siat 3.10.131__py3-none-any.whl → 3.10.132__py3-none-any.whl

build/lib/siat/risk_free_rate.py

@@ -0,0 +1,351 @@
+# -*- coding: utf-8 -*-
+"""
+版权：王德宏，北京外国语大学国际商学院
+功能：
+1、基于股票或股票组合计算无风险收益率
+2、绘制无风险收益率的变化趋势图：日，周，月
+3、与实际的无风险收益率比较
+版本：1.0，2021-2-6
+"""
+
+#==============================================================================
+#关闭所有警告
+import warnings; warnings.filterwarnings('ignore')
+import siat.common as com
+from siat.translate import *
+#==============================================================================
+#==============================================================================
+#==============================================================================
+def compare_rf(rf1,col1,rf2,col2,fromdate,todate,power=0,zeroline=True,twinx=False):
+    """
+    功能：比较两个无风险收益率的时间序列，并绘制趋势线
+    """
+    #检查日期期间的合理性
+    result,start,end=com.check_period(fromdate,todate)
+    if not result:
+        print(" Error(compare_rf): invalide date period from",fromdate,"to",todate)
+        return None
+    
+    #检查并筛选两个无风险收益率的时间序列
+    if rf1 is None:
+        print(" Error(compare_rf): 1st risk-free-rate series is empty")
+        return None        
+    if rf2 is None:
+        print(" Error(compare_rf): 2nd risk-free-rate series is empty")
+        return None  
+    df1a=rf1[rf1.index >= start]
+    df1b=df1a[df1a.index <= end]
+    df1b[col1+'%']=df1b[col1]*100.0
+    
+    df2a=rf2[rf2.index >= start]
+    df2b=df2a[df2a.index <= end]
+    df2b[col2+'%']=df2b[col2]*100.0
+
+    #绘制对比图
+    import siat.grafix as g
+    ticker1=g.ticker_name(df1b['ticker'][0])
+    colname1=col1+'%'
+    label1=col1+'%'
+    ticker2=g.ticker_name(df2b['ticker'][0])
+    colname2=col2+'%'
+    label2=col2+'%'   
+    ylabeltxt='无风险收益率%'
+    titletxt="基于CAPM计算的无风险收益率变化趋势"
+    
+    import datetime; today = datetime.date.today()
+    rf1note=df1b['footnote'][0]
+    rf2note=df2b['footnote'][0]
+    footnote="无风险收益率1："+str(rf1note)+ \
+             "\n无风险收益率2："+str(rf2note)+ \
+             "\n数据来源：雅虎财经, "+str(today)
+    
+    g.plot_line2(df1b,ticker1,colname1,label1, \
+               df2b,ticker2,colname2,label2, \
+               ylabeltxt,titletxt,footnote, \
+               power=power,zeroline=zeroline,twinx=twinx)
+
+    return
+
+if __name__=='__main__':
+    df1=get_rf_capm('AAPL','^GSPC','2018-1-1','2020-12-31',window=40)
+    df2=get_rf_capm('MSFT','^GSPC','2018-1-1','2020-12-31',window=40)
+    compare_rf(df1,'Rf',df2,'Rf','2019-1-1','2019-12-31')
+    compare_rf(df1,'Rf',df2,'Rf','2019-1-1','2019-12-31',twinx=True)
+
+    df1['Rf_20']=df1['Rf'].rolling(window=20).mean()
+    compare_rf(df1,'Rf',df1,'Rf_20','2019-1-1','2019-12-31')
+
+#==============================================================================
+
+def calc_rolling_cumret(dfc,col,period='Weekly'):
+    """
+    传入日收益率col的数据表dfc
+    传出不同期间的累计收益率序列cumret
+    """
+    df=dfc.copy()
+    #检查period类型
+    periodlist = ["Weekly","Biweekly","Monthly","Quarterly","Semiannual","Annual"]
+    if not (period in periodlist):
+        print("  Error(calc_rolling_cumret): only supports:",periodlist)
+        return None
+
+    #换算期间对应的实际交易天数
+    perioddays=[5,10,20,60,120,240]
+    rollingnum=perioddays[periodlist.index(period)]      
+    
+    import numpy as np
+    df['logdret']=np.log(df[col]+1)
+    df['cumret']=np.exp(df['logdret'].rolling(rollingnum).sum())-1.0
+    
+    return df['cumret']
+
+#==============================================================================
+
+if __name__ =="__main__":
+    col='Rf'
+    limits=[0.01,0.01]
+    
+def winsor(df,col,limits=[0.01,0.01]):
+    """
+    功能：对于数据表df1中的列col进行下1%（参数1）和上1%（参数2）的处理
+    """
+    import numpy as np
