6장 파마-프렌치 모델¶

6.5.1 팩터 데이터 구하기¶

In [1]:

import pandas_datareader.data as web  # module for reading datasets directly from the web
import pandas_datareader.famafrench as ff 

In [2]:

datasets = ff.get_available_datasets()
print('No. of datasets:{0}'.format(len(datasets)))

No. of datasets:297

In [3]:

print(datasets)

['F-F_Research_Data_Factors', 'F-F_Research_Data_Factors_weekly', 'F-F_Research_Data_Factors_daily', 'F-F_Research_Data_5_Factors_2x3', 'F-F_Research_Data_5_Factors_2x3_daily', 'Portfolios_Formed_on_ME', 'Portfolios_Formed_on_ME_Wout_Div', 'Portfolios_Formed_on_ME_Daily', 'Portfolios_Formed_on_BE-ME', 'Portfolios_Formed_on_BE-ME_Wout_Div', 'Portfolios_Formed_on_BE-ME_Daily', 'Portfolios_Formed_on_OP', 'Portfolios_Formed_on_OP_Wout_Div', 'Portfolios_Formed_on_OP_Daily', 'Portfolios_Formed_on_INV', 'Portfolios_Formed_on_INV_Wout_Div', 'Portfolios_Formed_on_INV_Daily', '6_Portfolios_2x3', '6_Portfolios_2x3_Wout_Div', '6_Portfolios_2x3_weekly', '6_Portfolios_2x3_daily', '25_Portfolios_5x5', '25_Portfolios_5x5_Wout_Div', '25_Portfolios_5x5_Daily', '100_Portfolios_10x10', '100_Portfolios_10x10_Wout_Div', '100_Portfolios_10x10_Daily', '6_Portfolios_ME_OP_2x3', '6_Portfolios_ME_OP_2x3_Wout_Div', '6_Portfolios_ME_OP_2x3_daily', '25_Portfolios_ME_OP_5x5', '25_Portfolios_ME_OP_5x5_Wout_Div', '25_Portfolios_ME_OP_5x5_daily', '100_Portfolios_ME_OP_10x10', '100_Portfolios_10x10_ME_OP_Wout_Div', '100_Portfolios_ME_OP_10x10_daily', '6_Portfolios_ME_INV_2x3', '6_Portfolios_ME_INV_2x3_Wout_Div', '6_Portfolios_ME_INV_2x3_daily', '25_Portfolios_ME_INV_5x5', '25_Portfolios_ME_INV_5x5_Wout_Div', '25_Portfolios_ME_INV_5x5_daily', '100_Portfolios_ME_INV_10x10', '100_Portfolios_10x10_ME_INV_Wout_Div', '100_Portfolios_ME_INV_10x10_daily', '25_Portfolios_BEME_OP_5x5', '25_Portfolios_BEME_OP_5x5_Wout_Div', '25_Portfolios_BEME_OP_5x5_daily', '25_Portfolios_BEME_INV_5x5', '25_Portfolios_BEME_INV_5x5_Wout_Div', '25_Portfolios_BEME_INV_5x5_daily', '25_Portfolios_OP_INV_5x5', '25_Portfolios_OP_INV_5x5_Wout_Div', '25_Portfolios_OP_INV_5x5_daily', '32_Portfolios_ME_BEME_OP_2x4x4', '32_Portfolios_ME_BEME_OP_2x4x4_Wout_Div', '32_Portfolios_ME_BEME_INV_2x4x4', '32_Portfolios_ME_BEME_INV_2x4x4_Wout_Div', '32_Portfolios_ME_OP_INV_2x4x4', '32_Portfolios_ME_OP_INV_2x4x4_Wout_Div', 'Portfolios_Formed_on_E-P', 'Portfolios_Formed_on_E-P_Wout_Div', 'Portfolios_Formed_on_CF-P', 'Portfolios_Formed_on_CF-P_Wout_Div', 'Portfolios_Formed_on_D-P', 'Portfolios_Formed_on_D-P_Wout_Div', '6_Portfolios_ME_EP_2x3', '6_Portfolios_ME_EP_2x3_Wout_Div', '6_Portfolios_ME_CFP_2x3', '6_Portfolios_ME_CFP_2x3_Wout_Div', '6_Portfolios_ME_DP_2x3', '6_Portfolios_ME_DP_2x3_Wout_Div', 'F-F_Momentum_Factor', 