CAPM 实证数据和代码

'''import tushare as ts pro = ts.pro_api('a492ea4daxxxxxxxxxxxxxxxxxxx12f4a655e21') # 获取shibor（1个月）数据作为无风险收益率 shibor_history=pro.shibor(start_date='20170101',end_date='20211231',fields='date,1m') shibor_history.to_csv('E:\\python_outputs\\CAPM\\shibor_history.csv',index=False) # 获取沪深300指数月线数据 HS300_history=pro.index_monthly(ts_code='000300.SH',start_date='20161201',end_date='20211231', fields='ts_code,trade_date,close') HS300_history.to_csv('E:\\python_outputs\\CAPM\\HS300_history.csv',index=False) # 获取沪深300指数成份股（2021年12月31日） cons_HS300=pro.index_weight(index_code='000300.SH',trade_date='20211231', fields='con_code,trade_date') cons_HS300.to_csv('E:\\python_outputs\\CAPM\\cons_HS300.csv',index=False) # 获取沪深300所有成份股的月线数据 # 抱歉，您每分钟最多访问该接口120次 import time for i in cons_HS300.truncate(before=0,after=99).con_code: con=pro.monthly(ts_code=i,start_date='20161201',end_date='20211231',fields='ts_code,trade_date,close') con.to_csv('E:\\python_outputs\\CAPM\\cons_monthly\\{}_monthly.csv'.format(i),index=False) time.sleep(61) # 等待61秒 for i in cons_HS300.truncate(before=100,after=199).con_code: con=pro.monthly(ts_code=i,start_date='20161201',end_date='20211231',fields='ts_code,trade_date,close') con.to_csv('E:\\python_outputs\\CAPM\\cons_monthly\\{}_monthly.csv'.format(i),index=False) time.sleep(61) # 等待61秒 for i in cons_HS300.truncate(before=200,after=299).con_code: con=pro.monthly(ts_code=i,start_date='20161201',end_date='20211231',fields='ts_code,trade_date,close') con.to_csv('E:\\python_outputs\\CAPM\\cons_monthly\\{}_monthly.csv'.format(i),index=False)''' # ======================================== ^ Step1 数据获取 ^ ================================================= # import glob import numpy as np import pandas as pd HS300=pd.read_csv('E:\\python_outputs\\CAPM\\HS300_history.csv',parse_dates=['trade_date']) HS300.trade_date=HS300.trade_date.apply(lambda x:x.strftime('%Y%m')) # 日期转为年月格式 # print(HS300) df=pd.read_csv('E:\\python_outputs\\CAPM\\shibor_history.csv',parse_dates=['date'],index_col='date') df0=(df.resample('M') # 调整周期，以月为周期 .mean() .div(12) # shibor利率为单利年利，直接出除以12 .reset_index()) df0.date=df0.date.apply(lambda x:x.strftime('%Y%m')) shibor=df0.rename(columns={'date':'trade_date','1m':'rf_shibor'}).set_index('trade_date') shibor.to_csv('E:\\python_outputs\\CAPM\\rf_shibor.csv') # print(shibor) files=glob.glob('E:\\python_outputs\\CAPM\\cons_monthly\\*.csv') cons_monthly=pd.concat([pd.read_csv(j,parse_dates=['trade_date']) for j in files]) cons_monthly.trade_date=cons_monthly.trade_date.apply(lambda x:x.strftime('%Y%m')) # print(cons_monthly) data=(pd.concat([cons_monthly,HS300]) .set_index('trade_date') .pivot(columns='ts_code',values='close') .dropna(axis=1) .apply(np.log) .diff() .dropna() .mul(100)) regression_data_set=(data.T-shibor.T.loc['rf_shibor']).T # regression_data_set为成分股和沪深300的历史超额收益率 regression_data_set.to_csv('E:\\python_outputs\\CAPM\\regression_data_set.csv') # print(regression_data_set) # ======================================== ^ Step2 数据整理 ^ ================================================= # import statsmodels.api as sm import matplotlib.pyplot as plt regression_data_set=pd.read_csv('E:\\python_outputs\\CAPM\\regression_data_set.csv',index_col='trade_date') print(regression_data_set) X=sm.add_constant(regression_data_set['000300.SH']) params_list=[] r2_adj_list=[] cons=regression_data_set.columns