def getResult(strategyName, strategy_class, symbolinfo, K_MIN, rawdataDic,
para, result_para_dic, indexcols, timestart):
time1 = time.time()
setname = para['Setname']
print("%s %s %d %s Enter %.3f" % (strategyName, symbolinfo.domain_symbol,
K_MIN, setname, time1 - timestart))
initialCash = result_para_dic['initialCash']
positionRatio = result_para_dic['positionRatio']
remove_polar_switch = result_para_dic['remove_polar_switch']
remove_polar_rate = result_para_dic['remove_polar_rate']
symbollist = symbolinfo.getSymbolList()
symbolDomainDic = symbolinfo.getSymbolDomainDic()
result = pd.DataFrame()
last_domain_utc = None
for symbol in symbollist:
if last_domain_utc:
# 如果上一个合约的最后一次平仓时间超过其主力合约结束时间，则要修改本次合约的开始时间为上一次平仓后
symbol_domain_start = last_domain_utc
symbolDomainDic[symbol][0] = last_domain_utc
else:
symbol_domain_start = symbolDomainDic[symbol][0]
symbol_domain_end = symbolDomainDic[symbol][1]
rawdata = rawdataDic[symbol]
r = strategy_class.run_trade_logic(symbolinfo, rawdata, para)
r['symbol'] = symbol # 增加主力全约列
r = r.loc[(r['openutc'] >= symbol_domain_start)
& (r['openutc'] <= symbol_domain_end)]
last_domain_utc = None
if r.shape[0] > 0:
last_close_utc = r.iloc[-1]['closeutc']
if last_close_utc > symbol_domain_end:
# 如果本合约最后一次平仓时间超过其主力合约结束时间，则要修改本合约的主力结束时间为平仓后
symbolDomainDic[symbol][1] = last_close_utc
last_domain_utc = last_close_utc
result = pd.concat([result, r])
result.reset_index(drop=True, inplace=True)
# 去极值操作
if remove_polar_switch:
result = RS.opr_result_remove_polar(result, remove_polar_rate)
# 全部操作结束后，要根据修改完的主力时间重新接出一份主连来计算dailyK
domain_bar = pd.DataFrame()
for symbol in symbollist[:-1]:
symbol_domain_start = symbolDomainDic[symbol][0]
symbol_domain_end = symbolDomainDic[symbol][1]
rbar = rawdataDic[symbol]
bars = rbar.loc[(rbar['utc_time'] >= symbol_domain_start)
& (rbar['utc_endtime'] < symbol_domain_end)]
domain_bar = pd.concat([domain_bar, bars])
# 最后一个合约只截前不截后
symbol = symbollist[-1]
symbol_domain_start = symbolDomainDic[symbol][0]
rbar = rawdataDic[symbol]
bars = rbar.loc[rbar['utc_time'] >= symbol_domain_start]
domain_bar = pd.concat([domain_bar, bars])
dailyK = DI.generatDailyClose(domain_bar)
result['commission_fee'], result['per earn'], result['own cash'], result[
'hands'] = RS.calcResult(result, symbolinfo, initialCash,
positionRatio)
bt_folder = "%s %d backtesting\\" % (symbolinfo.domain_symbol, K_MIN)
result.to_csv(bt_folder + strategyName + ' ' + symbolinfo.domain_symbol +
str(K_MIN) + ' ' + setname + ' result.csv',
index=False)
dR = RS.dailyReturn(symbolinfo, result, dailyK, initialCash) # 计算生成每日结果
dR.calDailyResult()
dR.dailyClose.to_csv(
(bt_folder + strategyName + ' ' + symbolinfo.domain_symbol +
str(K_MIN) + ' ' + setname + ' dailyresult.csv'))
results = RS.getStatisticsResult(result, False, indexcols, dR.dailyClose)
del result
print results
return [setname] + results # 在这里附上setname
def single_sl(strategy_name, symbol_info, bar_type, setname, bar1m_dic,