'F-F_Momentum_Factor_daily', '6_Portfolios_ME_Prior_12_2', '6_Portfolios_ME_Prior_12_2_Daily', '25_Portfolios_ME_Prior_12_2', '25_Portfolios_ME_Prior_12_2_Daily', '10_Portfolios_Prior_12_2', '10_Portfolios_Prior_12_2_Daily', 'F-F_ST_Reversal_Factor', 'F-F_ST_Reversal_Factor_daily', '6_Portfolios_ME_Prior_1_0', '6_Portfolios_ME_Prior_1_0_Daily', '25_Portfolios_ME_Prior_1_0', '25_Portfolios_ME_Prior_1_0_Daily', '10_Portfolios_Prior_1_0', '10_Portfolios_Prior_1_0_Daily', 'F-F_LT_Reversal_Factor', 'F-F_LT_Reversal_Factor_daily', '6_Portfolios_ME_Prior_60_13', '6_Portfolios_ME_Prior_60_13_Daily', '25_Portfolios_ME_Prior_60_13', '25_Portfolios_ME_Prior_60_13_Daily', '10_Portfolios_Prior_60_13', '10_Portfolios_Prior_60_13_Daily', 'Portfolios_Formed_on_AC', '25_Portfolios_ME_AC_5x5', 'Portfolios_Formed_on_BETA', '25_Portfolios_ME_BETA_5x5', 'Portfolios_Formed_on_NI', '25_Portfolios_ME_NI_5x5', 'Portfolios_Formed_on_VAR', '25_Portfolios_ME_VAR_5x5', 'Portfolios_Formed_on_RESVAR', '25_Portfolios_ME_RESVAR_5x5', '5_Industry_Portfolios', '5_Industry_Portfolios_Wout_Div', '5_Industry_Portfolios_daily', '10_Industry_Portfolios', '10_Industry_Portfolios_Wout_Div', '10_Industry_Portfolios_daily', '12_Industry_Portfolios', '12_Industry_Portfolios_Wout_Div', '12_Industry_Portfolios_daily', '17_Industry_Portfolios', '17_Industry_Portfolios_Wout_Div', '17_Industry_Portfolios_daily', '30_Industry_Portfolios', '30_Industry_Portfolios_Wout_Div', '30_Industry_Portfolios_daily', '38_Industry_Portfolios', '38_Industry_Portfolios_Wout_Div', '38_Industry_Portfolios_daily', '48_Industry_Portfolios', '48_Industry_Portfolios_Wout_Div', '48_Industry_Portfolios_daily', '49_Industry_Portfolios', '49_Industry_Portfolios_Wout_Div', '49_Industry_Portfolios_daily', 'ME_Breakpoints', 'BE-ME_Breakpoints', 'OP_Breakpoints', 'INV_Breakpoints', 'E-P_Breakpoints', 'CF-P_Breakpoints', 'D-P_Breakpoints', 'Prior_2-12_Breakpoints', 'Developed_3_Factors', 'Developed_3_Factors_Daily', 'Developed_ex_US_3_Factors', 'Developed_ex_US_3_Factors_Daily', 'Europe_3_Factors', 'Europe_3_Factors_Daily', 'Japan_3_Factors', 'Japan_3_Factors_Daily', 'Asia_Pacific_ex_Japan_3_Factors', 'Asia_Pacific_ex_Japan_3_Factors_Daily', 'North_America_3_Factors', 'North_America_3_Factors_Daily', 'Developed_5_Factors', 'Developed_5_Factors_Daily', 'Developed_ex_US_5_Factors', 'Developed_ex_US_5_Factors_Daily', 'Europe_5_Factors', 'Europe_5_Factors_Daily', 'Japan_5_Factors', 'Japan_5_Factors_Daily', 'Asia_Pacific_ex_Japan_5_Factors', 'Asia_Pacific_ex_Japan_5_Factors_Daily', 'North_America_5_Factors', 'North_America_5_Factors_Daily', 'Developed_Mom_Factor', 'Developed_Mom_Factor_Daily', 'Developed_ex_US_Mom_Factor', 'Developed_ex_US_Mom_Factor_Daily', 