def dailyCalculation(cur_date,T,N):
gl.logger.info('Start to do daily caculation!')
#日期获取
date_list=om.getNtradeDate(cur_date,-(T+N+1))
gl.logger.info('Date list get successfully,date interval is [%d,%d]'%(date_list[0],date_list[-1]))
#收盘价数据、因子数据准备
gl.logger.info('Start to do data prepare!')
date_time = datetime.datetime.now()
[return_df,train_factor_df,test_factor_df]=dataPrepare(date_list,T,N)
cost_seconds=(datetime.datetime.now()-date_time).seconds
gl.logger.info('Data preparation is done,consumes %d seconds.'%cost_seconds)
#模型预测
model_task_flag=modelPredict(train_factor_df,return_df,test_factor_df)
#标准化的因子数据
normalized_flag=factorDataNormalized(test_factor_df)
if(model_task_flag==1 and normalized_flag==1):
gl.logger.info('Daily stock rating task successfully! ')
else:
gl.logger.error('The model falg is %d, the normalized data flag is %d,daily task is failed, please check!'%(model_task_flag,normalized_flag))
def factor_prepare(cur_date):
gl.logger.info("start to prepare factor data----------")
n = -365 # 主要因为dq_turn会返回周六周日的日期,数据为na,但是占了行数
# 取某个日期前(负，-7)or后(正，+7)N天的交易日日期,date_list[0]为最近的一天
date_list = om.getNtradeDate(cur_date, n)
# 12个月的个股收益对沪深300 线性回归
gl.logger.info('start to prepare regress data')
regress_result = stock_regress(cur_date, date_list)
regress_result = pd.DataFrame.from_dict(regress_result, orient='columns')
# index分别为权重,常数项,经过处理的残差
regress_result.index = ['coef', 'const', 'beta_consis']
# 获取500天的复权收益列表,为降序排列,第0个位最新的
gl.logger.info('start to prepare adjclose data')
adjclose = om.get_ndays_adjclose(cur_date, date_list)
# 格式化index为日期,columns为股票代码,降序,第0个位最新的
date_adjclose = fh.transform_large_df(adjclose)
# 换手率加权*日均收益率
# 获取换手率，不带index，为了格式化df成为index为日期，columns为股票代码
gl.logger.info('start to prepare dq_turn data')
dq_turn = om.get_dq_turn(cur_date, date_list)
date_dq_turn = fh.transform_large_df(dq_turn)
# 这个会返回周六周日,数据为na,所以导致n取365
date_dq_turn.dropna(how='all', inplace=True)
# -----------------------------------------------------------------------------
# 估值因子,value factor
gl.logger.info('start to run factor_value function')
factor_value_flag = factor_value(cur_date)
# -----------------------------------------------------------------------------
# 成长因子,growth factor
gl.logger.info('start to run factor_growth function')
factor_growth_flag = factor_growth(cur_date)
# -----------------------------------------------------------------------------
# 财务质量因子,financial quality factor
gl.logger.info('start to run factor_financial_quality function')
factor_financial_flag = factor_financial_quality(cur_date)
# -----------------------------------------------------------------------------
# 杠杠因子,leverage factor
# 规模因子,size factor
gl.logger.info('start to run factor_leverage function')
factor_leverage_flag = factor_leverage(cur_date)
# -----------------------------------------------------------------------------
# 动量因子,momentum factor
gl.logger.info('start to run factor_momentum function')
factor_momentum_flag = factor_momentum(cur_date, regress_result, date_adjclose)
# -----------------------------------------------------------------------------
# 波动率因子,volatility factor
gl.logger.info('start to run factor_volatility function')
factor_volatility_flag = factor_volatility(cur_date, date_list, regress_result, date_adjclose)
# -----------------------------------------------------------------------------
# 换手率因子,turnover factor
gl.logger.info('start to run factor_turnover function')
factor_turnover_flag = factor_turnover(cur_date, date_list)
# factor_turnover_flag = 1
# -----------------------------------------------------------------------------
# 改进的动量因子,modified momentum factor
gl.logger.info('start to run factor_modified_momentum function')
factor_modified_flag = factor_modified_momentum(cur_date, date_list, date_adjclose, date_dq_turn)
# -----------------------------------------------------------------------------
gl.logger.info('start to run factor_close_bak function')
close_bak_flag = factor_close_bak(cur_date)
flag = factor_value_flag & factor_growth_flag & factor_financial_flag & factor_leverage_flag & factor_momentum_flag \
& factor_volatility_flag & factor_turnover_flag & factor_modified_flag & close_bak_flag
# 所有函数的返回值是不是都成功了
work_time = time.strftime('%Y%m%d%H%M%S')
work_id = 'FactorPrepare_{}'.format(work_time)
entry_time = time.strftime('%Y-%m-%d %H:%M:%S')
if flag != 1:
status_flag = om.status_insert([[work_id, 0, entry_time]])
gl.logger.error('some function are failed----------')
return 0
else:
status_flag = om.status_insert([[work_id, 1, entry_time]])
gl.logger.info('all function are successful----------')
return 1