(f'{code} factor update start!')
#获取数据
IPOInfo = StockIPOInfoProcess.getStockIPOInfoByCode(code)
listDate = IPOInfo['listDate'].iloc[0]
delistDate = IPOInfo['delistDate'].iloc[0]
#mydata=self.__updateStockDailyBasicData(code)
for factor in factorList:
fileName = code.replace('.', '_') + ".h5"
factorFilePath = os.path.join(self.localFileFolder, factor,
fileName)
path = os.path.join(self.localFileFolder, factor)
HDF5Utility.pathCreate(path)
exists = os.path.isfile(factorFilePath)
if exists == False:
startDate = listDate
endDate = self.endDate
lastDate = EMPTY_STRING
pass
else:
with pd.HDFStore(path=factorFilePath,
mode='r',
complib='blosc:zstd',
append=True,
complevel=9) as store:
existsDate = store['date']
lastDate = existsDate.max()
startDate = lastDate
endDate = self.endDate
if startDate >= endDate:
continue
pass
startDate = TradedayDataProcess.getPreviousTradeday(
startDate, 100)
if startDate < listDate:
startDate = listDate
pass
if endDate > delistDate:
endDate = delistDate
pass
#marketData=self.__updateStockDailyBasicData(code,startDate,endDate)
dailyData = KLineDataProcess('daily')
marketData = dailyData.getDataByDate(code, startDate, endDate)
endDate = marketData['date'].max() #因子日期统一计算到日线数据最后一天
mydata = marketData
#tradedays=TradedayDataProcess.getTradedays(startDate,endDate)
#mydata=pd.DataFrame(tradedays,columns=['date'])
#mydata=pd.merge(mydata,marketData,how='left',left_on='date',right_on='date')
if factor == 'closeStd':
myReturnFun = ComputeUtility.computeReturn
myStdFun = ComputeUtility.computeStandardDeviation
myTSRank = ComputeUtility.computeTimeSeriesRank
mydata = self.__computeStockDailyFactor(
mydata, myReturnFun, ['close', 'preClose', 'return'])
mydata.loc[mydata['status'] == '停牌', 'return'] = np.nan
mydata.loc[mydata['date'] == listDate, 'return'] = np.nan
mydata = self.__computeStockDailyFactor(
mydata, myStdFun, ['return', 'closeStd20', 20, 0.8])
mydata = self.__computeStockDailyFactor(
mydata, myTSRank,
['closeStd20', 'ts_closeStd20', 50, 0.4])
mycolumns = [
'date', 'return', 'closeStd20', 'ts_closeStd20'
]
mydata = mydata[mycolumns]
elif factor == 'industry':
mydata.set_index('date', drop=True, inplace=True)
myindustry = IndustryClassification.getIndustryByCode(
code, startDate, endDate)
mydata['industry'] = myindustry['industry']
mydata['industryName'] = myindustry['name']
mydata.reset_index(drop=False, inplace=True)
mycolumns = ['date', 'industry', 'industryName']
mydata = mydata[mycolumns]
pass
elif factor == 'index':
myindex = IndexComponentDataProcess()
myIndexBelongs50 = myindex.getStockBelongs(
code, SSE50, startDate, endDate)
myIndexBelongs300 = myindex.getStockBelongs(
code, HS300, startDate, endDate)
myIndexBelongs500 = myindex.getStockBelongs(
code, CSI500, startDate, endDate)
myStockWeightOf50 = IndexComponentDataProcess.getStockPropertyInIndex(
code, SSE50, startDate, endDate)
myStockWeightOf300 = IndexComponentDataProcess.getStockPropertyInIndex(
code, HS300, startDate, endDate)
myStockWeightOf500 = IndexComponentDataProcess.getStockPropertyInIndex(
code, CSI500, startDate, endDate)
mydata.set_index('date', drop=True, inplace=True)
mydata['is50'] = myIndexBelongs50['exists']
mydata['is300'] = myIndexBelongs300['exists']
mydata['is500'] = myIndexBelongs500['exists']
mydata['weight50'] = myStockWeightOf50['weight']
mydata['weight300'] = myStockWeightOf300['weight']
mydata['weight500'] = myStockWeightOf500['weight']
mydata.reset_index(drop=False, inplace=True)
mycolumns = [
'date', 'is50', 'is300', 'is500', 'weight50',
'weight300', 'weight500'
]
mydata = mydata[mycolumns]
#mydata[['is50','is300','is500']]=mydata[['is50','is300','is500']].astype('int64')
pass
elif factor == 'marketValue':
mydata.set_index('date', drop=True, inplace=True)
myDailyDerivative = KLineDataProcess('dailyDerivative')
mydataDerivative = myDailyDerivative.getDataByDate(
