-
Notifications
You must be signed in to change notification settings - Fork 86
/
bsz.py
181 lines (151 loc) · 8.08 KB
/
bsz.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
#!/usr/bin/env python
from regress import *
from loaddata import *
from util import *
def wavg(group):
b = group['pbeta']
d = group['log_ret']
w = group['mkt_cap_y'] / 1e6
res = b * ((d * w).sum() / w.sum())
return res
def wavg2(group):
b = group['pbeta']
d = group['cur_log_ret']
w = group['mkt_cap_y'] / 1e6
res = b * ((d * w).sum() / w.sum())
return res
def calc_bsz_daily(daily_df, horizon):
print "Caculating daily bsz..."
result_df = filter_expandable(daily_df)
print "Calculating bsz0..."
result_df['rv'] = result_df['meanBidSize'].astype(float) / result_df['meanAskSize']
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['bsz0'] = result_df['rv'] * result_df['badjret']
result_df['bsz0'] = ((result_df['meanAskSize'] - result_df['meanBidSize']) / (result_df['meanBidSize'] + result_df['meanAskSize'])) / np.sqrt(result_df['spread_bps'])
result_df['bsz0_B'] = winsorize_by_date(result_df[ 'bsz0' ] / 10000)
demean = lambda x: (x - x.mean())
indgroups = result_df[['bsz0_B', 'gdate', 'ind1']].groupby(['gdate', 'ind1'], sort=False).transform(demean)
result_df['bsz0_B_ma'] = indgroups['bsz0_B']
print "Calulating lags..."
for lag in range(1,horizon+1):
shift_df = result_df.unstack().shift(lag).stack()
result_df['bsz'+str(lag)+'_B_ma'] = shift_df['bsz0_B_ma']
result_df['bsz'+str(lag)+'_B'] = shift_df['bsz0_B']
return result_df
def calc_bsz_intra(intra_df):
print "Calculating bsz intra..."
result_df = filter_expandable(intra_df)
print "Calulating bszC..."
result_df['cur_log_ret'] = result_df['overnight_log_ret'] + (np.log(result_df['iclose']/result_df['bopen']))
# result_df['c2c_badj'] = result_df['cur_log_ret'] / result_df['pbeta']
result_df['bret'] = result_df[['cur_log_ret', 'pbeta', 'mkt_cap_y', 'giclose_ts']].groupby(['giclose_ts'], sort=False).apply(wavg2).reset_index(level=0)['pbeta']
result_df['badjret'] = result_df['cur_log_ret'] - result_df['bret']
result_df['rv_i'] = result_df['meanBidSize'].astype(float) / result_df['meanAskSize']
# result_df['bszC'] = result_df['rv_i'] * result_df['badjret']
result_df['bszC'] = ((result_df['meanAskSize'] - result_df['meanBidSize']) / (result_df['meanBidSize'] + result_df['meanAskSize'])) / np.sqrt(result_df['meanSpread'])
result_df['bszC_B'] = winsorize_by_ts(result_df[ 'bszC' ] / 10000)
print "Calulating bszC_ma..."
demean = lambda x: (x - x.mean())
indgroups = result_df[['bszC_B', 'giclose_ts', 'ind1']].groupby(['giclose_ts', 'ind1'], sort=False).transform(demean)
result_df['bszC_B_ma'] = indgroups['bszC_B']
return result_df
def bsz_fits(daily_df, intra_df, horizon, name, middate):
insample_intra_df = intra_df
insample_daily_df = daily_df
outsample_intra_df = intra_df
outsample = False
if middate is not None:
outsample = True
insample_intra_df = intra_df[ intra_df['date'] < middate ]
insample_daily_df = daily_df[ daily_df.index.get_level_values('date') < middate ]
outsample_intra_df = intra_df[ intra_df['date'] >= middate ]
outsample_intra_df['bsz'] = np.nan
outsample_intra_df['bszma'] = np.nan
outsample_intra_df[ 'bszC_B_ma_coef' ] = np.nan
outsample_intra_df[ 'bszC_B_ma_coef' ] = np.nan
outsample_intra_df[ 'bszC_B_coef' ] = np.nan
outsample_intra_df[ 'bszC_B_coef' ] = np.nan
for lag in range(0, horizon+1):
outsample_intra_df[ 'bsz' + str(lag) + '_B_ma_coef' ] = np.nan
outsample_intra_df[ 'bszma' + str(lag) + '_B_coef' ] = np.nan
fits_df = pd.DataFrame(columns=['horizon', 'coef', 'indep', 'tstat', 'nobs', 'stderr'])
fitresults_df = regress_alpha(insample_intra_df, 'bszC_B_ma', horizon, True, 'intra')
fits_df = fits_df.append(fitresults_df, ignore_index=True)
