def risk_multiple(model):
# 风险因子业绩归因模块
# fof_multi_attr_riskmodel
if not del_data('fof_multi_attr_riskmodel'):
raise Error("删除历史数据失败，任务结束")
stockIndus, indusReturn, factorExposure, factorReturn = publicData_riskModel(
)
# 批量运行
'''
跨期归因结果需要存储多个周期值
单期归因结果仅存储成立以来的值即可
跨期归因和单期归因可以用两张表来存储数据
'''
# dataframe 周期类型和数据库周期类型映射关系
cyleFormula = {'1': '1', '2': '2', '3': '3', '5': '4', '成立以来': '5'}
fundSymbols = getData_fundSymbols(['股票型', '混合型'])
cycleList = ['1', '2', '3', '5', '成立以来']
holdingTypeList = ['mainStockHolding', 'allStockHolding']
for holdingType in holdingTypeList:
for symbol in fundSymbols:
for cycle in cycleList:
btParms = {
'symbol': symbol,
'cycle': cycle,
'holdingType': holdingType
}
self = attribution_riskModel(btParms, stockIndus, indusReturn,
factorExposure, factorReturn)
output = self.output
tp = "1" if holdingType == 'mainStockHolding' else '2' # 持仓类型
param = output['multiAttr']
if len(param) == 0:
continue
for idx in param.index:
di = param.loc[idx].to_dict()
indexName = idx # 因子名称
indexType = di['factorType'] # 因子分类
indexAttr = di['multiAttr'] # 因子贡献收益
cyleTp = cyleFormula[cycle] # 周期类型
lsd = []
id_ = uuid_util.gen_uuid()
lsd.append(id_)
lsd.append(output['symbol'])
lsd.append(tp)
lsd.append(cyleTp)
lsd.append(indexType)
lsd.append(indexName)
lsd.append(indexAttr)
lsd.append("sys")
lsd.append(datetime.now())
lsd.append("sys")
lsd.append(datetime.now())
lsd = [transformFloatIfAvaliable4(l) for l in lsd]
sql = 'INSERT INTO `fof`.`fof_multi_attr_riskmodel`(`OBJECT_ID`, `S_INFO_WINDCODE`, `DATA_TYPE`, `CYCLE_TYPE`, `FACTOR_TYPE`, `FACTOR_NAME`, `FACTOR_VALUE`, `CREATE_USER_ID`, `CREATE_TIME`, `UPDATE_USER_ID`, `UPDATE_TIME`) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)'
mysqlops.insert_one(MysqlConf.DB.fof, sql, tuple(lsd))
def risk_single(model):
# 风险因子业绩归因模块
# fof_single_attr_riskmodel
if not del_data('fof_single_attr_riskmodel'):
raise Error("删除历史数据失败，任务结束")
stockIndus, indusReturn, factorExposure, factorReturn = publicData_riskModel(
)
# 批量运行
'''
跨期归因结果需要存储多个周期值
单期归因结果仅存储成立以来的值即可
跨期归因和单期归因可以用两张表来存储数据
'''
fundSymbols = getData_fundSymbols(['股票型', '混合型'])
cycleList = ['成立以来'] # 这个程序跟之前的程序有一点特殊，就是多期归因要存多个周期的结果，单期归因只存成立以来的结果
holdingTypeList = ['mainStockHolding', 'allStockHolding']
for holdingType in holdingTypeList:
for symbol in fundSymbols:
for cycle in cycleList:
btParms = {
'symbol': symbol,
'cycle': cycle,
'holdingType': holdingType
}
self = attribution_riskModel(btParms, stockIndus, indusReturn,
factorExposure, factorReturn)
output = self.output
tp = "1" if holdingType == 'mainStockHolding' else '2' # 持仓类型