code, startDate, endDate)
mydataDerivative.set_index('date', inplace=True)
mydata['freeShares'] = mydataDerivative['freeShares']
mydata['freeMarketValue'] = mydataDerivative[
'freeMarketValue']
mydata.reset_index(drop=False, inplace=True)
mycolumns = ['date', 'freeShares', 'freeMarketValue']
mydata = mydata[mycolumns]
try:
with pd.HDFStore(path=factorFilePath,
mode='a',
complib='blosc:zstd',
append=True,
complevel=9) as store:
if lastDate == EMPTY_STRING:
mydate = mydata['date']
else:
mydata = mydata[(mydata['date'] > lastDate)]
mydate = mydata['date']
if mydata.empty == False:
store.append('date',
mydate,
append=True,
format="table",
data_columns=['date'],
complevel=9)
store.append('factors',
mydata,
append=True,
format="table",
data_columns=mydata.columns,
complevel=9)
except Exception as excp:
logger.error(f'{code} in {date} error! {excp}')
logger.error(traceback.format_exc())
pass
pass
def updateMinuteIndexKLines(self, code, startDate, endDate):
index = KLineDataProcess('minuteIndex', True)
index.getDataByDate(code, startDate, endDate)
pass
def updateMultipleStocksDailyDerivatives(self, codeList, startDate,
endDate):
dailyStock = KLineDataProcess('dailyDerivative', True)
dailyStock.parallelizationUpdateDataByDate(codeList, startDate,
endDate)
pass
def updateSingleStockDailyDerivatives(self, code, startDate, endDate):
stock = KLineDataProcess('dailyDerivative', True)
stock.getDataByDate(code, startDate, endDate)
pass
def updateMultipleStocksMinuteKLines(self, codeList, startDate, endDate):
stock = KLineDataProcess('minute', True)
stock.parallelizationUpdateDataByDate(codeList, startDate, endDate)
pass
def saveAllFactorsByCodeAndDate(self, code, date):
mypath = os.path.join(self.path, str(code).replace('.', '_'))
HDF5Utility.pathCreate(mypath)
fileName = os.path.join(mypath, str(date) + '.h5')
exists = HDF5Utility.fileCheck(fileName)
if exists == True: #如果文件已将存在，直接返回
return
myfactor = factorBase()
mydata = pd.DataFrame()
factors = self.factorsUsed
#获取tick因子数据
mydata = self.getFactorsUsedByDateFromLocalFile(code, date, factors)
#获取tick行情数据
tick = TickDataProcess()
tickData = tick.getDataByDateFromLocalFile(code, date)
mydata = pd.merge(mydata,
tickData,
how='left',
left_index=True,
right_index=True)
if mydata.shape[0] == 0:
return
#获取日线数据
dailyRepo = dailyFactorsProcess()
dailyData = dailyRepo.getSingleStockDailyFactors(code, date, date)
dailyKLineRepo = KLineDataProcess('daily')
dailyKLineData = dailyKLineRepo.getDataByDate(code, date, date)
mydata['preClose'] = dailyKLineData['preClose'].iloc[0]
mydata['increaseToday'] = mydata['midPrice'] / mydata['preClose'] - 1
mydata = mydata[mydata['tick'] < '145700000']
#删去涨跌停之后的数据
ceiling = mydata[(mydata['B1'] == 0) | (mydata['S1'] == 0)]
if ceiling.shape[0] > 0:
ceilingTime = ceiling['tick'].iloc[0]
mydata = mydata[mydata['tick'] < ceilingTime]
pass
excludedColumns = [
'preClose', 'buyVolume2', 'buyVolume5', 'buyVolume10',
'sellVolume2', 'sellVolume5', 'sellVolume10'
]
mycolumns = list(
set(mydata.columns).difference(
set(list(tickData.columns) + excludedColumns)))
mycolumns.sort()
mydata = mydata[mycolumns]
if mydata.shape[0] == 0:
return
try:
logger.info(f'Recording factors of {code} in {date}!')
with pd.HDFStore(fileName,
'a',
complib='blosc:zstd',
append=True,
complevel=9) as store:
store.append('data',
mydata,
append=True,
format="table",
data_columns=mydata.columns)
except Exception as excp:
logger.error(f'{fileName} error! {excp}')
def __updateStockDailyBasicData(self, code, startDate, endDate):
marketData = KLineDataProcess('daily', True)
mydata = marketData.getDataByDate(code, startDate, endDate)
return mydata
pass
def updateSingleStockDailyKLines(self, code, startDate, endDate):
dailyStock = KLineDataProcess('daily', True)
dailyStock.getDataByDate(code, startDate, endDate)
pass