-
Notifications
You must be signed in to change notification settings - Fork 9
/
Copy path2a_calculate_mom_port_betas.py
291 lines (248 loc) · 10.6 KB
/
2a_calculate_mom_port_betas.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
# ======================================================================================================================
# Replicate Barberis, Jin, and Wang (2021)
# Part 2a: calculate momemtum portfolio betas
#
# Author: Gen Li
# 03/14/2021
#
# ======================================================================================================================
import pandas as pd
import os
import wrds
import numpy as np
from fuzzywuzzy import fuzz
import sqlite3
import glob
from pandas.tseries.offsets import *
from scipy import stats
from time import process_time
# import modin.pandas as pd
from distributed import Client
client = Client()
import statsmodels.api as sm
# Directory set up
project_dir = "/Users/genli/Dropbox/UBC/Course/2020 Term2/COMM 673/COMM673_paper_replica" # Change to your project directory
data_folder = project_dir + "/data"
os.chdir(project_dir + "/_temp")
#%%
# ======================================================================================================================
# Part 1: Merge portfolio and CRSP daily return data
# ======================================================================================================================
umd = pd.read_pickle("momentum_10_portfolio.pkl")
crsp_d = pd.read_pickle(data_folder + "/CRSP_d_19620701_20151231.pkl")
factor = pd.read_csv(data_folder + "/F-F_Research_Data_Factors_daily.csv", skiprows=3)
# Clean factor data
factor = factor.iloc[:-1, :]
factor.columns = ['date_new', 'Mkt_RF', 'SMB', 'HML', 'RF']
factor['date_new'] = pd.to_datetime(factor['date_new'], format='%Y%m%d')
for c in ['Mkt_RF', 'SMB', 'HML', 'RF']:
factor[c] = pd.to_numeric(factor[c])
factor[c] = factor[c] / 100
# Convert permno data format
umd['permno'] = umd['permno'].astype(int)
umd['permno'].isnull().sum()
crsp_d['permno'] = crsp_d['permno'].astype(int)
crsp_d['permno'].isnull().sum()
# Add factor data to crsp
crsp_d['date_new'] = pd.to_datetime(crsp_d.date)
crsp_d = crsp_d.merge(factor, how='left', on=['date_new'])
crsp_d['ret_rf'] = crsp_d.ret - crsp_d.RF
# Create umd one year window variable
umd['one_year_start'] = umd.hdate1
umd['one_year_end'] = umd['one_year_start'] + pd.Timedelta("365D")
umd['one_year_start_date'] = umd['one_year_start'].dt.date
umd['one_year_end_date'] = umd['one_year_end'].dt.date
# Create group number for each combination of permno and date
umd['group_num'] = umd.groupby(['permno', 'date']).ngroup()
# # Get news through SQL
# conn = sqlite3.connect(':memory:')
# umd.to_sql('umd', conn, index=False, if_exists="replace")
# crsp_d.to_sql('crsp_d', conn, index=False, if_exists="replace")
#
# qry = '''
# select
# a.permno, a.date, a.momr, a.momr_bp, a.hdate1, a.hdate2, b.*
# from
# umd a, crsp_d b
# where
# a.permno = b.permno and (b.date between a.one_year_start_date and a.one_year_end_date)
# '''
# umd_1Y_daily_ret = pd.read_sql_query(qry, conn)
# umd_1Y_daily_ret.to_pickle("umd_1Y_daily_ret.pkl")
#%%
# ======================================================================================================================
# Part 2: Calculate beta for each portfolio-month
# ======================================================================================================================
group_groups = np.int(np.floor(umd.group_num.max() / 10000) + 1)