param = output['singleAttr']
if len(param) == 0:
continue
for idx in param.index:
dt = idx # time
di = param.loc[idx].to_dict()
style = di['style'] # 风格因子归因
industry = di['industry'] # 行业因子归因
idiosyn = di['idiosyn'] # 特质因子归因
fundRet = di['fundRet'] # 基金当月收益率
lsd = []
id_ = uuid_util.gen_uuid()
lsd.append(id_)
lsd.append(output['symbol'])
lsd.append(dt._short_repr.replace("-", ""))
lsd.append(tp)
lsd.append(style)
lsd.append(industry)
lsd.append(idiosyn)
lsd.append(fundRet)
lsd.append("sys")
lsd.append(datetime.now())
lsd.append("sys")
lsd.append(datetime.now())
lsd = [transformFloatIfAvaliable4(l) for l in lsd]
sql = 'INSERT INTO `fof`.`fof_single_attr_riskmodel`(`OBJECT_ID`, `S_INFO_WINDCODE`, `TRADE_DT`, `DATA_TYPE`, `STYLE`, `INDUSTRY`, `IDIOSYN`, `FUND_RETURN`, `CREATE_USER_ID`, `CREATE_TIME`, `UPDATE_USER_ID`, `UPDATE_TIME`) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)'
mysqlops.insert_one(MysqlConf.DB.fof, sql, tuple(lsd))
def other_indicator(model: OfflineTaskModel):
# 离线计算其他指标
indicatorInfo = model.taskModel
indicatorId = model.extVal
if del_data(sqlPa="delete from fof_index_value where INDICATOR_ID = '%s'" % indicatorId) == False:
raise Error("删除历史数据失败，任务结束")
if indicatorId not in INDEX_CODE_NAME_CACHE.keys():
raise Error("指标id不存在:{}".format(indicatorId))
indexName = INDEX_CODE_NAME_CACHE[indicatorId]
# fundAdjNav = getPublicData(['股票型', '混合型'])
fundAdjNav = getPublicData()
fundSymbols = list(fundAdjNav.columns)
updatedType = []
fundClass_1 = pd.Series('混合型基金', index=fundSymbols)
fundClass_tmp = getData_fundInformation(fundSymbols)['FUND_INVESTTYPE']
fundClass_1[fundClass_tmp.index] = fundClass_tmp
for symbol in fundSymbols:
allIndicators = []
fundType = fundClass_1[symbol]
if fundType not in updatedType:
for cycle in n_cycle_tp:
btParms = {'indexName': indexName, 'cycle': cycle, 'symbol': symbol, 'sample': '一级',
'marketIndex': '',
'otherPar': ''}
self = indexScore_fund(btParms, fundAdjNav, 'auto')
op = self.output
val = op['factorValue']
allIndicators.append(val)
updatedType.append(fundType)
df = pd.DataFrame(allIndicators).fillna(bus_const.blank)
cols = df.columns.tolist()
for col in cols:
res = df[col]
lst = res.values.tolist()
lst = [transformFloatIfAvaliable(l) for l in lst]
objId = uuid_util.gen_uuid()
lst.insert(0, str(objId))
lst.insert(1, str(indicatorId))
lst.insert(2, col)
date = formatDate2yyyymmdd()
lst.insert(3, str(date))
lst.append("sys")
lst.append(datetime.datetime.now())
lst.append("sys")
lst.append(datetime.datetime.now())
tp = tuple(lst)
