|
1
|
|
|
# |
|
2
|
|
|
# Copyright 2014 Quantopian, Inc. |
|
3
|
|
|
# |
|
4
|
|
|
# Licensed under the Apache License, Version 2.0 (the "License"); |
|
5
|
|
|
# you may not use this file except in compliance with the License. |
|
6
|
|
|
# You may obtain a copy of the License at |
|
7
|
|
|
# |
|
8
|
|
|
# http://www.apache.org/licenses/LICENSE-2.0 |
|
9
|
|
|
# |
|
10
|
|
|
# Unless required by applicable law or agreed to in writing, software |
|
11
|
|
|
# distributed under the License is distributed on an "AS IS" BASIS, |
|
12
|
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
|
13
|
|
|
# See the License for the specific language governing permissions and |
|
14
|
|
|
# limitations under the License. |
|
15
|
|
|
import datetime |
|
16
|
|
|
import hashlib |
|
17
|
|
|
import os |
|
18
|
|
|
|
|
19
|
|
|
import numpy as np |
|
20
|
|
|
import pandas as pd |
|
21
|
|
|
import pytz |
|
22
|
|
|
import xlrd |
|
23
|
|
|
import requests |
|
24
|
|
|
|
|
25
|
|
|
from six.moves import map |
|
26
|
|
|
|
|
27
|
|
|
|
|
28
|
|
|
def col_letter_to_index(col_letter): |
|
29
|
|
|
# Only supports single letter, |
|
30
|
|
|
# but answer key doesn't need multi-letter, yet. |
|
31
|
|
|
index = 0 |
|
32
|
|
|
for i, char in enumerate(reversed(col_letter)): |
|
33
|
|
|
index += ((ord(char) - 65) + 1) * pow(26, i) |
|
34
|
|
|
return index |
|
35
|
|
|
|
|
36
|
|
|
DIR = os.path.dirname(os.path.realpath(__file__)) |
|
37
|
|
|
|
|
38
|
|
|
ANSWER_KEY_CHECKSUMS_PATH = os.path.join(DIR, 'risk-answer-key-checksums') |
|
39
|
|
|
ANSWER_KEY_CHECKSUMS = open(ANSWER_KEY_CHECKSUMS_PATH, 'r').read().splitlines() |
|
40
|
|
|
|
|
41
|
|
|
ANSWER_KEY_FILENAME = 'risk-answer-key.xlsx' |
|
42
|
|
|
|
|
43
|
|
|
ANSWER_KEY_PATH = os.path.join(DIR, ANSWER_KEY_FILENAME) |
|
44
|
|
|
|
|
45
|
|
|
ANSWER_KEY_BUCKET_NAME = 'zipline-test_data' |
|
46
|
|
|
|
|
47
|
|
|
ANSWER_KEY_DL_TEMPLATE = """ |
|
48
|
|
|
https://s3.amazonaws.com/zipline-test-data/risk/{md5}/risk-answer-key.xlsx |
|
49
|
|
|
""".strip() |
|
50
|
|
|
|
|
51
|
|
|
LATEST_ANSWER_KEY_URL = ANSWER_KEY_DL_TEMPLATE.format( |
|
52
|
|
|
md5=ANSWER_KEY_CHECKSUMS[-1]) |
|
53
|
|
|
|
|
54
|
|
|
|
|
55
|
|
|
def answer_key_signature(): |
|
56
|
|
|
with open(ANSWER_KEY_PATH, 'rb') as f: |
|
57
|
|
|
md5 = hashlib.md5() |
|
58
|
|
|
buf = f.read(1024) |
|
59
|
|
|
md5.update(buf) |
|
60
|
|
|
while buf != b"": |
|
61
|
|
|
buf = f.read(1024) |
|
62
|
|
|
md5.update(buf) |
|
63
|
|
|
return md5.hexdigest() |
|
64
|
|
|
|
|
65
|
|
|
|
|
66
|
|
|
def ensure_latest_answer_key(): |
|
67
|
|
|
""" |
|
68
|
|
|
Get the latest answer key from a publically available location. |
|
69
|
|
|
|
|
70
|
|
|
Logic for determining what and when to download is as such: |
|
71
|
|
|
|
|
72
|
|
|
- If there is no local spreadsheet file, then get the lastest answer key, |
|
73
|
|
|