for g in range(group_groups):
print("==========================================")
print("I AM REGRESSING GROUP " + str(g))
print("==========================================")
if g != group_groups - 1:
start_ind = g * 10000
end_ind = (g + 1) * 10000
cri = (umd.group_num >= start_ind) & (umd.group_num <= end_ind)
umd_sub = umd.loc[cri].copy()
crsp_d_sub = crsp_d.loc[crsp_d.permno.isin(umd_sub.permno)].copy()
else:
start_ind = g * 10000
cri = (umd.group_num >= start_ind)
umd_sub = umd.loc[cri].copy()
crsp_d_sub = crsp_d.loc[crsp_d.permno.isin(umd_sub.permno)].copy()
# Get news through SQL
conn = sqlite3.connect(':memory:')
umd_sub.to_sql('umd', conn, index=False, if_exists="replace")
crsp_d_sub.to_sql('crsp_d', conn, index=False, if_exists="replace")
qry = '''
select
a.group_num, a.permno, a.date, a.momr, a.momr_bp, a.hdate1, a.hdate2, a.one_year_start_date, a.one_year_end_date, b.ret, b.ret_rf, b.Mkt_RF
from
umd a, crsp_d b
where
a.permno = b.permno and (b.date between a.one_year_start_date and a.one_year_end_date)
'''
umd_1Y_daily_ret = pd.read_sql_query(qry, conn)
# umd_1Y_daily_ret.to_pickle("umd_1Y_daily_ret.pkl")
def get_beta_alpha(df):
try:
X = sm.add_constant(df['Mkt_RF'])
Y = df['ret_rf']
result = sm.OLS(Y, X, missing='drop').fit()
result.params
output = pd.Series({'group_num': df.group_num.iloc[1]})
output = output.append(result.params)
except:
output = pd.Series({'group_num': df.group_num.iloc[1], 'const':np.NaN, 'Mkt_RF':np.NaN})
# return result.params.get('const'), result.params.get('Mkt_RF')
return output
beta_alpha = umd_1Y_daily_ret.groupby('group_num').apply(get_beta_alpha)
# beta_alpha.to_pickle('beta_alpha_group_' + str(g) + '.pkl')
#%%
# ======================================================================================================================
# Part 3: Aggregate beta for all portfolio-month
# ======================================================================================================================
beta_files = glob.glob('beta_alpha_group_*')
# Aggregate all beta output files
beta_alpha_all = pd.DataFrame()
for f in beta_files:
temp = pd.read_pickle(f)
beta_alpha_all = beta_alpha_all.append(temp, ignore_index=True)
beta_alpha_all = beta_alpha_all.sort_values('group_num')
beta_alpha_all = beta_alpha_all.reset_index(drop=True)
# Merge with UMD dataset
beta_alpha_all = beta_alpha_all.merge(umd, how='right', on=['group_num'])
beta_alpha_all = beta_alpha_all.drop_duplicates()
# Export
beta_alpha_all.to_pickle(data_folder + '/beta_alpha_all.pkl')
#%%
# ======================================================================================================================
# Part 4: Calculate average beta by mom group and month
# ======================================================================================================================
beta_alpha_all = pd.read_pickle(data_folder + '/beta_alpha_all.pkl')
# Average beta of each momr portfolio-month
avg_beta_momr = beta_alpha_all.groupby(['momr','date']).Mkt_RF.mean().to_frame('avg_beta').reset_index()
temp = beta_alpha_all.groupby(['momr','date']).Mkt_RF.count().to_frame('num_stock').reset_index()
avg_beta_momr = avg_beta_momr.merge(temp, how='left', on=['momr','date'])