'Europe_Mom_Factor', 'Europe_Mom_Factor_Daily', 'Japan_Mom_Factor', 'Japan_Mom_Factor_Daily', 'Asia_Pacific_ex_Japan_MOM_Factor', 'Asia_Pacific_ex_Japan_MOM_Factor_Daily', 'North_America_Mom_Factor', 'North_America_Mom_Factor_Daily', 'Developed_6_Portfolios_ME_BE-ME', 'Developed_6_Portfolios_ME_BE-ME_daily', 'Developed_ex_US_6_Portfolios_ME_BE-ME', 'Developed_ex_US_6_Portfolios_ME_BE-ME_daily', 'Europe_6_Portfolios_ME_BE-ME', 'Europe_6_Portfolios_ME_BE-ME_daily', 'Japan_6_Portfolios_ME_BE-ME', 'Japan_6_Portfolios_ME_BE-ME_daily', 'Asia_Pacific_ex_Japan_6_Portfolios_ME_BE-ME', 'Asia_Pacific_ex_Japan_6_Portfolios_ME_BE-ME_daily', 'North_America_6_Portfolios_ME_BE-ME', 'North_America_6_Portfolios_ME_BE-ME_daily', 'Developed_25_Portfolios_ME_BE-ME', 'Developed_25_Portfolios_ME_BE-ME_daily', 'Developed_ex_US_25_Portfolios_ME_BE-ME', 'Developed_ex_US_25_Portfolios_ME_BE-ME_daily', 'Europe_25_Portfolios_ME_BE-ME', 'Europe_25_Portfolios_ME_BE-ME_daily', 'Japan_25_Portfolios_ME_BE-ME', 'Japan_25_Portfolios_ME_BE-ME_daily', 'Asia_Pacific_ex_Japan_25_Portfolios_ME_BE-ME', 'Asia_Pacific_ex_Japan_25_Portfolios_ME_BE-ME_daily', 'North_America_25_Portfolios_ME_BE-ME', 'North_America_25_Portfolios_ME_BE-ME_daily', 'Developed_6_Portfolios_ME_OP', 'Developed_6_Portfolios_ME_OP_Daily', 'Developed_ex_US_6_Portfolios_ME_OP', 'Developed_ex_US_6_Portfolios_ME_OP_Daily', 'Europe_6_Portfolios_ME_OP', 'Europe_6_Portfolios_ME_OP_Daily', 'Japan_6_Portfolios_ME_OP', 'Japan_6_Portfolios_ME_OP_Daily', 'Asia_Pacific_ex_Japan_6_Portfolios_ME_OP', 'Asia_Pacific_ex_Japan_6_Portfolios_ME_OP_Daily', 'North_America_6_Portfolios_ME_OP', 'North_America_6_Portfolios_ME_OP_Daily', 'Developed_25_Portfolios_ME_OP', 'Developed_25_Portfolios_ME_OP_Daily', 'Developed_ex_US_25_Portfolios_ME_OP', 'Developed_ex_US_25_Portfolios_ME_OP_Daily', 'Europe_25_Portfolios_ME_OP', 'Europe_25_Portfolios_ME_OP_Daily', 'Japan_25_Portfolios_ME_OP', 'Japan_25_Portfolios_ME_OP_Daily', 'Asia_Pacific_ex_Japan_25_Portfolios_ME_OP', 'Asia_Pacific_ex_Japan_25_Portfolios_ME_OP_Daily', 'North_America_25_Portfolios_ME_OP', 'North_America_25_Portfolios_ME_OP_Daily', 'Developed_6_Portfolios_ME_INV', 'Developed_6_Portfolios_ME_INV_Daily', 'Developed_ex_US_6_Portfolios_ME_INV', 'Developed_ex_US_6_Portfolios_ME_INV_Daily', 'Europe_6_Portfolios_ME_INV', 'Europe_6_Portfolios_ME_INV_Daily', 'Japan_6_Portfolios_ME_INV', 'Japan_6_Portfolios_ME_INV_Daily', 'Asia_Pacific_ex_Japan_6_Portfolios_ME_INV', 'Asia_Pacific_ex_Japan_6_Portfolios_ME_INV_Daily', 'North_America_6_Portfolios_ME_INV', 'North_America_6_Portfolios_ME_INV_Daily', 'Developed_25_Portfolios_ME_INV', 'Developed_25_Portfolios_ME_INV_Daily', 'Developed_ex_US_25_Portfolios_ME_INV', 