单因子 & 多因子策略（基于JoinQuant）

一份朴实的声明。。。

1. 基于joinquant，大部分代码是原作者所写，俺只是改写、补充。。。

2. 源码见：https://www.joinquant.com/post/775

3. 基于因子打分，不是因子回归

（思路：单因子打分，分值赋权个股形成单因子组合，再等权加权单因子组合，变为多因子策略）

----------------------------------------------------------------------------------------

## step1：导入所需的库

import pandas as pdfrom pandas

import Series, DataFrame

import numpy as npimport statsmodels.api as sm

import scipy.stats as scs

import matplotlib.pyplot as plt

## step2：三个函数

# 函数1：获取因子

factors = ['B/M','EPS','PEG','ROE','ROA','GP/R','P/R','L/A','FAP','CMV']

#月初取出因子数值

def get_factors(fdate,factors):

    stock_set = get_index_stocks('000001.XSHG',fdate)

    q = query( valuation.code, balance.total_owner_equities/valuation.market_cap/100000000, income.basic_eps, valuation.pe_ratio, income.net_profit/balance.total_owner_equities, income.net_profit/balance.total_assets, income.total_profit/income.operating_revenue, income.net_profit/income.operating_revenue, balance.total_liability/balance.total_assets, balance.fixed_assets/balance.total_assets, valuation.circulating_market_cap ).filter( valuation.code.in_(stock_set), valuation.circulating_market_cap )

    fdf = get_fundamentals(q, date=fdate)

    fdf.index = fdf['code']

    fdf.columns = ['code'] + factors

    return fdf.iloc[:,-10:]

# 函数2和3：计算因子及benchmark的月收益

def caculate_port_monthly_return(port,startdate,enddate,nextdate,CMV):

    close1 = get_price(port, startdate, enddate, 'daily', ['close']) #面板数据

    close2 = get_price(port, enddate, nextdate, 'daily',['close']) #面板数据

    weighted_m_return = ((close2['close'].ix[0,:]/close1['close'].ix[0,:]-1)).mean() #等权加权

    return weighted_m_return

def caculate_benchmark_monthly_return(startdate,enddate,nextdate):

    close1 = get_price(['000300.XSHG'],startdate,enddate,'daily',['close'])['close']

    close2 = get_price(['000300.XSHG'],enddate, nextdate, 'daily',['close'])['close']

    benchmark_return = (close2.ix[0,:]/close1.ix[0,:]-1).sum()

    print close1

    return benchmark_return

## step3：核心策略

factors = ['B/M','EPS','PEG','ROE','ROA','GP/R','P/R','L/A','FAP','CMV']#因为研究模块取fundmental数据默认date为研究日期的前一天。所以要自备时间序列。按月取

year = ['2011','2012','2013','2014','2015','2016','2017']

month = ['01','02','03','04','05','06','07','08','09','10','11','12']

result = {}

for i in range(7*12):

    startdate = year[i/12] + '-' + month[i%12] + '-01'

    try:

        enddate = year[(i+1)/12] + '-' + month[(i+1)%12] + '-01'

    except IndexError:

        enddate = '2018-01-01'

    try:

        nextdate = year[(i+2)/12] + '-' + month[(i+2)%12] + '-01'

    except IndexError:

        if enddate == '2018-01-01':

            nextdate = '2018-02-01'

        else:

            nextdate = '2018-01-01'

            #print 'time %s'%startdate

fdf = get_factors(startdate,factors)

CMV = fdf['CMV'] #5个组合，10个因子

df = DataFrame(np.zeros(6*10).reshape(6,10),index = ['port1','port2','port3','port4','port5','benchmark'],columns = factors)

for fac in factors:

    score = fdf[fac].order()

    port1 = list(score.index)[: len(score)/5]

    port2 = list(score.index)[ len(score)/5+1: 2*len(score)/5]

    port3 = list(score.index)[ 2*len(score)/5+1: -2*len(score)/5]

    port4 = list(score.index)[ -2*len(score)/5+1: -len(score)/5]

    port5 = list(score.index)[ -len(score)/5+1: ]

    df.ix['port1',fac] = caculate_port_monthly_return(port1,startdate,enddate,nextdate,CMV)

    df.ix['port2',fac] = caculate_port_monthly_return(port2,startdate,enddate,nextdate,CMV)

    df.ix['port3',fac] = caculate_port_monthly_return(port3,startdate,enddate,nextdate,CMV)

    df.ix['port4',fac] = caculate_port_monthly_return(port4,startdate,enddate,nextdate,CMV)

    df.ix['port5',fac] = caculate_port_monthly_return(port5,startdate,enddate,nextdate,CMV)

    df.ix['benchmark',fac] = caculate_benchmark_monthly_return(startdate,enddate,nextdate)

    #print 'factor %s'%fac

result[i+1]=dfmonthly_return = pd.Panel(result)

## step4：模型评价

total_return = {}

annual_return = {}

excess_return = {}

annual_volatility={}

sharpe={}

information_ratio={}

win_prob = {}

drawdown={}

tr={}

for fac in factors:

    monthly = monthly_return[:,:,fac]

    total_return[fac] = (monthly+1).T.cumprod().iloc[-1,:]-1

    annual_return[fac] = (total_return[fac]+1)**(1./6)-1

    excess_return[fac] = annual_return[fac]*0.7- annual_return[fac][-1]

    annual_volatility[fac]=monthly.T.std()

    sharpe[fac]=(annual_return[fac]-0.04)/annual_volatility[fac]

    information_ratio[fac]=(annual_return[fac]-annual_return[fac][-1])/annual_volatility[fac]

    drawdown[fac]=monthly.T.min()

    win_excess = monthly.iloc[0,:]*0.7-monthly.iloc[-1,:]

    win_prob[fac]=win_excess[win_excess>0].count()/float(len(win_excess))

### step5：Smart Beta指数的编制 #把一个三维的panel按因子降维，即对因子求平均，使因子这一维度消失掉 # 也即将因子类子指数等权加权，编制成指数 columns= ['port1','port2','port3','port4','port5','benchmark'] avg_data= DataFrame(np.zeros(85*6).reshape(85,6),columns= ['port1','port2','port3','port4','port5','benchmark']) for i in columns: avg_data[i]=pd.DataFrame((monthly_return[:,i,:]).mean()) #(monthly_return[:,2,:]).mean() #avg_data avg_total_return=(avg_data+1).cumprod().iloc[-1,:]-1 avg_annual_return= (avg_total_return+1)**(1./6)-1 avg_excess_return= avg_annual_return*0.7- avg_annual_return[-1] avg_annual_volatility=avg_data.std() avg_information_ratio=avg_excess_return/avg_annual_volatility avg_sharpe=(avg_annual_return-0.04)/avg_annual_volatility avg_drawdown=avg_data.min() avg_win_excess = avg_data.T.iloc[5,:]*0.7-avg_data.T.iloc[-1,:] avg_win_prob=avg_win_excess[avg_win_excess>0].count()/float(len(avg_win_excess)) avg_win_prob plt.figure(figsize =(8,4)) plt.plot((avg_data+1).cumprod().ix[:,0], 'r',label = '%s'%fac ,alpha=0.5) plt.plot((monthly.ix[-1,:].T+1).cumprod(), 'blue',label = '沪深300指数',alpha=0.5) plt.xlabel('净值曲线') plt.legend(loc='upper left',ncol=2)