# Average beta of each momr_bp port.-month
avg_beta_momr_bp = beta_alpha_all.groupby(['momr_bp','date']).Mkt_RF.mean().to_frame('avg_beta_bp').reset_index()
temp = beta_alpha_all.groupby(['momr_bp','date']).Mkt_RF.count().to_frame('num_stock_bp').reset_index()
avg_beta_momr_bp = avg_beta_momr_bp.merge(temp, how='left', on=['momr_bp','date'])
# Aggregate two results
avg_beta_momr_all = avg_beta_momr.merge(avg_beta_momr_bp, how='left', left_on=['momr','date'], right_on=['momr_bp','date'])
# Export
avg_beta_momr_all.to_pickle(data_folder + '/avg_beta_mom_port.pkl')
#%% Parallel processing
# group_groups = np.int(np.floor(umd.group_num.max() / 10000) + 1)
#
# umd_split = np.array_split(umd, group_groups)
#
# def single_work(g, umd, crsp_d):
# # crsp_d_sub = crsp_d.loc[crsp_d.permno.isin(umd_sub.permno)].copy()
#
# if g != group_groups - 1:
# start_ind = g * 10000
# end_ind = (g + 1) * 10000
#
# cri = (umd.group_num >= start_ind) & (umd.group_num <= end_ind)
# umd_sub = umd.loc[cri].copy()
# crsp_d_sub = crsp_d.loc[crsp_d.permno.isin(umd_sub.permno)].copy()
#
# else:
# start_ind = g * 10000
#
# cri = (umd.group_num >= start_ind)
# umd_sub = umd.loc[cri].copy()
# crsp_d_sub = crsp_d.loc[crsp_d.permno.isin(umd_sub.permno)].copy()
#
# # Get news through SQL
# conn = sqlite3.connect(':memory:')
# umd_sub.to_sql('umd', conn, index=False, if_exists="replace")
# crsp_d_sub.to_sql('crsp_d', conn, index=False, if_exists="replace")
#
# qry = '''
# select
# a.group_num, a.permno, a.date, a.momr, a.momr_bp, a.hdate1, a.hdate2, a.one_year_start_date, a.one_year_end_date, b.ret, b.ret_rf, b.Mkt_RF
# from
# umd a, crsp_d b
# where
# a.permno = b.permno and (b.date between a.one_year_start_date and a.one_year_end_date)
# '''
# umd_1Y_daily_ret = pd.read_sql_query(qry, conn)
# # umd_1Y_daily_ret.to_pickle("umd_1Y_daily_ret.pkl")
#
# def get_beta_alpha(df):
# try:
# X = sm.add_constant(df['Mkt_RF'])
# Y = df['ret_rf']
#
# result = sm.OLS(Y, X, missing='drop').fit()
# result.params
# output = pd.Series({'group_num': df.group_num.iloc[1]})
# output = output.append(result.params)
# except:
# output = pd.Series({'group_num': df.group_num.iloc[1], 'const':np.NaN, 'Mkt_RF':np.NaN})
#
# # return result.params.get('const'), result.params.get('Mkt_RF')
# return output
#
#
# beta_alpha = umd_1Y_daily_ret.groupby('group_num').apply(get_beta_alpha)
# beta_alpha.to_pickle('beta_alpha_group_' + str(g) + '.pkl')
#
# return beta_alpha
#
#
# # import multiprocessing as mp
# # from multiprocessing import Pool
# # from functools import partial
# # cores = mp.cpu_count()
# # pool = Pool(cores)
# # # for n, frame in enumerate(pool.imap(single_work, (umd_split,crsp_d)), start=1):
# # # frame.to_pickle('{}'.format(n))
# # # pool.close()
# # # pool.join()
# #
# # if __name__ == '__main__':
# # N= mp.cpu_count()
# #
# # with mp.Pool(processes = N) as p:
# # prod_x = partial(single_work, crsp_d=crsp_d)
# # results = p.map(prod_x, umd_split[:2])
# # results.to_pickle("beta_alpha_result_all.pkl")
#
# Pros = []
# all_reg_results = pd.DataFrame()
# from multiprocessing import Process
# import single_work
# def main():
# for i in range(3):
# print("Thread Started")
# p = Process(target=single_work, args=(i, umd, crsp_d))
# Pros.append(p)
# p.start()
#
# # block until all the threads finish (i.e. block until all function_x calls finish)
# for t in Pros:
# t.join()
#
# main()