|
1
|
|
|
import numpy as np |
|
2
|
|
|
import pandas as pd |
|
3
|
|
|
from six import iteritems |
|
4
|
|
|
from six.moves import zip |
|
5
|
|
|
|
|
6
|
|
|
from zipline.utils.numpy_utils import np_NaT |
|
7
|
|
|
|
|
8
|
|
|
|
|
9
|
|
|
def next_date_frame(dates, announcement_dates): |
|
10
|
|
|
""" |
|
11
|
|
|
Make a DataFrame representing simulated next earnings dates. |
|
12
|
|
|
|
|
13
|
|
|
Parameters |
|
14
|
|
|
---------- |
|
15
|
|
|
dates : pd.DatetimeIndex. |
|
16
|
|
|
The index of the returned DataFrame. |
|
17
|
|
|
announcement_dates : dict[int -> pd.Series] |
|
18
|
|
|
Dict mapping sids to an index of dates on which earnings were announced |
|
19
|
|
|
for that sid. |
|
20
|
|
|
|
|
21
|
|
|
Returns |
|
22
|
|
|
------- |
|
23
|
|
|
next_earnings: pd.DataFrame |
|
24
|
|
|
A DataFrame representing, for each (label, date) pair, the first entry |
|
25
|
|
|
in `earnings_calendars[label]` on or after `date`. Entries falling |
|
26
|
|
|
after the last date in a calendar will have `np_NaT` as the result in |
|
27
|
|
|
the output. |
|
28
|
|
|
|
|
29
|
|
|
See Also |
|
30
|
|
|
-------- |
|
31
|
|
|
previous_earnings_date_frame |
|
32
|
|
|
""" |
|
33
|
|
|
cols = { |
|
34
|
|
|
equity: np.full_like(dates, np_NaT) for equity in announcement_dates |
|
35
|
|
|
} |
|
36
|
|
|
raw_dates = dates.values |
|
37
|
|
|
for equity, earnings_dates in iteritems(announcement_dates): |
|
38
|
|
|
data = cols[equity] |
|
39
|
|
|
if not earnings_dates.index.is_monotonic_increasing: |
|
40
|
|
|
earnings_dates = earnings_dates.sort_index() |
|
41
|
|
|
|
|
42
|
|
|
# Iterate over the raw Series values, since we're comparing against |
|
43
|
|
|
# numpy arrays anyway. |
|
44
|
|
|
iterkv = zip(earnings_dates.index.values, earnings_dates.values) |
|
45
|
|
|
for timestamp, announce_date in iterkv: |
|
46
|
|
|
date_mask = (timestamp <= raw_dates) & (raw_dates <= announce_date) |
|
47
|
|
|
value_mask = (announce_date <= data) | (data == np_NaT) |
|
48
|
|
|
data[date_mask & value_mask] = announce_date |
|
49
|
|
|
|
|
50
|
|
|
return pd.DataFrame(index=dates, data=cols) |
|
51
|
|
|
|
|
52
|
|
|
|
|
53
|
|
|
def previous_date_frame(dates, announcement_dates): |
|
54
|
|
|
""" |
|
55
|
|
|
Make a DataFrame representing simulated next earnings dates. |
|
56
|
|
|
|
|
57
|
|
|
Parameters |
|
58
|
|
|
---------- |
|
59
|
|
|
dates : DatetimeIndex. |
|
60
|
|
|
The index of the returned DataFrame. |
|
61
|
|
|
announcement_dates : dict[int -> DatetimeIndex] |
|
62
|
|
|
Dict mapping sids to an index of dates on which earnings were announced |
|
63
|
|
|
for that sid. |
|
64
|
|
|
|
|
65
|
|
|
Returns |
|
66
|
|
|
------- |
|
67
|
|
|
prev_earnings: pd.DataFrame |
|
68
|
|
|
A DataFrame representing, for (label, date) pair, the first entry in |
|
69
|
|
|
`announcement_dates[label]` strictly before `date`. Entries falling |
|
70
|
|
|
before the first date in a calendar will have `NaT` as the result in |
|
71
|
|
|
the output. |
|
72
|
|
|
|
|
73
|
|
|
See Also |
|
74
|
|
|
-------- |
|
75
|
|
|
next_earnings_date_frame |
|
76
|
|
|
""" |
|
77
|
|
|
sids = list(announcement_dates) |
|
78
|
|
|
out = np.full((len(dates), len(sids)), np_NaT, dtype='datetime64[ns]') |
|
79
|
|
|
dn = dates[-1].asm8 |
|
80
|
|
|
for col_idx, sid in enumerate(sids): |
|
81
|
|
|
# announcement_dates[sid] is Series mapping knowledge_date to actual |
|
82
|
|
|
# announcement date. We don't care about the knowledge date for |
|
83
|
|
|
# computing previous earnings. |
|
84
|
|
|
values = announcement_dates[sid].values |
|
85
|
|
|
values = values[values <= dn] |
|
86
|
|
|
out[dates.searchsorted(values), col_idx] = values |
|
87
|
|
|
|
|
88
|
|
|
frame = pd.DataFrame(out, index=dates, columns=sids) |
|
89
|
|
|
frame.ffill(inplace=True) |
|
90
|
|
|
return frame |
|
91
|
|
|
|
quantopian /
zipline