barxm_dic, stop_loss_class_list, result_para_dic, indexcols,
timestart):
print "%s %s %d %s" % (strategy_name, symbol_info.domain_symbol, bar_type,
setname)
symbol = symbol_info.domain_symbol
bt_folder = "%s %d backtesting\\" % (symbol, bar_type)
oprdf = pd.read_csv(bt_folder + strategy_name + ' ' + symbol +
str(bar_type) + ' ' + setname + ' result.csv')
close_type_list = []
all_final_result_dic = {} # 这个是用来保存每个文件的RS结果，返回给外部调用的
all_stop_loss_opr_result_dic = {} # 这个是用来保存每个参数每次操作的止损结果
for stop_loss_class in stop_loss_class_list:
sl_name = stop_loss_class.get_sl_name()
close_type_list.append(sl_name)
final_result_dic = {}
stop_loss_opr_result_dic = {}
for para in stop_loss_class.get_para_dic_list():
final_result_dic[para['para_name']] = []
stop_loss_opr_result_dic[para['para_name']] = []
all_stop_loss_opr_result_dic[sl_name] = stop_loss_opr_result_dic
all_final_result_dic[sl_name] = final_result_dic
for stop_loss_class in stop_loss_class_list:
if stop_loss_class.need_data_process_before_domain:
bar1m_dic, barxm_dic = stop_loss_class.data_process_before_domain(
bar1m_dic, barxm_dic)
symbolDomainDic = symbol_info.amendSymbolDomainDicByOpr(oprdf)
bar1m = DI.getDomainbarByDomainSymbol(symbol_info.getSymbolList(),
bar1m_dic, symbolDomainDic)
bar1m = bar1m_prepare(bar1m)
barxm = DI.getDomainbarByDomainSymbol(symbol_info.getSymbolList(),
barxm_dic, symbolDomainDic)
barxm.set_index('utc_time', drop=False, inplace=True) # 开始时间对齐
bar1m.set_index('utc_time', drop=False, inplace=True)
for stop_loss_class in stop_loss_class_list:
if stop_loss_class.need_data_process_after_domain:
bar1m, barxm = stop_loss_class.data_process_after_domain(
bar1m, barxm)
positionRatio = result_para_dic['positionRatio']
initialCash = result_para_dic['initialCash']
oprnum = oprdf.shape[0]
worknum = 0
for i in range(oprnum):
opr = oprdf.iloc[i]
#startutc = barxm.loc[opr['openutc'], 'utc_endtime'] # 从开仓的10m线结束后开始
#endutc = barxm.loc[opr['closeutc'], 'utc_endtime'] - 60 # 一直到平仓的10m线结束
startutc = opr['openutc']
endutc = opr['closeutc']
data_1m = bar1m.loc[startutc:endutc]
data1m = data_1m.drop(data_1m.index[-1]) # 因为loc取数是含头含尾的，所以要去掉最后一行
for stop_loss_class in stop_loss_class_list:
sl_name = stop_loss_class.get_sl_name()
stop_loss_opr_result_dic = all_stop_loss_opr_result_dic[sl_name]
opr_result_dic = stop_loss_class.get_opr_sl_result(opr, data1m)
for para in stop_loss_class.get_para_dic_list():
stop_loss_opr_result_dic[para['para_name']].append(
opr_result_dic[para['para_name']])
slip = symbol_info.getSlip()
olddailydf = pd.read_csv(bt_folder + strategy_name + ' ' + symbol +
str(bar_type) + ' ' + setname +
' dailyresult.csv',
index_col='date')
oldr = RS.getStatisticsResult(oprdf, False, indexcols, olddailydf)
dailyK = DI.generatDailyClose(barxm)
for stop_loss_class in stop_loss_class_list:
sl_name = stop_loss_class.get_sl_name()
stop_loss_opr_result_dic = all_stop_loss_opr_result_dic[sl_name]
