|
1
|
|
|
""" |
|
2
|
|
|
Reference implementation for EarningsCalendar loaders. |
|
3
|
|
|
""" |
|
4
|
|
|
from numpy import full_like |
|
5
|
|
|
import pandas as pd |
|
6
|
|
|
from six import iteritems |
|
7
|
|
|
|
|
8
|
|
|
from zipline.utils.memoize import lazyval |
|
9
|
|
|
|
|
10
|
|
|
from .base import PipelineLoader |
|
11
|
|
|
from .frame import DataFrameLoader |
|
12
|
|
|
from ..data.earnings import EarningsCalendar |
|
13
|
|
|
|
|
14
|
|
|
|
|
15
|
|
|
class EarningsCalendarLoader(PipelineLoader): |
|
16
|
|
|
""" |
|
17
|
|
|
Reference loader for `zipline.pipeline.data.earnings.EarningsCalendar`. |
|
18
|
|
|
|
|
19
|
|
|
Does not currently support adjustments to the dates of known earnings. |
|
20
|
|
|
|
|
21
|
|
|
Parameters |
|
22
|
|
|
---------- |
|
23
|
|
|
all_dates : pd.DatetimeIndex |
|
24
|
|
|
Index of dates for which we can serve queries. |
|
25
|
|
|
announcement_dates : dict[int -> DatetimeIndex] |
|
26
|
|
|
Dict mapping column labels to an index of dates on which earnings were |
|
27
|
|
|
announced. |
|
28
|
|
|
""" |
|
29
|
|
|
def __init__(self, all_dates, announcement_dates): |
|
30
|
|
|
self._all_dates = all_dates |
|
31
|
|
|
self._announcment_dates = announcement_dates |
|
32
|
|
|
|
|
33
|
|
|
def get_loader(self, column): |
|
34
|
|
|
""" |
|
35
|
|
|
Dispatch to the loader for `column`. |
|
36
|
|
|
""" |
|
37
|
|
|
if column is EarningsCalendar.next_announcement: |
|
38
|
|
|
return self.next_announcement_loader |
|
39
|
|
|
elif column is EarningsCalendar.previous_announcement: |
|
40
|
|
|
return self.previous_annoucement_loader |
|
41
|
|
|
else: |
|
42
|
|
|
raise ValueError("Don't know how to load column %s." % column) |
|
43
|
|
|
|
|
44
|
|
|
@lazyval |
|
45
|
|
|
def next_announcement_loader(self): |
|
46
|
|
|
return DataFrameLoader( |
|
47
|
|
|
EarningsCalendar.next_announcement, |
|
48
|
|
|
next_earnings_date_frame( |
|
49
|
|
|
self._all_dates, |
|
50
|
|
|
self._announcement_dates, |
|
51
|
|
|
), |
|
52
|
|
|
adjustments=None, |
|
53
|
|
|
) |
|
54
|
|
|
|
|
55
|
|
|
@lazyval |
|
56
|
|
|
def previous_announcement_loader(self): |
|
57
|
|
|
return DataFrameLoader( |
|
58
|
|
|
EarningsCalendar.previous_announcement, |
|
59
|
|
|
previous_earnings_date_frame( |
|
60
|
|
|
self._all_dates, |
|
61
|
|
|
self._announcement_dates, |
|
62
|
|
|
), |
|
63
|
|
|
adjustments=None, |
|
64
|
|
|
) |
|
65
|
|
|
|
|
66
|
|
|
def load_adjusted_array(self, columns, dates, assets, mask): |
|
67
|
|
|
return { |
|
68
|
|
|
column: self.get_loader(column).load_adjusted_array( |
|
69
|
|
|
[column], dates, assets, mask |
|
70
|
|
|
) |
|
71
|
|
|
for column in columns |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
|
|
75
|
|
|
def next_earnings_date_frame(dates, announcement_dates): |
|
76
|
|
|
""" |
|
77
|
|
|