as defined by the last row in the checksum file. |
|
74
|
|
|
- If there is a local spreadsheet file: |
|
75
|
|
|
-- If the spreadsheet's checksum is in the checksum file: |
|
76
|
|
|
--- If the spreadsheet's checksum does not match the latest, then grab the |
|
77
|
|
|
the latest checksum and replace the local checksum file. |
|
78
|
|
|
--- If the spreadsheet's checksum matches the latest, then skip download, |
|
79
|
|
|
and use the local spreadsheet as a cached copy. |
|
80
|
|
|
-- If the spreadsheet's checksum is not in the checksum file, then leave |
|
81
|
|
|
the local file alone, assuming that the local xls's md5 is not in the list |
|
82
|
|
|
due to local modifications during development. |
|
83
|
|
|
|
|
84
|
|
|
It is possible that md5's could collide, if that is ever case, we should |
|
85
|
|
|
then find an alternative naming scheme. |
|
86
|
|
|
|
|
87
|
|
|
The spreadsheet answer sheet is not kept in SCM, as every edit would |
|
88
|
|
|
increase the repo size by the file size, since it is treated as a binary. |
|
89
|
|
|
""" |
|
90
|
|
|
|
|
91
|
|
|
answer_key_dl_checksum = None |
|
92
|
|
|
|
|
93
|
|
|
local_answer_key_exists = os.path.exists(ANSWER_KEY_PATH) |
|
94
|
|
|
if local_answer_key_exists: |
|
95
|
|
|
local_hash = answer_key_signature() |
|
96
|
|
|
|
|
97
|
|
|
if local_hash in ANSWER_KEY_CHECKSUMS: |
|
98
|
|
|
# Assume previously downloaded version. |
|
99
|
|
|
# Check for latest. |
|
100
|
|
|
if local_hash != ANSWER_KEY_CHECKSUMS[-1]: |
|
101
|
|
|
# More recent checksum, download |
|
102
|
|
|
answer_key_dl_checksum = ANSWER_KEY_CHECKSUMS[-1] |
|
103
|
|
|
else: |
|
104
|
|
|
# Assume local copy that is being developed on |
|
105
|
|
|
answer_key_dl_checksum = None |
|
106
|
|
|
else: |
|
107
|
|
|
answer_key_dl_checksum = ANSWER_KEY_CHECKSUMS[-1] |
|
108
|
|
|
|
|
109
|
|
|
if answer_key_dl_checksum: |
|
110
|
|
|
res = requests.get( |
|
111
|
|
|
ANSWER_KEY_DL_TEMPLATE.format(md5=answer_key_dl_checksum)) |
|
112
|
|
|
with open(ANSWER_KEY_PATH, 'wb') as f: |
|
113
|
|
|
f.write(res.content) |
|
114
|
|
|
|
|
115
|
|
|
# Get latest answer key on load. |
|
116
|
|
|
ensure_latest_answer_key() |
|
117
|
|
|
|
|
118
|
|
|
|
|
119
|
|
|
class DataIndex(object): |
|
120
|
|
|
""" |
|
121
|
|
|
Coordinates for the spreadsheet, using the values as seen in the notebook. |
|
122
|
|
|
The python-excel libraries use 0 index, while the spreadsheet in a GUI |
|
123
|
|
|
uses a 1 index. |
|
124
|
|
|
""" |
|
125
|
|
|
def __init__(self, sheet_name, col, row_start, row_end, |
|
126
|
|
|
value_type='float'): |
|
127
|
|
|
self.sheet_name = sheet_name |
|
128
|
|
|
self.col = col |
|
129
|
|
|
self.row_start = row_start |
|
130
|
|
|
self.row_end = row_end |
|
131
|
|
|
self.value_type = value_type |
|
132
|
|
|
|
|
133
|
|
|
@property |
|
134
|
|
|
def col_index(self): |
|
135
|
|
|
return col_letter_to_index(self.col) - 1 |