+    from scipy.stats.mstats import winsorize
+
+    a = np.array(df[col])
+    aw=winsorize(a,limits=limits)
+    
+    return aw.data
+#==============================================================================
+if __name__=='__main__':
+    ticker='399001.SZ'
+    mktidx='000300.SS'
+    fromdate='2020-1-1'
+    todate='2020-12-31'
+    window=240
+
+def get_rf_capm(ticker,mktidx,fromdate,todate,window=240,sharelist=[]):
+    """
+    功能：计算无风险收益率的时间序列
+    ticker：股票或股票组合
+    mktidx：股票市场指数
+    fromdate：开始时间
+    todate：截止时间
+    window：每次回归的样本个数
+    sharelist：第一个参数为投资组合的持股比例，默认为等权重；为单个股票时无用
+    """
+    #检查日期期间的合理性
+    result,start,end=com.check_period(fromdate,todate)
+    if not result:
+        print(" Error(get_rf_capm): invalide date period from",fromdate,"to",todate)
+        return None
+
+    #提前开始日期，留出回归窗口
+    start1=com.date_adjust(start,adjust=-window*2)
+    
+    #获得股票或股票组合的历史收益率
+    import siat.security_prices as sp
+    #单个股票情形
+    if isinstance(ticker,str):
+        spdf=sp.get_prices(ticker,start1,end)
+    #股票组合情形
+    if isinstance(ticker,list):
+        if sharelist == []:
+            num=len(ticker)
+            sharelist=[1]*num
+        spdf=sp.get_prices_portfolio(ticker,sharelist,start1,end)
+    if spdf is None:
+        print("  Error(get_rf_capm): info not found or unavailable for",ticker)
+        return None
+    
+    spdf['Stock_dailyRet']=spdf['Close'].pct_change()  
+    
+    #获取市场指数的历史收益率
+    rmdf=sp.get_prices(mktidx,start1,end)
+    if rmdf is None:
+        print("  Error(get_rf_capm): info not found or unavailable for",mktidx)
+        return None    
+    rmdf['Market_dailyRet']=rmdf['Close'].pct_change()  
+    
+    #合并股票（组合）与市场指数的收益率为一个数据集
+    import pandas as pd
+    df=pd.merge(rmdf['Market_dailyRet'],spdf['Stock_dailyRet'],how='inner',left_index=True,right_index=True)
+    df['Date']=df.index.strftime("%Y-%m-%d")
+    
+    datelist_ts=list(df.index)    
+    datelist=list(df['Date'])
+    fromdate=start.strftime("%Y-%m-%d")
+    #start_pos=lookup_datelist(datelist,fromdate,direction='more')
+    todate=end.strftime("%Y-%m-%d")
+    #end_pos=lookup_datelist(datelist,todate,direction='less')
+    
+    #用于滚动的日期期间
+    datelist_rolling=[]
+    for d in datelist:
+        if (d >= fromdate) and (d <= todate):
+            datelist_rolling=datelist_rolling+[d]
+    
+    #滚动回归
+    if sharelist == []:
+        #footnote=[ticker,mktidx,fromdate,todate,window]
+        footnote=[ticker,mktidx,window]
+    else:
+        #footnote=[ticker,mktidx,fromdate,todate,window,sharelist]
+        footnote=[ticker,mktidx,window,sharelist]
+    betas=pd.DataFrame(columns=('date','Beta','alpha','R-sqr','p-value','sig','Rf','ticker','footnote'))    
+    from scipy import stats
+    import numpy as np
+    
+    for d in datelist_rolling:
+        pos2=lookup_datelist(datelist,d)
+        pos1=pos2 - window
+        sdate2=datelist_ts[pos2]
+        sdate1=datelist_ts[pos1]
+
+        sampledf=df[df.index >= sdate1].copy()
+        sampledf=sampledf[sampledf.index < sdate2]
+        
+        (beta,alpha,r_value,p_value,std_err)= \
+            stats.linregress(sampledf['Market_dailyRet'],sampledf['Stock_dailyRet'])
+        sig=sig_stars(p_value)
+        try:
+            rf=alpha/(1-beta)
+        except:
+            rf=np.nan    
+        
+        row=pd.Series({'date':sdate2,'Beta':beta,'alpha':alpha, \
+            'R-sqr':r_value**2,'p-value':p_value,'sig':sig,'Rf':rf,'ticker':ticker,'footnote':footnote})
+        try:
+            betas=betas.append(row,ignore_index=True)
+        except:
+            betas=betas._append(row,ignore_index=True)
+    
+    betas.set_index('date',inplace=True)
+    
+    return betas