'Developed_ex_US_25_Portfolios_ME_INV_Daily', 'Europe_25_Portfolios_ME_INV', 'Europe_25_Portfolios_ME_INV_Daily', 'Japan_25_Portfolios_ME_INV', 'Japan_25_Portfolios_ME_INV_Daily', 'Asia_Pacific_ex_Japan_25_Portfolios_ME_INV', 'Asia_Pacific_ex_Japan_25_Portfolios_ME_INV_Daily', 'North_America_25_Portfolios_ME_INV', 'North_America_25_Portfolios_ME_INV_Daily', 'Developed_6_Portfolios_ME_Prior_12_2', 'Developed_6_Portfolios_ME_Prior_250_20_daily', 'Developed_ex_US_6_Portfolios_ME_Prior_12_2', 'Developed_ex_US_6_Portfolios_ME_Prior_250_20_daily', 'Europe_6_Portfolios_ME_Prior_12_2', 'Europe_6_Portfolios_ME_Prior_250_20_daily', 'Japan_6_Portfolios_ME_Prior_12_2', 'Japan_6_Portfolios_ME_Prior_250_20_daily', 'Asia_Pacific_ex_Japan_6_Portfolios_ME_Prior_12_2', 'Asia_Pacific_ex_Japan_6_Portfolios_ME_Prior_250_20_daily', 'North_America_6_Portfolios_ME_Prior_12_2', 'North_America_6_Portfolios_ME_Prior_250_20_daily', 'Developed_25_Portfolios_ME_Prior_12_2', 'Developed_25_Portfolios_ME_Prior_250_20_daily', 'Developed_ex_US_25_Portfolios_ME_Prior_12_2', 'Developed_ex_US_25_Portfolios_ME_Prior_250_20_daily', 'Europe_25_Portfolios_ME_Prior_12_2', 'Europe_25_Portfolios_ME_Prior_250_20_daily', 'Japan_25_Portfolios_ME_Prior_12_2', 'Japan_25_Portfolios_ME_Prior_250_20_daily', 'Asia_Pacific_ex_Japan_25_Portfolios_ME_Prior_12_2', 'Asia_Pacific_ex_Japan_25_Portfolios_ME_Prior_250_20_daily', 'North_America_25_Portfolios_ME_Prior_12_2', 'North_America_25_Portfolios_ME_Prior_250_20_daily', 'Developed_32_Portfolios_ME_BE-ME_OP_2x4x4', 'Developed_ex_US_32_Portfolios_ME_BE-ME_OP_2x4x4', 'Europe_32_Portfolios_ME_BE-ME_OP_2x4x4', 'Japan_32_Portfolios_ME_BE-ME_OP_2x4x4', 'Asia_Pacific_ex_Japan_32_Portfolios_ME_BE-ME_OP_2x4x4', 'North_America_32_Portfolios_ME_BE-ME_OP_2x4x4', 'Developed_32_Portfolios_ME_BE-ME_INV(TA)_2x4x4', 'Developed_ex_US_32_Portfolios_ME_BE-ME_INV(TA)_2x4x4', 'Europe_32_Portfolios_ME_BE-ME_INV(TA)_2x4x4', 'Japan_32_Portfolios_ME_BE-ME_INV(TA)_2x4x4', 'Asia_Pacific_ex_Japan_32_Portfolios_ME_BE-ME_INV(TA)_2x4x4', 'North_America_32_Portfolios_ME_BE-ME_INV(TA)_2x4x4', 'Developed_32_Portfolios_ME_INV(TA)_OP_2x4x4', 'Developed_ex_US_32_Portfolios_ME_INV(TA)_OP_2x4x4', 'Europe_32_Portfolios_ME_INV(TA)_OP_2x4x4', 'Japan_32_Portfolios_ME_INV(TA)_OP_2x4x4', 'Asia_Pacific_ex_Japan_32_Portfolios_ME_INV(TA)_OP_2x4x4', 'North_America_32_Portfolios_ME_INV(TA)_OP_2x4x4', 'Emerging_5_Factors', 'Emerging_MOM_Factor', 'Emerging_Markets_6_Portfolios_ME_BE-ME', 'Emerging_Markets_6_Portfolios_ME_OP', 'Emerging_Markets_6_Portfolios_ME_INV', 'Emerging_Markets_6_Portfolios_ME_Prior_12_2', 'Emerging_Markets_4_Portfolios_BE-ME_OP', 'Emerging_Markets_4_Portfolios_OP_INV', 'Emerging_Markets_4_Portfolios_BE-ME_INV']