sql = "INSERT INTO fof_index_value (`OBJECT_ID`, `INDICATOR_ID`, `J_WINDCODE`, `TRADE_DT`, `THISYEAR_VALUE`, `QUARTER_VALUE`, `HALFYEAR_VALUE`, `YEAR_VALUE`, `TWOYEA_VALUE`, `THREEYEAR_VALUE`, `FIVEYEAR_VALUE`, `N1_VALUE`, `N2_VALUE`, `CREATE_USER_ID`, `CREATE_TIME`, `UPDATE_USER_ID`, `UPDATE_TIME`, `DELETE_FLAG`) VALUES ( %s, %s, %s, %s, %s, %s,%s,%s, %s,%s, %s, %s, %s, %s, %s, %s, %s,0)"
mysqlops.insert_one(MysqlConf.DB.fof, sql, tp)
allIndicators.clear()
def volatility(model: OfflineTaskModel):
# 离线计算年化波动率
if del_data('fof_variance') == False:
raise Error("删除历史数据失败，任务结束")
indicatorInfo = model.taskModel
taskName = indicatorInfo.value[0]
indiId = get_indicator_id(taskName)
# fundAdjNav = getPublicData(['股票型', '混合型'])
fundAdjNav = getPublicData()
fundSymbols = list(fundAdjNav.columns)
fundClass_1 = pd.Series('混合型基金', index=fundSymbols)
fundClass_tmp = getData_fundInformation(fundSymbols)['FUND_INVESTTYPE']
fundClass_1[fundClass_tmp.index] = fundClass_tmp
halfs = indicatorInfo.value[1]
for half in halfs:
updatedType = []
allIndicators = []
for symbol in fundSymbols:
fundType = fundClass_1[symbol]
if fundType not in updatedType:
for cycle in n_cycle_tp:
btParms = {'indexName': taskName, 'cycle': cycle, 'symbol': symbol, 'sample': '一级',
'marketIndex': '',
'otherPar': half}
self = indexScore_fund(btParms, fundAdjNav, 'auto')
op = self.output
val = op['factorValue']
allIndicators.append(val)
updatedType.append(fundType)
df = pd.DataFrame(allIndicators).fillna(bus_const.blank)
cols = df.columns.tolist()
for col in cols:
res = df[col]
lst = res.values.tolist()
lst = [transformFloatIfAvaliable(l) for l in lst]
objId = uuid_util.gen_uuid()
lst.insert(0, str(objId))
lst.insert(1, str(indiId))
lst.insert(2, str(col))
date = formatDate2yyyymmdd()
lst.insert(3, str(date))
lst.insert(4, transformString2Decimal(half))
lst.append("sys")
lst.append(datetime.datetime.now())
lst.append("sys")
lst.append(datetime.datetime.now())
tp = tuple(lst)
sql = "INSERT INTO fof_variance (`OBJECT_ID`, `INDICATOR_ID`, `J_WINDCODE`, `TRADE_DT`, `F_WEIGHT`, `THISYEAR_VALUE`, `QUARTER_VALUE`, `HALFYEAR_VALUE`, `YEAR_VALUE`, `TWOYEA_VALUE`, `THREEYEAR_VALUE`, `FIVEYEAR_VALUE`, `N1_VALUE`, `N2_VALUE`, `CREATE_USER_ID`, `CREATE_TIME`, `UPDATE_USER_ID`, `UPDATE_TIME`, `DELETE_FLAG`) VALUES ( %s, %s, %s, %s, %s, %s, %s,%s, %s, %s, %s, %s, %s, %s, %s, %s, %s,%s,0)"
mysqlops.insert_one(MysqlConf.DB.fof, sql, tp)
allIndicators.clear()
def stockexpousre(model):
# factorName即为mongodb数据库相关collection的名字，一次仅支持传入一个因子值
sql = "delete from fof_stockexpousre "
mysqlops.fetch_one(MysqlConf.DB.fof, sql)
if not del_data('fof_stockexpousre'):