final_result_dic = all_final_result_dic[sl_name]
folder_prefix = stop_loss_class.get_folder_prefix()
file_suffix = stop_loss_class.get_file_suffix()
for para_name, opr_result_dic_list in stop_loss_opr_result_dic.items():
result_df = pd.DataFrame(opr_result_dic_list)
oprdf_temp = pd.concat([oprdf, result_df], axis=1)
oprdf_temp['new_ret'] = (
(oprdf_temp['new_closeprice'] - oprdf_temp['openprice']) *
oprdf_temp['tradetype']) - slip
oprdf_temp[
'new_ret_r'] = oprdf_temp['new_ret'] / oprdf_temp['openprice']
oprdf_temp['new_commission_fee'], oprdf_temp['new_per earn'], oprdf_temp['new_own cash'], oprdf_temp['new_hands'] = \
RS.calcResult(oprdf_temp, symbol_info, initialCash, positionRatio, ret_col='new_ret')
# 保存新的result文档
tofolder = "%s%s\\" % (folder_prefix, para_name)
oprdf_temp.to_csv(tofolder + strategy_name + ' ' + symbol +
str(bar_type) + ' ' + setname + ' ' +
file_suffix + para_name + '.csv',
index=False)
dR = RS.dailyReturn(symbol_info, oprdf_temp, dailyK,
initialCash) # 计算生成每日结果
dR.calDailyResult()
dR.dailyClose.to_csv(tofolder + strategy_name + ' ' + symbol +
str(bar_type) + ' ' + setname + ' daily' +
file_suffix + para_name + '.csv')
newr = RS.getStatisticsResult(oprdf_temp, True, indexcols,
dR.dailyClose)
worknum = oprdf_temp.loc[oprdf_temp['new_closeindex'] !=
oprdf_temp['closeindex']].shape[0]
final_result_dic[para_name] = [setname, para_name, worknum
] + oldr + newr
return all_final_result_dic
def multi_stop_loss(strategyName, symbolInfo, K_MIN, setname,
stopLossTargetDictList, barxmdic, result_para_dic,
tofolder, indexcols):
print 'setname:', setname
symbol = symbolInfo.domain_symbol
bt_folder = "%s %d backtesting\\" % (symbol, K_MIN)
oprdf = pd.read_csv(bt_folder + strategyName + ' ' + symbol + str(K_MIN) +
' ' + setname + ' result.csv')
symbolDomainDic = symbolInfo.amendSymbolDomainDicByOpr(oprdf)
barxm = DI.getDomainbarByDomainSymbol(symbolInfo.getSymbolList(), barxmdic,
symbolDomainDic)
dailyK = DI.generatDailyClose(barxm)
positionRatio = result_para_dic['positionRatio']
initialCash = result_para_dic['initialCash']
oprlist = []
sltnum = len(stopLossTargetDictList)
for i in range(sltnum):
slt = stopLossTargetDictList[i]
# 遍历读取各止损目标的结果文件,按名称将结果写入oprdf中
sltdf = pd.read_csv("%s%s %s%d %s %s" %
(slt['folder'], strategyName, symbol, K_MIN,
setname, slt['fileSuffix']))
sltName = slt['name']
oprdf[sltName + '_closeprice'] = sltdf['new_closeprice']
oprdf[sltName + '_closetime'] = sltdf['new_closetime']
oprdf[sltName + '_closeindex'] = sltdf['new_closeindex']
oprdf[sltName + '_closeutc'] = sltdf['new_closeutc']
oprdf[sltName + '_ret'] = sltdf['new_ret']
oprdf[sltName + '_own cash'] = sltdf['new_own cash']
oprlist.append(sltdf)
# dsloprname=stratetyName+' '+symbol + str(K_MIN) + ' ' + setname + ' resultDSL_by_tick.csv'
# ownloprname=stratetyName+' '+symbol + str(K_MIN) + ' ' + setname + ' resultOWNL_by_tick.csv'
# dsloprdf=pd.read_csv(dslFolder+dsloprname)
# ownloprdf=pd.read_csv(ownlFolder+ownloprname)
oprdf['new_closeprice'] = oprdf['closeprice']