10개의 인더스트리 포트폴리오에 대한 데이터셋 얻기¶

In [4]:

df_10_industry = [dataset for dataset in datasets if '10' in dataset and 'Industry' in dataset]
print(df_10_industry)

['10_Industry_Portfolios', '10_Industry_Portfolios_Wout_Div', '10_Industry_Portfolios_daily']

In [22]:

ds_industry = web.DataReader(df_10_industry[0],'famafrench',start='2017-06-23',end='2019-11-01') # Taking [0] as extracting '10_Industry_Portfolios'

In [23]:

print(type(ds_industry))
ds_industry.keys()

<class 'dict'>

Out[23]:

dict_keys([0, 1, 2, 3, 4, 5, 6, 7, 'DESCR'])

In [24]:

print(ds_industry['DESCR'])

10 Industry Portfolios
----------------------

This file was created by CMPT_IND_RETS using the 202009 CRSP database. It contains value- and equal-weighted returns for 10 industry portfolios. The portfolios are constructed at the end of June. The annual returns are from January to December. Missing data are indicated by -99.99 or -999. Copyright 2020 Kenneth R. French

  0 : Average Value Weighted Returns -- Monthly (30 rows x 10 cols)
  1 : Average Equal Weighted Returns -- Monthly (30 rows x 10 cols)
  2 : Average Value Weighted Returns -- Annual (3 rows x 10 cols)
  3 : Average Equal Weighted Returns -- Annual (3 rows x 10 cols)
  4 : Number of Firms in Portfolios (30 rows x 10 cols)
  5 : Average Firm Size (30 rows x 10 cols)
  6 : Sum of BE / Sum of ME (3 rows x 10 cols)
  7 : Value-Weighted Average of BE/ME (3 rows x 10 cols)

In [25]:

ds_industry[0].head()

Out[25]:

	NoDur	Durbl	Manuf	Enrgy	HiTec	Telcm	Shops	Hlth	Utils	Other
Date
2017-06	-0.98	3.51	1.28	-0.21	-2.11	-2.27	-1.90	5.56	-1.89	4.18
2017-07	-0.12	-1.12	2.16	2.11	3.78	5.30	0.06	0.53	2.96	1.44
2017-08	-1.71	-0.16	0.17	-5.17	3.04	-2.56	-1.66	1.79	2.15	-0.40
2017-09	-0.33	5.29	4.21	11.14	0.58	-1.68	2.50	1.50	-1.93	4.31
2017-10	0.10	1.20	2.85	0.29	6.86	-5.74	2.69	-2.24	3.05	2.03

5개 팩터 데이터셋¶

In [ ]:

df_5_factor = [dataset for dataset in datasets if '5' in dataset and 'Factor' in dataset]
print(df_5_factor)
ds_factors = web.DataReader(df_5_factor[0],'famafrench',start='2017-06-23',end='2019-11-01') # Taking [0] as extracting 1F-F-Research_Data_Factors_2x3')
print('\nKEYS\n{0}'.format(ds_factors.keys()))
print('DATASET DESCRIPTION \n {0}'.format(ds_factors['DESCR']))
ds_factors[0].head()

6.5.2 펀드 수익률과 팩터 데이터 회귀분석¶

In [66]:

import pandas_datareader.data as web
import pandas_datareader.famafrench as ff
import pandas as pd

# 이용가능한 데이터셋을 구한다
datasets = ff.get_available_datasets()

In [67]:

# 월간 3팩터(datasets[0])를 구한다
df_3_factor = datasets[0]
print(df_3_factor)

F-F_Research_Data_Factors

In [68]:

df_3_factor

Out[68]:

'F-F_Research_Data_Factors'

In [ ]:

# 1980년 2월부터 2019년 6월 팩터 데이터를 다운로드한다
ds_factors = web.DataReader(df_3_factor,'famafrench', start='1980-02-01', end='2019-06-30')
print('\nKEYS\n{0}'.format(ds_factors.keys()))
print('DATASET DESCRIPTION \n {0}'.format(ds_factors['DESCR']))

In [70]:

print(ds_factors)

{0:          Mkt-RF   SMB   HML    RF
Date                             
1980-02   -1.22 -1.82  0.62  0.89
1980-03  -12.90 -6.64 -1.06  1.21
1980-04    3.97  0.97  1.06  1.26
1980-05    5.26  2.16  0.39  0.81
1980-06    3.06  1.67 -0.89  0.61
...         ...   ...   ...   ...
2019-02    3.40  2.10 -2.72  0.18
2019-03    1.10 -3.06 -4.20  0.19
2019-04    3.96 -1.77  2.00  0.21
2019-05   -6.94 -1.25 -2.15  0.21
2019-06    6.93  0.20 -0.82  0.18

