# 导入函数库
import statsmodels.api as sm
from statsmodels import regression
import numpy as np
import pandas as pd
import jqdata
import matplotlib.pyplot as plt

#得到标的与时间数据
security = get_industry_stocks('801192','2016-12-31')
#创建一个pandas数据结构，用于储存各家银行的数据信息
DF = pd.DataFrame(np.zeros((len(security),6)),index=security,columns=['EVA','RORAC','PE_TTM','PE_LYR','PB','MC'])

#对于任何一家银行，获取税前利润和财务指标
for k in range(len(security)):
    df = get_fundamentals(query(
        income.total_profit, 
        valuation.pe_ratio,
        valuation.pe_ratio_lyr,
        valuation.pb_ratio,
        valuation.circulating_market_cap
    ).filter(
        valuation.code == security[k]
    ), statDate = '2016')
    ebt = df['total_profit']
    PE_TTM = df['pe_ratio']
    PE_LYR = df['pe_ratio_lyr']
    PB = df['pb_ratio']
    MC = df['circulating_market_cap']
    #获取RWA数据
    bank_ind = get_fundamentals(query(
        bank_indicator.weighted_risky_asset, 
    ).filter(
        bank_indicator.code == security[k]
    ), statDate = '2016')
    rwa = bank_ind['weighted_risky_asset']
    #利用最近5年的数据回归计算β，即银行业股票和大盘之间的相关性
    market_portofolio = get_price('000300.XSHG', start_date='2012-01-01', end_date='2016-12-30', frequency='daily', fields=['close'])['close']
    bank_portofolio = get_price('399387.XSHE', start_date='2012-01-01', end_date='2016-12-30', frequency='daily', fields=['close'])['close']
    market_return =[[0]for i in range(len(market_portofolio)-1)]
    bank_return =[[0] for i in range(len(market_portofolio)-1)]
    for i in range(len(market_portofolio)-1):
        market_return[i] =[(market_portofolio[i+1]-market_portofolio[i])/market_portofolio[i]]
        bank_return[i] = [(bank_portofolio[i+1]-bank_portofolio[i])/bank_portofolio[i]]
    Y = market_return
    X = bank_return
    results = regression.linear_model.OLS(Y, X).fit()
    beta = results.params
    
    #计算几何平均市场收益率,有两种方法，第一种方法是几年增长率的几何平均
    market_re_exp = (market_portofolio.iloc[-1]/market_portofolio.iloc[1])**(0.2)-1
    #第二种方法是每一年的增长率的算数平均
    market_re_one = market_portofolio.iloc[365]/market_portofolio.iloc[1]-1
    market_re_two = market_portofolio.iloc[730]/market_portofolio.iloc[366]-1
    market_re_three = market_portofolio.iloc[1095]/market_portofolio.iloc[731]-1
    market_re_four = market_portofolio.iloc[-1]/market_portofolio.iloc[1096]-1
    #这里我们选用第二种方法，以算数平均作为市场收益率的取值
    market_re = np.mean([market_re_one,market_re_two,market_re_three,market_re_four])
    #取2016年长期国债利率作为无风险利率，为3.01%
    risk_free_return = 0.0301
    #利用CAPM模型，计算COE
    COE = risk_free_return +beta[0]*(market_re-risk_free_return)
    #根据2017银行业监管规定，商业银行资本充足率为10.5%
    CAR = 0.105
    #计算经济利润eva
    eva = ebt - rwa * CAR *COE
    RORAC = ebt/(rwa*CAR)
    
    #以亿元人民币为单位,放入以pandas为数据结构的DF中
    DF.ix[k,0]=eva[0]/10e7
    DF.ix[k,1]=RORAC[0] 
    DF.ix[k,2]=PE_TTM[0]
    DF.ix[k,3]=PE_LYR[0]
    DF.ix[k,4]=PB[0]
    DF.ix[k,5]=MC[0]

#下面开始回归！比如想找经济利润与pe的关系,为了方便使用线性函数拟合，这里先取log
x = log(DF.ix[:,0].values)
X = sm.add_constant(x)
y = log(DF.ix[:,3].values)
model = sm.OLS(y,X)
results = model.fit()
y_fitted = results.fittedvalues

#最后一步！反映在图上
fig, ax = plt.subplots(figsize=(8,6))
ax.plot(x, y, 'o', label='data')
ax.plot(x, y_fitted, 'r--',label='OLS')
ax.legend(loc='best')