+
+if __name__=='__main__':   
+    df1=get_rf_capm('399001.SZ','000001.SS','2020-1-1','2020-12-31',window=240)
+    df1['Rf'].plot()
+
+    df2=get_rf_capm('000001.SS','399001.SZ','2020-1-1','2020-12-31',window=240)
+    df2['Rf'].plot()
+
+    df3=get_rf_capm('000001.SS','000300.SS','2020-1-1','2020-12-31',window=240)
+    df3['Rf'].plot()    
+
+    df4=get_rf_capm('AAPL','^GSPC','2020-1-1','2020-12-31',window=240)
+    df4['Rf'].plot()  
+
+    df5=get_rf_capm('^DJI','^GSPC','2020-1-1','2020-12-31',window=240)
+    df5['Rf'].plot()  
+#==============================================================================
+if __name__=='__main__':   
+    datelist=['2019-12-30','2019-12-31','2020-01-02','2020-01-03','2020-01-06']
+    adate='2020-01-01'
+    adate='2020-01-04'
+    direction='less'
+    direction='more'
+    
+def lookup_datelist(datelist,adate,direction='less'):
+    """
+    功能：在日期列表datelist查找与日期adate最接近日期的位置
+    direction='more'：若无匹配的日期，则往日期增加的方向查找最接近日期的位置
+    direction='less'：若无匹配的日期，则往日期减少的方向查找最接近日期的位置
+    """
+    i=0
+    found=False
+    while not found:
+        try:
+            pos=datelist.index(adate)
+        except:
+            if direction == 'more':
+                i=i+1
+            else:
+                i=i-1
+            adate=com.date_adjust(adate,adjust=i)
+        else:
+            found=True    
+
+    return pos
+
+#==============================================================================
+if __name__=='__main__':   
+    p_value=0.07
+    p_value=0.02
+    p_value=0.0009
+    criteria='accounting'
+    criteria='financial'
+    
+def sig_stars(p_value,criteria='accounting'):
+    """
+    功能：基于p_value给出星号的个数
+    p_value：显著性水平
+    criteria='accounting'：默认的显著性基准，<0.1为一颗星；若为'financial'，<0.05为一颗星
+    """
+    sig=''
+    if criteria == 'accounting':
+        if p_value < 0.1: sig='*'*1
+        if p_value < 0.05: sig='*'*2
+        if p_value < 0.01: sig='*'*3
+    else:
+        if p_value < 0.05: sig='*'*1
+        if p_value < 0.01: sig='*'*2
+        if p_value < 0.001: sig='*'*3        
+
+    return sig
+
+if __name__=='__main__': 
+    sig_stars(0.1)
+    sig_stars(0.07)
+    sig_stars(0.04)
+    sig_stars(0.009)
+
+#==============================================================================
+def calc_capm_rf(rmdf,rdf):
+    """
+    功能：CAPM回归
+    返回：截距项，贝塔系数，无风险收益率
+    """
+    #OLS回归
+    from scipy import stats 
+    (beta,alpha,r_value,p_value,std_err)=stats.linregress(rmdf,rdf)    
+    
+    rf=alpha/(1-beta)
+    
+    return [beta,alpha,r_value,p_value,std_err,rf]
+
+#==============================================================================
+#==============================================================================
+#==============================================================================
+#==============================================================================
+if __name__=='__main__': 
+    start='2018-1-1'
+    end='2020-12-31'
+    scope='US'
+    freq='daily'
+
+def get_rf_kfdl(start,end,scope='US',freq='daily'):
+    """
+    功能：从Kenneth R. French's Data Library获得无风险收益率
+    start/end：日期期间
+    scope：国家/地区，支持美国/北美/欧洲/日本/不含日本的亚太/不含美国的全球。全球
+    freq：支持日/月/年收益率，其中美国还支持周数据
+    返回：无风险收益率，市场收益率
+    """
+    import siat.fama_french as ff
+    factor='FF3'
+    df=ff.get_ff_factors(start,end,scope,factor,freq)
+    
+    df['Market_dailyRet']=df['Mkt-RF']+df['RF']
+    df['Rf']=df['RF']
+    df['ticker']=scope
+    footnote=[scope,freq]
+    df['footnote']=df['ticker'].apply(lambda x:footnote)
+    
+    df1=df[['Market_dailyRet','Rf','ticker','footnote']]
+    
+    return df1
+
+
+#==============================================================================