[473 rows x 4 columns], 1:       Mkt-RF    SMB    HML     RF
Date                             
1980   22.13   5.57 -25.06  11.24
1981  -18.13   7.23  25.01  14.71
1982   10.66   8.89  13.59  10.54
1983   13.74  13.67  20.85   8.80
1984   -6.05  -8.31  19.63   9.85
1985   24.91   0.12   1.35   7.72
1986   10.12  -9.60   9.58   6.16
1987   -3.87 -11.00  -1.64   5.47
1988   11.55   5.90  14.77   6.35
1989   20.49 -12.80  -4.35   8.37
1990  -13.95 -14.17  -9.66   7.81
1991   29.18  16.14 -14.67   5.60
1992    6.23   7.55  24.62   3.51
1993    8.21   6.42  16.99   2.90
1994   -4.10  -0.40  -0.97   3.90
1995   31.22 -10.23   6.59   5.60
1996   15.96  -4.29   8.27   5.21
1997   25.96  -8.20  19.35   5.26
1998   19.46 -25.62  -9.64   4.86
1999   20.57  15.16 -32.18   4.68
2000  -17.60  -5.37  46.33   5.89
2001  -15.20  17.90  19.14   3.83
2002  -22.76   3.88   8.92   1.65
2003   30.75  26.11   4.21   1.02
2004   10.72   4.12   7.82   1.20
2005    3.09  -2.41   9.44   2.98
2006   10.60   0.12  11.95   4.80
2007    1.04  -6.98 -17.08   4.66
2008  -38.34   2.65   1.31   1.60
2009   28.26   9.46  -9.18   0.10
2010   17.37  13.85  -5.47   0.12
2011    0.44  -5.87  -8.07   0.04
2012   16.27  -1.25  10.01   0.06
2013   35.20   6.78   2.54   0.02
2014   11.71  -7.95  -1.54   0.02
2015    0.08  -3.92  -9.73   0.02
2016   13.30   6.59  22.42   0.20
2017   21.51  -4.90 -13.70   0.80
2018   -6.93  -3.35  -9.22   1.81
2019   28.28  -6.32 -10.88   2.14, 'DESCR': 'F-F Research Data Factors\n-------------------------\n\nThis file was created by CMPT_ME_BEME_RETS using the 202009 CRSP database. The 1-month TBill return is from Ibbotson and Associates, Inc. Copyright 2020 Kenneth R. French\n\n  0 : (473 rows x 4 cols)\n  1 : Annual Factors: January-December (40 rows x 4 cols)'}

In [71]:

# 나중에 구할 펀드수익률과 합치기 위해 인덱스 종류를 변경한다
#ds_factors[0].index=ds_factors[0].index.strftime('%Y-%m')

# ds_3_factors에 저장한다
ds_3_factors = ds_factors[0]

In [72]:

from pandas_datareader import data as web
import pandas as pd

In [73]:

# Fidelity Contrafund Fund (FCNTX) 
ticker = "FCNTX" 

In [74]:

# 1980년 1월부터 2019년 6월까지 펀드 월간(interval='m') 과거 주가 중 수정주가()['Adj Close'])를 구한다.
pxclose = web.get_data_yahoo(ticker, start='1980-01-01',end='2019-06-30',interval='m')['Adj Close']

In [75]:

# 앞서 팩터 자료와 합치기 위해 인덱스 종류를 변경한다
pxclose.index=pxclose.index.strftime('%Y-%m')

In [76]:

# 수익률을 계산한다
ret_data = pxclose.pct_change()[1:]

In [77]:

ret_data = pd.DataFrame(ret_data)
ret_data.columns = ['portfolio']
ret_data

Out[77]:

	portfolio
Date
1980-02	-0.016874
1980-03	-0.089431
1980-04	0.017857
1980-05	0.078948
1980-06	0.011743
...	...
2019-02	0.014938
2019-03	0.031163
2019-04	0.048800
2019-05	-0.057208
2019-06	0.066343