oprdf['new_closetime'] = oprdf['closetime']
oprdf['new_closeindex'] = oprdf['closeindex']
oprdf['new_closeutc'] = oprdf['closeutc']
oprdf['min_closeutc'] = oprdf['closeutc']
oprdf['max_closeutc'] = oprdf['closeutc']
for i in range(sltnum):
# 先取最早平仓的时间，再根据时间去匹配类型
slt = stopLossTargetDictList[i]
utcname = slt['name'] + '_closeutc'
oprdf['min_closeutc'] = oprdf.loc[:, ['min_closeutc', utcname]].min(
axis=1)
oprdf['max_closeutc'] = oprdf.loc[:, ['max_closeutc', utcname]].max(
axis=1)
# 根据最早平仓时间的结果，匹配平仓类型,不处理时间相同的情况
oprdf['closetype'] = 'Normal'
oprdf.loc[oprdf['max_closeutc'] != oprdf['closeutc'],
'min_closeutc'] = oprdf['max_closeutc']
for i in range(sltnum):
slt = stopLossTargetDictList[i]
name = slt['name']
utcname = name + '_closeutc'
utcnamebuf = name + '_closeutc_buf'
oprdf[utcnamebuf] = oprdf[utcname]
oprdf.loc[(oprdf['max_closeutc'] != oprdf['closeutc']) &
(oprdf[utcname] == oprdf['closeutc']),
utcnamebuf] = oprdf['max_closeutc']
for i in range(sltnum):
# 先取最早平仓的时间，再根据时间去匹配类型
slt = stopLossTargetDictList[i]
name = slt['name']
utcnamebuf = name + '_closeutc_buf'
oprdf['min_closeutc'] = oprdf.loc[:, ['min_closeutc', utcnamebuf]].min(
axis=1)
for i in range(sltnum):
# 先按与最小相同的标识名称，因为止损文件中没有生效的操作的值与原值相同
# 所以标识完后剩下的Normal就是原时间比止损时间早的值（也就是使用最小值匹配不出来的值，需要特殊处理）
slt = stopLossTargetDictList[i]
name = slt['name']
utcname = name + '_closeutc'
oprdf.loc[oprdf['min_closeutc'] == oprdf[utcname],
'closetype'] = slt['name']
oprdf.loc[oprdf['min_closeutc'] == oprdf[utcname],
'new_closeprice'] = oprdf[name + '_closeprice']
oprdf.loc[oprdf['min_closeutc'] == oprdf[utcname],
'new_closetime'] = oprdf[name + '_closetime']
oprdf.loc[oprdf['min_closeutc'] == oprdf[utcname],
'new_closeindex'] = oprdf[name + '_closeindex']
oprdf.loc[oprdf['min_closeutc'] == oprdf[utcname],
'new_closeutc'] = oprdf[name + '_closeutc']
oprdf.drop(name + '_closeutc_buf', axis=1, inplace=True) # 删掉buf列
# 标识正常止损
oprdf.loc[oprdf['min_closeutc'] == oprdf['closeutc'],
'closetype'] = 'Normal'
oprdf.drop('min_closeutc', axis=1, inplace=True)
oprdf.drop('max_closeutc', axis=1, inplace=True)
slip = symbolInfo.getSlip()
# 2017-12-08:加入滑点
oprdf['new_ret'] = ((oprdf['new_closeprice'] - oprdf['openprice']) *
oprdf['tradetype']) - slip
oprdf['new_ret_r'] = oprdf['new_ret'] / oprdf['openprice']
oprdf['new_commission_fee'], oprdf['new_per earn'], oprdf[
'new_own cash'], oprdf['new_hands'] = RS.calcResult(oprdf,
symbolInfo,
initialCash,
positionRatio,
ret_col='new_ret')
oprdf.to_csv(tofolder + '\\' + strategyName + ' ' + symbol + str(K_MIN) +
' ' + setname + ' result_multiSLT.csv',
index=False)
# 计算统计结果
slWorkNum = oprdf.loc[oprdf['closetype'] != 'Normal'].shape[0]
olddailydf = pd.read_csv(bt_folder + strategyName + ' ' + symbol +
str(K_MIN) + ' ' + setname + ' dailyresult.csv',
index_col='date')
oldr = RS.getStatisticsResult(oprdf, False, indexcols, olddailydf)