raise Error("删除历史数据失败，任务结束")
fs = [
'value',
'size',
'beta',
'earning',
# 'factorReturn',
'growth',
'leverage',
'liquidity',
'momentum',
'nonlinear_size',
'size',
'volatility'
]
sql = 'insert into ' \
'fof_stockexpousre (OBJECT_ID,trade_dt,s_windcode,indicator_code,factor_value,CREATE_USER_ID,CREATE_TIME,UPDATE_USER_ID,UPDATE_TIME)' \
'values (%s,%s,%s,%s,%s,%s,%s,%s,%s)'
for fName in fs:
valueExp = getData_factorExposure(fName)
valueExp = valueExp.fillna("9999.000000")
di = valueExp.to_dict()
for k in di:
# print("K===%s" % k)
for v in di.get(k):
s = uuid_util.gen_uuid()
lsd = []
oid = s[:5] + '-' + s[5:9] + '-' + s[9:13] + '-' + s[
13:18] + '-' + s[18:]
lsd.append(oid)
r = v._date_repr.replace("-", "")
lsd.append(r)
lsd.append(k[:-2] + "." + k[-2:])
lsd.append(fName)
lsd.append(transformFloatIfAvaliable(di.get(k).get(v)))
lsd.append("sys")
lsd.append(datetime.now())
lsd.append("sys")
lsd.append(datetime.now())
mysqlops.insert_one(MysqlConf.DB.fof, sql, tuple(lsd))
def indicator_score_rank(model: OfflineTaskModel):
# 离线计算指标得分
indicatorId = model.extVal
if indicatorId not in INDEX_CODE_NAME_CACHE.keys():
raise Error("指标id不存在:{}".format(indicatorId))
if not del_data(
sqlPa="delete from fof_index_score where INDICATOR_ID = '%s' " %
indicatorId):
raise Error("删除历史数据失败，任务结束")
iName = INDEX_CODE_NAME_CACHE[indicatorId]
# fundAdjNav = getPublicData(['股票型', '混合型'])
fundAdjNav = getPublicData()
fundSymbols = list(fundAdjNav.columns)
updatedType = []
fundClass_1 = pd.Series('混合型基金', index=fundSymbols)
fundClass_tmp = getData_fundInformation(fundSymbols)['FUND_INVESTTYPE']
fundClass_1[fundClass_tmp.index] = fundClass_tmp
for symbol in fundSymbols:
indicators = []
ranks = []
fundType = fundClass_1[symbol]
if fundType not in updatedType:
for cycle in cycle_score:
btParms = {
'indexName': iName,
'cycle': cycle,
'symbol': symbol,
'sample': '一级',
'marketIndex': '',
'otherPar': ''
}
self = indexScore_fund(btParms, fundAdjNav, 'auto')
op = self.output
score = op['factorScore']
rank = op['factorRank']
# 这个值是用来记录当前cycle及分类下参与排名的样本数量，因为前端显示的排名都是5 / 200
# 这种形式，本意是要存入数据库的，但我跟平赞沟通过了，他说他们前端直接根据数据库里的数据处理即可，不用存这个字段了
# sc = op['sampleCounts']
# tp = score
# for ti in tp.index:
# tp[ti] = str(sc)
indicators.append(score)
ranks.append(rank)
ranks.append(pd.Series()) # 一级分类排名样本
ranks.append(pd.Series()) # 二级分类排名一期不上，先做个空值
ranks.append(pd.Series()) # 二级分类排名样本一期不上，先做个空值
updatedType.append(fundType)
indicators.extend(ranks)
df = pd.DataFrame(indicators).fillna('9999999999999.0')
cols = df.columns.tolist()
for col in cols:
res = df[col] # [基金id,score,score,...]
