forked from noterminusgit/statarb
-
Notifications
You must be signed in to change notification settings - Fork 0
/
prod_sal.py
213 lines (167 loc) · 10.2 KB
/
prod_sal.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
#!/usr/bin/env python
from regress import *
from load_data_live import *
from loaddata import *
from util import *
from pandas.stats.moments import ewma
ESTIMATE = "SAL"
def wavg(group):
b = group['pbeta']
d = group['log_ret']
w = group['mkt_cap_y'] / 1e6
print "Mkt return: {} {}".format(group['gdate'], ((d * w).sum() / w.sum()))
res = b * ((d * w).sum() / w.sum())
return res
def calc_sal_daily(daily_df, horizon):
print "Caculating daily sal..."
result_df = filter_expandable(daily_df)
print "Calculating sal0..."
halflife = horizon / 2
# result_df['dk'] = np.exp( -1.0 * halflife * (result_df['gdate'] - result_df['last']).astype('timedelta64[D]').astype(int) )
result_df['bret'] = result_df[['log_ret', 'pbeta', 'mkt_cap_y', 'gdate']].groupby('gdate').apply(wavg).reset_index(level=0)['pbeta']
result_df['badjret'] = result_df['log_ret'] - result_df['bret']
result_df['badj0_B'] = winsorize_by_date(result_df[ 'badjret' ])
result_df['cum_ret'] = pd.rolling_sum(result_df['log_ret'], horizon)
print result_df[ESTIMATE + '_diff_mean'].describe()
result_df['std_diff'] = result_df[ESTIMATE + '_std'].unstack().diff().stack()
result_df.loc[ result_df['std_diff'] <= 0, ESTIMATE + '_diff_mean'] = 0
result_df['sal0'] = result_df[ESTIMATE + '_diff_mean'] / result_df[ESTIMATE + '_median']
# print result_df.columns
# result_df['sum'] = result_df['SAL_median']
# result_df['det_diff'] = (result_df['sum'].diff())
# result_df['det_diff_sum'] = pd.rolling_sum( result_df['det_diff'], window=2)
# #result_df['det_diff_dk'] = ewma(result_df['det_diff'], halflife=horizon )
# result_df['sal0'] = result_df['det_diff']
# result_df['median'] = -1.0 * (result_df['median'] - 3)
# result_df['med_diff'] = result_df['median'].unstack().diff().stack()
# result_df['med_diff_dk'] = pd.rolling_sum( result_df['dk'] * result_df['med_diff'], window=horizon )
# result_df['sal0'] = (np.sign(result_df['med_diff_dk']) * np.sign(result_df['cum_ret'])).clip(lower=0) * result_df['med_diff_dk']
# demean = lambda x: (x - x.mean())
# indgroups = result_df[['sal0', 'gdate', 'ind1']].groupby(['gdate', 'ind1'], sort=True).transform(demean)
# result_df['sal0_ma'] = indgroups['sal0']
# result_df['sal0_ma'] = result_df['sal0_ma'] - result_df['sal0_ma'].dropna().mean()
# result_df['sal0_ma'] = result_df['sal0_ma'] * (np.sign(result_df['sal0_ma']) * np.sign(result_df['cum_ret']))
result_df['sal0_ma'] = result_df['sal0']
for lag in range(1,horizon+1):
shift_df = result_df.unstack().shift(lag).stack()
result_df['sal'+str(lag)+'_ma'] = shift_df['sal0_ma']
return result_df
def generate_coefs(daily_df, horizon, name, coeffile=None, intercepts=None):
insample_daily_df = daily_df
insample_up_df = insample_daily_df[ insample_daily_df[ESTIMATE + "_diff_mean"] > 0 ]
fits_df = pd.DataFrame(columns=['horizon', 'coef', 'indep', 'tstat', 'nobs', 'stderr', 'intercept'])
for ii in range(1, horizon+1):
fitresults_df = regress_alpha(insample_up_df, 'sal0_ma', ii, False, 'daily', True)
print "INTERCEPT {} {}".format(ii, intercepts[ii])