for a in cons: y=regression_data_set[a] model=sm.OLS(y,X) result=model.fit() param = result.params.to_list() r2_adj = result.rsquared_adj params_list.append(param) r2_adj_list.append(r2_adj) r2_adj = pd.DataFrame(r2_adj_list, index=cons, columns=['rsquared_adj']) params = pd.DataFrame(params_list, columns=['const', '000300.SH'], index=cons) regression_result = pd.concat([params, r2_adj], axis=1) # print(regression_result) regression_result.to_csv('E:\\python_outputs\\CAPM\\regression_result.csv') # 个例展示（贵州茅台） X1=sm.add_constant(regression_data_set['000300.SH']) y1=regression_data_set['600519.SH'] model1=sm.OLS(y1,X1) result1=model1.fit() print(result1.summary()) plt.rcParams['font.sans-serif'] = ['SimHei'] plt.rcParams['axes.unicode_minus'] = False regression_data_set1=regression_data_set.assign(observes_month=range(1,61)) plt.figure(figsize=(9,5)) plt.plot(regression_data_set1.observes_month,[0]*60,color='black') plt.plot(regression_data_set1.observes_month,regression_data_set['600519.SH'],label='贵州茅台',linestyle='--') plt.plot(regression_data_set1.observes_month,regression_data_set['000300.SH'],label='沪深300') plt.title('沪深300和贵州茅台的超额收益') plt.xlabel('观察样本(月)') plt.ylabel('超额收益(%)') plt.legend() plt.savefig('E:\\python_outputs\\CAPM\\沪深300和贵州茅台的超额收益.png') X1=regression_data_set['000300.SH'] Y1=result1.predict() plt.figure(figsize=(9,6)) plt.scatter(X1,y1,label='观测值') plt.plot(X1,Y1,label='预测值',color='orange') plt.title('贵州茅台证券特征线') plt.xlabel('沪深300超额收益(%)',position=(1,1)) plt.ylabel('贵州茅台超额收益(%)',position=(-5,0.95),rotation=0) plt.legend() ax=plt.gca() ax.spines['right'].set_color('none') ax.spines['top'].set_color('none') ax.spines['bottom'].set_position(('data',0)) ax.spines['left'].set_position(('data',0)) plt.savefig('E:\\python_outputs\\CAPM\\贵州茅台证券特征线.png') # plt.show() # ===================================== ^ Step3 一阶回归和个例（贵州茅台） ^ ====================================== # regression_result=pd.read_csv('E:\\python_outputs\\CAPM\\regression_result.csv',index_col='Unnamed: 0') regression_result.rename(columns={'000300.SH':'beta'},inplace=True) # print(regression_result) df1=regression_data_set.div(100).add(1).apply(np.cumprod) # print(df1) mean_return=df1.loc[202112].apply(lambda s:np.power(s,1/60)).sub(1).mul(100).rename('mean_Return') X2=sm.add_constant(regression_result['beta']) y2=mean_return # print(y2) model2=sm.OLS(y2,X2) result2=model2.fit() print(result2.summary()) # print(df1.loc[202112]) # print(mean_return) s=regression_data_set.mean() # print(s) rf=pd.read_csv('E:\\python_outputs\\CAPM\\rf_shibor.csv',index_col='trade_date') mean_rf=rf.rf_shibor.mean() print('mean_rf:','{:.4f}'.format(mean_rf)) X3=regression_result['beta'] Y3=pd.Series(result2.predict()).add(mean_rf) y3=y2.add(mean_rf) # print(y3) plt.figure(figsize=(6,6)) plt.scatter(X3,y3,label='observes') plt.plot(X3,Y3,label='CML',color='orange') plt.title('证券市场线') plt.xlabel('beta',position=(1,1)) plt.ylabel('E(R)',position=(-5,0.95),rotation=0) plt.legend() ax=plt.gca() ax.spines['right'].set_color('none') ax.spines['top'].set_color('none') ax.spines['bottom'].set_position(('data',0)) ax.spines['left'].set_position(('data',0)) plt.savefig('E:\\python_outputs\\CAPM\\证券市场线.png') plt.show() # ======================================== ^ Step2 二阶回归 ^ ================================================= #