dR = RS.dailyReturn(symbolInfo, oprdf, dailyK, initialCash) # 计算生成每日结果
dR.calDailyResult()
dR.dailyClose.to_csv(tofolder + '\\' + strategyName + ' ' + symbol +
str(K_MIN) + ' ' + setname +
' dailyresult_multiSLT.csv')
newr = RS.getStatisticsResult(oprdf, True, indexcols, dR.dailyClose)
print newr
return [
setname,
tofolder,
slWorkNum,
] + oldr + newr
def calOprResult(strategyName,
rawpath,
symbolinfo,
K_MIN,
nextmonth,
columns,
barxmdic,
positionRatio,
initialCash,
indexcols,
indexcolsFlag,
resultfilesuffix='result.csv'):
'''
根据灰区的取值，取出各灰区的操作列表，组成目标集组的操作表，并计算各个评价指标
:return:
'''
symbol = symbolinfo.domain_symbol
graydf = pd.read_csv(rawpath + 'ForwardOprAnalyze\\' + strategyName + ' ' +
symbol + str(K_MIN) + 'multiTargetForwardSetname.csv',
index_col='Group')
cols = graydf.columns.tolist()[3:]
cols.append(nextmonth)
groupResult = []
closetime_col = columns['closetime_col']
closeindex_col = columns['closeindex_col']
closeprice_col = columns['closeprice_col']
closeutc_col = columns['closeutc_col']
retr_col = columns['retr_col']
ret_col = columns['ret_col']
cash_col = columns['cash_col']
hands_col = columns['hands_col']
for i in range(graydf.shape[0]):
gray = graydf.iloc[i]
oprdf = pd.DataFrame(columns=[
'opentime', 'openutc', 'openindex', 'openprice', closetime_col,
closeutc_col, closeindex_col, closeprice_col, 'tradetype', ret_col,
retr_col, 'symbol'
])
print gray.name, gray.Target, gray.Windows
for l in range(len(cols) - 1):
startmonth = cols[l]
endmonth = cols[l + 1]
setname = gray[startmonth]
oprdf = pd.concat([
oprdf,
getOprlistByMonth(strategyName, rawpath, symbol, K_MIN,
setname, startmonth, endmonth, columns,
resultfilesuffix)
])
oprdf = oprdf.reset_index(drop=True)
oprdf['commission_fee'], oprdf['per earn'], oprdf[cash_col], oprdf[
hands_col] = RS.calcResult(oprdf, symbolinfo, initialCash,
positionRatio, ret_col)
tofilename = ('%s %s%d_%s_win%d_oprResult.csv' %
(strategyName, symbol, K_MIN, gray.Target, gray.Windows))
oprdf.to_csv(rawpath + 'ForwardOprAnalyze\\' + tofilename)
symbolDomainDic = symbolinfo.amendSymbolDomainDicByOpr(
oprdf, closeutc_col=closeutc_col)
barxm = DC.getDomainbarByDomainSymbol(symbolinfo.getSymbolList(),
barxmdic, symbolDomainDic)
dailyK = DC.generatDailyClose(barxm) # 生成按日的K线
dR = RS.dailyReturn(symbolinfo, oprdf, dailyK, initialCash) # 计算生成每日结果
dR.calDailyResult()
tofilename = ('%s %s%d_%s_win%d_oprdailyResult.csv' %
(strategyName, symbol, K_MIN, gray.Target, gray.Windows))
dR.dailyClose.to_csv(rawpath + 'ForwardOprAnalyze\\' + tofilename)
r = RS.getStatisticsResult(oprdf, indexcolsFlag, indexcols,
dR.dailyClose)
groupResult.append([gray.name, gray.Target, gray.Windows] + r)
groupResultDf = pd.DataFrame(groupResult,
columns=['Group', 'Target', 'Windows'] +
indexcols)
groupResultDf.to_csv(rawpath + 'ForwardOprAnalyze\\' + strategyName + ' ' +
symbol + '_' + str(K_MIN) + '_groupOprResult.csv',
index=False)
pass