fitresults_df['intercept'] = fitresults_df['intercept'] - float(intercepts[ii])
fits_df = fits_df.append(fitresults_df, ignore_index=True)
fits_df.set_index(keys=['indep', 'horizon'], inplace=True)
coef0 = fits_df.ix['sal0_ma'].ix[horizon].ix['coef']
intercept0 = fits_df.ix['sal0_ma'].ix[horizon].ix['intercept']
coef_list = list()
coef_list.append( { 'name': 'tgt0_ma_coef', 'group': "up", 'coef': coef0 } )
coef_list.append( { 'name': 'tgt0_ma_intercept', 'group': 'up', 'coef': intercept0 } )
print "Coef{}: {}".format(0, coef0)
for lag in range(1,horizon):
coef = coef0 - fits_df.ix['sal0_ma'].ix[lag].ix['coef']
intercept = intercept0 - fits_df.ix['sal0_ma'].ix[lag].ix['intercept']
print "Coef{}: {}".format(lag, coef)
coef_list.append( { 'name': 'sal' + str(lag) + '_ma_coef', 'group': "up", 'coef': coef } )
coef_list.append( { 'name': 'sal' + str(lag) + '_ma_intercept', 'group': "up", 'coef': intercept } )
insample_dn_df = insample_daily_df[ insample_daily_df[ESTIMATE + "_diff_mean"] <= 0 ]
fits_df = pd.DataFrame(columns=['horizon', 'coef', 'indep', 'tstat', 'nobs', 'stderr', 'intercept'])
for ii in range(1, horizon+1):
fitresults_df = regress_alpha(insample_dn_df, 'sal0_ma', ii, False, 'daily', True)
fitresults_df['intercept'] = fitresults_df['intercept'] - intercepts[ii]
fits_df = fits_df.append(fitresults_df, ignore_index=True)
fits_df.set_index(keys=['indep', 'horizon'], inplace=True)
coef0 = fits_df.ix['sal0_ma'].ix[horizon].ix['coef']
intercept0 = fits_df.ix['sal0_ma'].ix[horizon].ix['intercept']
coef_list.append( { 'name': 'tgt0_ma_coef', 'group': "dn", 'coef': coef0 } )
coef_list.append( { 'name': 'tgt0_ma_intercept', 'group': 'dn', 'coef': intercept0 } )
print "Coef{}: {}".format(0, coef0)
for lag in range(1,horizon):
coef = coef0 - fits_df.ix['sal0_ma'].ix[lag].ix['coef']
intercept = intercept0 - fits_df.ix['sal0_ma'].ix[lag].ix['intercept']
print "Coef{}: {}".format(lag, coef)
coef_list.append( { 'name': 'sal' + str(lag) + '_ma_coef', 'group': "dn", 'coef': coef } )
coef_list.append( { 'name': 'sal' + str(lag) + '_ma_intercept', 'group': "dn", 'coef': intercept } )
coef_df = pd.DataFrame(coef_list)
coef_df.to_csv(coeffile)
return
def sal_alpha(daily_df, horizon, name, coeffile):
coef_df = pd.read_csv(coeffile, header=0, index_col=['name', 'group'])
outsample_daily_df = daily_df
outsample_daily_df['sal'] = 0.0
coef0 = coef_df.ix['sal0_ma_coef'].ix["up"].ix['coef']
intercept0 = coef_df.ix['sal0_ma_intercept'].ix["up"].ix['coef']
print "Coef{}: {}".format(0, coef0)
outsample_daily_df.loc[ outsample_daily_df[ESTIMATE + '_diff_mean'] > 0, 'sal0_ma_coef' ] = coef0
outsample_daily_df.loc[ outsample_daily_df[ESTIMATE + '_diff_mean'] > 0, 'sal0_ma_intercept' ] = intercept0
for lag in range(1,horizon):
coef = coef_df.ix['sal' + str(lag) + '0_ma_coef'].ix["up"].ix['coef']
intercept = coef_df.ix['sal' + str(lag) + '_ma_intercept'].ix["up"].ix['coef']
outsample_daily_df.loc[ outsample_daily_df[ESTIMATE + '_diff_mean'] > 0, 'sal'+str(lag)+'_ma_coef' ] = coef
outsample_daily_df.loc[ outsample_daily_df[ESTIMATE + '_diff_mean'] > 0, 'sal'+str(lag)+'_ma_intercept' ] = intercept
coef0 = coef_df.ix['sal0_ma_coef'].ix["dn"].ix['coef']