plot_fit(fits_df, "bsz_intra_"+name+"_" + df_dates(insample_intra_df))
fits_df.set_index(keys=['indep', 'horizon'], inplace=True)
unstacked = outsample_intra_df[ ['ticker'] ].unstack()
coefs = dict()
coefs[1] = unstacked.between_time('09:30', '10:31').stack().index
coefs[2] = unstacked.between_time('10:30', '11:31').stack().index
coefs[3] = unstacked.between_time('11:30', '12:31').stack().index
coefs[4] = unstacked.between_time('12:30', '13:31').stack().index
coefs[5] = unstacked.between_time('13:30', '14:31').stack().index
coefs[6] = unstacked.between_time('14:30', '15:59').stack().index
print fits_df.head()
for ii in range(1,7):
outsample_intra_df.ix[ coefs[ii], 'bszC_B_ma_coef' ] = fits_df.ix['bszC_B_ma'].ix[ii].ix['coef']
fits_df = pd.DataFrame(columns=['horizon', 'coef', 'indep', 'tstat', 'nobs', 'stderr'])
for lag in range(1,horizon+1):
fitresults_df = regress_alpha(insample_daily_df, 'bsz0_B_ma', lag, outsample, 'daily')
fits_df = fits_df.append(fitresults_df, ignore_index=True)
plot_fit(fits_df, "bsz_daily_"+name+"_" + df_dates(insample_daily_df))
fits_df.set_index(keys=['indep', 'horizon'], inplace=True)
coef0 = fits_df.ix['bsz0_B_ma'].ix[horizon].ix['coef']
print "Coef0: {}".format(coef0)
for lag in range(1,horizon):
coef = coef0 - fits_df.ix['bsz0_B_ma'].ix[lag].ix['coef']
print "Coef{}: {}".format(lag, coef)
outsample_intra_df[ 'bsz'+str(lag)+'_B_ma_coef' ] = coef
outsample_intra_df['bsz'] = outsample_intra_df['bszC_B_ma'] * outsample_intra_df['bszC_B_ma_coef']
for lag in range(1,horizon):
outsample_intra_df['bsz'] += outsample_intra_df['bsz'+str(lag)+'_B_ma'] * outsample_intra_df['bsz'+str(lag)+'_B_ma_coef']
return outsample_intra_df
def calc_bsz_forecast(daily_df, intra_df, horizon, middate):
daily_results_df = calc_bsz_daily(daily_df, horizon)
forwards_df = calc_forward_returns(daily_df, horizon)
daily_results_df = pd.concat( [daily_results_df, forwards_df], axis=1)
intra_results_df = calc_bsz_intra(intra_df)
intra_results_df = merge_intra_data(daily_results_df, intra_results_df)
full_df = bsz_fits(daily_results_df, intra_results_df, horizon, "", middate)
return full_df
if __name__=="__main__":
parser = argparse.ArgumentParser(description='G')
parser.add_argument("--start",action="store",dest="start",default=None)
parser.add_argument("--end",action="store",dest="end",default=None)
parser.add_argument("--mid",action="store",dest="mid",default=None)
parser.add_argument("--freq",action="store",dest="freq",default=15)
parser.add_argument("--horizon",action="store",dest="horizon",default=3)
args = parser.parse_args()
start = args.start
end = args.end
lookback = 30
horizon = int(args.horizon)
freq = args.freq
pname = "./bsz" + start + "." + end
start = dateparser.parse(start)
end = dateparser.parse(end)
middate = dateparser.parse(args.mid)
loaded = False
try:
daily_df = pd.read_hdf(pname+"_daily.h5", 'table')
intra_df = pd.read_hdf(pname+"_intra.h5", 'table')
loaded = True
except:
print "Did not load cached data..."
if not loaded:
uni_df = get_uni(start, end, lookback)
BARRA_COLS = ['ind1', 'pbeta']
barra_df = load_barra(uni_df, start, end, BARRA_COLS)
PRICE_COLS = ['close', 'overnight_log_ret', 'tradable_volume', 'tradable_med_volume_21']
price_df = load_prices(uni_df, start, end, PRICE_COLS)
BAR_COLS = ['meanAskSize', 'meanBidSize', 'meanSpread', 'bopen', 'spread_bps']
intra_df = load_bars(price_df[ ['ticker'] ], start, end, BAR_COLS, freq)
daily_df = merge_barra_data(price_df, barra_df)
daily_df = merge_intra_eod(daily_df, intra_df)
intra_df = merge_intra_data(daily_df, intra_df)
daily_df.to_hdf(pname+"_daily.h5", 'table', complib='zlib')
intra_df.to_hdf(pname+"_intra.h5", 'table', complib='zlib')
outsample_df = calc_bsz_forecast(daily_df, intra_df, horizon, middate)
dump_alpha(outsample_df, 'bsz')
# dump_all(outsample_df)