Make a DataFrame representing simulated next earnings dates. |
|
78
|
|
|
|
|
79
|
|
|
Parameters |
|
80
|
|
|
---------- |
|
81
|
|
|
dates : DatetimeIndex. |
|
82
|
|
|
The index of the returned DataFrame. |
|
83
|
|
|
announcement_dates : dict[int -> DatetimeIndex] |
|
84
|
|
|
Dict mapping sids to an index of dates on which earnings were announced |
|
85
|
|
|
for that sid. |
|
86
|
|
|
|
|
87
|
|
|
Returns |
|
88
|
|
|
------- |
|
89
|
|
|
next_earnings: pd.DataFrame |
|
90
|
|
|
A DataFrame representing, for each (label, date) pair, the first entry |
|
91
|
|
|
in `earnings_calendars[label]` on or after `date`. Entries falling |
|
92
|
|
|
after the last date in a calendar will have `NaT` as the result in the |
|
93
|
|
|
output. |
|
94
|
|
|
|
|
95
|
|
|
See Also |
|
96
|
|
|
-------- |
|
97
|
|
|
next_earnings_date_frame |
|
98
|
|
|
""" |
|
99
|
|
|
cols = {equity: full_like(dates, "NaT") for equity in announcement_dates} |
|
100
|
|
|
for equity, earnings_dates in iteritems(announcement_dates): |
|
101
|
|
|
next_dt_indices = earnings_dates.searchsorted(dates) |
|
102
|
|
|
mask = next_dt_indices < len(earnings_dates) |
|
103
|
|
|
cols[equity][mask] = earnings_dates[next_dt_indices[mask]] |
|
104
|
|
|
|
|
105
|
|
|
return pd.DataFrame(index=dates, data=cols) |
|
106
|
|
|
|
|
107
|
|
|
|
|
108
|
|
|
def previous_earnings_date_frame(dates, announcement_dates): |
|
109
|
|
|
""" |
|
110
|
|
|
Make a DataFrame representing simulated next earnings dates. |
|
111
|
|
|
|
|
112
|
|
|
Parameters |
|
113
|
|
|
---------- |
|
114
|
|
|
dates : DatetimeIndex. |
|
115
|
|
|
The index of the returned DataFrame. |
|
116
|
|
|
announcement_dates : dict[int -> DatetimeIndex] |
|
117
|
|
|
Dict mapping sids to an index of dates on which earnings were announced |
|
118
|
|
|
for that sid. |
|
119
|
|
|
|
|
120
|
|
|
Returns |
|
121
|
|
|
------- |
|
122
|
|
|
prev_earnings: pd.DataFrame |
|
123
|
|
|
A DataFrame representing, for (label, date) pair, the first entry in |
|
124
|
|
|
`announcement_dates[label]` strictly before `date`. Entries falling |
|
125
|
|
|
before the first date in a calendar will have `NaT` as the result in |
|
126
|
|
|
the output. |
|
127
|
|
|
|
|
128
|
|
|
See Also |
|
129
|
|
|
-------- |
|
130
|
|
|
next_earnings_date_frame |
|
131
|
|
|
""" |
|
132
|
|
|
cols = {equity: full_like(dates, "NaT") for equity in announcement_dates} |
|
133
|
|
|
for equity, earnings_dates in iteritems(announcement_dates): |
|
134
|
|
|
# Subtract one to roll back to the index of the previous date. |
|
135
|
|
|
prev_dt_indices = earnings_dates.searchsorted(dates) - 1 |
|
136
|
|
|
mask = prev_dt_indices > 0 |
|
137
|
|
|
cols[equity][mask] = earnings_dates[prev_dt_indices[mask]] |
|
138
|
|
|
|
|
139
|
|
|
return pd.DataFrame(index=dates, data=cols) |
|
140
|
|
|
|
quantopian /
zipline