intercept0 = coef_df.ix['sal0_ma_intercept'].ix["dn"].ix['coef']
print "Coef{}: {}".format(0, coef0)
outsample_daily_df.loc[ outsample_daily_df[ESTIMATE + '_diff_mean'] <= 0, 'sal0_ma_coef' ] = coef0
outsample_daily_df.loc[ outsample_daily_df[ESTIMATE + '_diff_mean'] <= 0, 'sal0_ma_intercept' ] = intercept0
for lag in range(1,horizon):
coef = coef_df.ix['sal' + str(lag) + '0_ma_coef'].ix["dn"].ix['coef']
intercept = coef_df.ix['sal' + str(lag) + '_ma_intercept'].ix["dn"].ix['coef']
print "Coef{}: {}".format(lag, coef)
outsample_daily_df.loc[ outsample_daily_df[ESTIMATE + '_diff_mean'] <= 0, 'sal'+str(lag)+'_ma_coef' ] = coef
outsample_daily_df.loc[ outsample_daily_df[ESTIMATE + '_diff_mean'] <= 0, 'sal'+str(lag)+'_ma_intercept' ] = intercept
outsample_daily_df[ 'sal' ] = (outsample_daily_df['sal0_ma'].fillna(0) * outsample_daily_df['sal0_ma_coef'] + outsample_daily_df['sal0_ma_intercept']).fillna(0)
for lag in range(1,horizon):
outsample_daily_df[ 'sal'] += (outsample_daily_df['sal'+str(lag)+'_ma'].fillna(0) * outsample_daily_df['sal'+str(lag)+'_ma_coef'] + outsample_daily_df['sal'+str(lag)+'_ma_intercept']).fillna(0)
return outsample_daily_df
def calc_sal_forecast(daily_df, horizon, coeffile, fit):
daily_results_df = calc_sal_daily(daily_df, horizon)
if fit:
forwards_df = calc_forward_returns(daily_df, horizon)
daily_results_df = pd.concat( [daily_results_df, forwards_df], axis=1)
intercept_d = get_intercept(daily_results_df, horizon, 'sal0_ma')
generate_coefs( daily_results_df, horizon, "all", coeffile, intercept_d)
return
else:
res = sal_alpha( daily_results_df, horizon, "all", coeffile)
result_df = pd.concat([res], verify_integrity=True)
return result_df
if __name__=="__main__":
parser = argparse.ArgumentParser(description='G')
parser.add_argument("--asof",action="store",dest="asof",default=None)
parser.add_argument("--inputfile",action="store",dest="inputfile",default=None)
parser.add_argument("--outputfile",action="store",dest="outputfile",default=None)
parser.add_argument("--logfile",action="store",dest="logfile",default=None)
parser.add_argument("--coeffile",action="store",dest="coeffile",default=None)
parser.add_argument("--fit",action="store",dest="fit",default=False)
args = parser.parse_args()
horizon = int(20)
end = datetime.strptime(args.asof, "%Y%m%d")
if args.fit:
print "Fitting..."
coeffile = args.coeffile + "/" + args.asof + ".sal.csv"
lookback = timedelta(days=720)
start = end - lookback
uni_df = get_uni(start, end, 30)
else:
print "Not fitting..."
coeffile = args.coeffile
lookback = timedelta(days=horizon+5)
start = end - lookback
uni_df = load_live_file(args.inputfile)
end = datetime.strptime(args.asof + '_' + uni_df['time'].min(), '%Y%m%d_%H:%M:%S')
print "Running between {} and {}".format(start, end)
BARRA_COLS = ['ind1', 'pbeta']
barra_df = load_barra(uni_df, start, end, BARRA_COLS)
PRICE_COLS = ['close']
price_df = load_prices(uni_df, start, end, PRICE_COLS)
daily_df = merge_barra_data(price_df, barra_df)
analyst_df = load_estimate_hist(price_df[['ticker']], start, end, "SAL")
daily_df = merge_daily_calcs(analyst_df, daily_df)
result_df = calc_sal_forecast(daily_df, horizon, coeffile, args.fit)
if not args.fit:
dump_prod_alpha(result_df, 'sal', args.outputfile)