473 rows × 1 columns

In [78]:

print(ret_data.head(), type(ret_data))
print(ds_3_factors.head(), type(ds_3_factors))

         portfolio
Date              
1980-02  -0.016874
1980-03  -0.089431
1980-04   0.017857
1980-05   0.078948
1980-06   0.011743 <class 'pandas.core.frame.DataFrame'>
         Mkt-RF   SMB   HML    RF
Date                             
1980-02   -1.22 -1.82  0.62  0.89
1980-03  -12.90 -6.64 -1.06  1.21
1980-04    3.97  0.97  1.06  1.26
1980-05    5.26  2.16  0.39  0.81
1980-06    3.06  1.67 -0.89  0.61 <class 'pandas.core.frame.DataFrame'>

In [79]:

# 회귀분석을 위해 두 개의 데이터프레임(ret_data, ds_3_factors)을 합친다
# 두 프레임의 인덱스를 기준으로 합치는 경우 매개변수 left_index와 right_index를 True로 한다.
# 또 합치려는 두 프레임의 컬럼 데이터나 인덱스끼리 서로 일치하는 데이터만 합치는 경우 how 매개변수를 'inner'로 한다.

regress_data=ret_data.merge(ds_3_factors, how = 'inner', left_index= True, right_index= True)
regress_data

Out[79]:

	portfolio	Mkt-RF	SMB	HML	RF
Date
1980-02	-0.016874	-1.22	-1.82	0.62	0.89
1980-03	-0.089431	-12.90	-6.64	-1.06	1.21
1980-04	0.017857	3.97	0.97	1.06	1.26
1980-05	0.078948	5.26	2.16	0.39	0.81
1980-06	0.011743	3.06	1.67	-0.89	0.61
...	...	...	...	...	...
2019-02	0.014938	3.40	2.10	-2.72	0.18
2019-03	0.031163	1.10	-3.06	-4.20	0.19
2019-04	0.048800	3.96	-1.77	2.00	0.21
2019-05	-0.057208	-6.94	-1.25	-2.15	0.21
2019-06	0.066343	6.93	0.20	-0.82	0.18

473 rows × 5 columns

In [80]:

# "Mkt-RF" 컬럼 이름을 "mkt_excess"으로 변경한다
regress_data.rename(columns={"Mkt-RF":"mkt_excess"}, inplace=True)
# 초과수익율(펀드수익률-무위험수익률)을 계산하고 'port_excess' 컬럼을 만들어 저장한다
regress_data['port_excess'] = regress_data['portfolio'] - regress_data['RF']
print(regress_data.tail())

         portfolio  mkt_excess   SMB   HML    RF  port_excess
Date                                                         
2019-02   0.014938        3.40  2.10 -2.72  0.18    -0.165062
2019-03   0.031163        1.10 -3.06 -4.20  0.19    -0.158837
2019-04   0.048800        3.96 -1.77  2.00  0.21    -0.161200
2019-05  -0.057208       -6.94 -1.25 -2.15  0.21    -0.267208
2019-06   0.066343        6.93  0.20 -0.82  0.18    -0.113657

In [85]:

import statsmodels.api as smf

In [92]:

model = smf.formula.ols(formula = "port_excess ~ mkt_excess + SMB + HML", data = regress_data).fit()

In [94]:

print(model.params)

Intercept    -0.311119
mkt_excess    0.008059
SMB           0.010100
HML           0.005635
dtype: float64

In [90]:

print(model.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:            port_excess   R-squared:                       0.005
Model:                            OLS   Adj. R-squared:                 -0.001
Method:                 Least Squares   F-statistic:                    0.8470
Date:                Thu, 12 Nov 2020   Prob (F-statistic):              0.469
Time:                        05:20:39   Log-Likelihood:                -472.48
No. Observations:                 473   AIC:                             953.0
Df Residuals:                     469   BIC:                             969.6
Df Model:                           3                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
Intercept     -0.3111      0.031    -10.058      0.000      -0.372      -0.250
mkt_excess     0.0081      0.007      1.110      0.268      -0.006       0.022
SMB            0.0101      0.011      0.944      0.346      -0.011       0.031
HML            0.0056      0.011      0.508      0.612      -0.016       0.027
==============================================================================
Omnibus:                      830.537   Durbin-Watson:                   1.671
Prob(Omnibus):                  0.000   Jarque-Bera (JB):           432316.055
Skew:                          10.778   Prob(JB):                         0.00
Kurtosis:                     149.530   Cond. No.                         4.78
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.