lst = res.values.tolist()
lst = [transformFloatIfAvaliable2(l) for l in lst]
objId = uuid_util.gen_uuid()
lst.insert(0, str(objId))
lst.insert(1, str(indicatorId))
lst.insert(2, col)
# lst.insert(2, list(map(lambda x: x[-2:] + x[:-2], [col]))[0])
date = formatDate2yyyymmdd()
lst.insert(3, str(date))
lst.append("sys")
lst.append(datetime.datetime.now())
lst.append("sys")
lst.append(datetime.datetime.now())
lst = [
'9999999999999.0' if type(ele) == pd.Series else ele
for ele in lst
]
tp = tuple(lst)
sql = "INSERT INTO fof_index_score VALUES ( %s, %s, %s, %s, %s, %s, %s,%s, %s, %s, %s, %s, %s, %s,%s, %s, %s, %s, %s, %s, %s, %s,%s,%s,%s, %s, %s, %s, %s, %s, %s, %s,%s,%s,%s, %s, %s, %s, %s, %s, %s, %s,%s,0)"
mysqlops.insert_one(MysqlConf.DB.fof, sql, tp)
indicators.clear()
ranks.clear()
def information_ratio(model: OfflineTaskModel):
# 离线计算信息比率
if del_data('fof_inforatio') == False:
raise Error("删除历史数据失败，任务结束")
# sql = "delete from fof_inforatio "
# mysqlops.fetch_one(MysqlConf.DB.fof, sql)
indicatorInfo = model.taskModel
taskName = indicatorInfo.value[0]
indiId = get_indicator_id(taskName)
# fundAdjNav = getPublicData(['股票型', '混合型'])
fundAdjNav = getPublicData()
fundSymbols = list(fundAdjNav.columns)
updatedType = []
fundClass_1 = pd.Series('混合型基金', index=fundSymbols)
fundClass_tmp = getData_fundInformation(fundSymbols)['FUND_INVESTTYPE']
fundClass_1[fundClass_tmp.index] = fundClass_tmp
for symbol in fundSymbols:
allIndicators = []
fundType = fundClass_1[symbol]
if fundType not in updatedType:
for cycle in n_cycle_tp:
btParms = {'indexName': taskName, 'cycle': cycle, 'symbol': symbol, 'sample': '一级',
'marketIndex': '',
'otherPar': ''}
self = indexScore_fund(btParms, fundAdjNav, 'auto')
op = self.output
val = op['factorValue']
allIndicators.append(val)
updatedType.append(fundType)
df = pd.DataFrame(allIndicators).fillna(bus_const.blank)
cols = df.columns.tolist()
for col in cols:
res = df[col]
dd = res.values.tolist()
lst = [transformFloatIfAvaliable(l) for l in dd]
objId = uuid_util.gen_uuid()
lst.insert(0, str(objId))
lst.insert(1, str(indiId))
lst.insert(2, col)
date = formatDate2yyyymmdd()
lst.insert(3, str(date))
# idxVal = None
# try:
# idxVal = CODE_INDEX_CACHE[col]
# except:
# # 缓存没有查一次数据库
# sql = "SELECT s_info_windcode,s_info_indexwindcode FROM chinamutualfundbenchmark where S_INFO_WINDCODE = '{}'".format(
# col)
# res = mysqlops.fetchmany(MysqlConf.DB.fof, sql)
# if res and 's_info_indexwindcode' in res and res['s_info_indexwindcode'] is not None:
# idxVal = res['s_info_indexwindcode'].decode()
lst.insert(4, indicatorInfo.value[1]) # 比较基准wind代码
lst.append("sys")
lst.append(datetime.datetime.now())
lst.append("sys")
lst.append(datetime.datetime.now())
tp = tuple(lst)
sql = "INSERT INTO fof_inforatio (`OBJECT_ID`, `INDICATOR_ID`, `J_WINDCODE`, `TRADE_DT`, `B_WINDCODE`, `THISYEAR_VALUE`, `QUARTER_VALUE`, `HALFYEAR_VALUE`, `YEAR_VALUE`, `TWOYEA_VALUE`, `THREEYEAR_VALUE`, `FIVEYEAR_VALUE`, `N1_VALUE`, `N2_VALUE`, `CREATE_USER_ID`, `CREATE_TIME`, `UPDATE_USER_ID`, `UPDATE_TIME`, `DELETE_FLAG`) VALUES ( %s, %s, %s, %s, %s, %s, %s,%s,%s,%s, %s, %s, %s, %s, %s, %s, %s, %s,0)"
mysqlops.insert_one(MysqlConf.DB.fof, sql, tp)
allIndicators.clear()