|
136
|
|
|
|
|
137
|
|
|
@property |
|
138
|
|
|
def row_start_index(self): |
|
139
|
|
|
return self.row_start - 1 |
|
140
|
|
|
|
|
141
|
|
|
@property |
|
142
|
|
|
def row_end_index(self): |
|
143
|
|
|
return self.row_end - 1 |
|
144
|
|
|
|
|
145
|
|
|
def __str__(self): |
|
146
|
|
|
return "'{sheet_name}'!{col}{row_start}:{col}{row_end}".format( |
|
147
|
|
|
sheet_name=self.sheet_name, |
|
148
|
|
|
col=self.col, |
|
149
|
|
|
row_start=self.row_start, |
|
150
|
|
|
row_end=self.row_end |
|
151
|
|
|
) |
|
152
|
|
|
|
|
153
|
|
|
|
|
154
|
|
|
class AnswerKey(object): |
|
155
|
|
|
|
|
156
|
|
|
INDEXES = { |
|
157
|
|
|
'RETURNS': DataIndex('Sim Period', 'D', 4, 255), |
|
158
|
|
|
|
|
159
|
|
|
'BENCHMARK': { |
|
160
|
|
|
'Dates': DataIndex('s_p', 'A', 4, 254, value_type='date'), |
|
161
|
|
|
'Returns': DataIndex('s_p', 'H', 4, 254) |
|
162
|
|
|
}, |
|
163
|
|
|
|
|
164
|
|
|
# Below matches the inconsistent capitalization in spreadsheet |
|
165
|
|
|
'BENCHMARK_PERIOD_RETURNS': { |
|
166
|
|
|
'Monthly': DataIndex('s_p', 'R', 8, 19), |
|
167
|
|
|
'3-Month': DataIndex('s_p', 'S', 10, 19), |
|
168
|
|
|
'6-month': DataIndex('s_p', 'T', 13, 19), |
|
169
|
|
|
'year': DataIndex('s_p', 'U', 19, 19), |
|
170
|
|
|
}, |
|
171
|
|
|
|
|
172
|
|
|
'BENCHMARK_PERIOD_VOLATILITY': { |
|
173
|
|
|
'Monthly': DataIndex('s_p', 'V', 8, 19), |
|
174
|
|
|
'3-Month': DataIndex('s_p', 'W', 10, 19), |
|
175
|
|
|
'6-month': DataIndex('s_p', 'X', 13, 19), |
|
176
|
|
|
'year': DataIndex('s_p', 'Y', 19, 19), |
|
177
|
|
|
}, |
|
178
|
|
|
|
|
179
|
|
|
'ALGORITHM_PERIOD_RETURNS': { |
|
180
|
|
|
'Monthly': DataIndex('Sim Period', 'Z', 23, 34), |
|
181
|
|
|
'3-Month': DataIndex('Sim Period', 'AA', 25, 34), |
|
182
|
|
|
'6-month': DataIndex('Sim Period', 'AB', 28, 34), |
|
183
|
|
|
'year': DataIndex('Sim Period', 'AC', 34, 34), |
|
184
|
|
|
}, |
|
185
|
|
|
|
|
186
|
|
|
'ALGORITHM_PERIOD_VOLATILITY': { |
|
187
|
|
|
'Monthly': DataIndex('Sim Period', 'AH', 23, 34), |
|
188
|
|
|
'3-Month': DataIndex('Sim Period', 'AI', 25, 34), |
|
189
|
|
|
'6-month': DataIndex('Sim Period', 'AJ', 28, 34), |
|
190
|
|
|
'year': DataIndex('Sim Period', 'AK', 34, 34), |
|
191
|
|
|
}, |
|
192
|
|
|
|
|
193
|
|
|
'ALGORITHM_PERIOD_SHARPE': { |
|
194
|
|
|
'Monthly': DataIndex('Sim Period', 'AL', 23, 34), |
|
195
|
|
|
'3-Month': DataIndex('Sim Period', 'AM', 25, 34), |
|
196
|
|
|
'6-month': DataIndex('Sim Period', 'AN', 28, 34), |
|
197
|
|
|
'year': DataIndex('Sim Period', 'AO', 34, 34), |
|
198
|
|
|
}, |
|
199
|
|
|
|
|
200
|
|
|
'ALGORITHM_PERIOD_BETA': { |
|
201
|
|
|
'Monthly': DataIndex('Sim Period', 'AP', 23, 34), |
|
202
|
|
|
'3-Month': DataIndex('Sim Period', 'AQ', 25, 34), |
|
203
|
|
|
'6-month': DataIndex('Sim Period', 'AR', 28, 34), |
|
204
|
|
|
'year': DataIndex('Sim Period', 'AS', 34, 34), |
|
205
|
|
|
}, |
|
206
|
|
|
|
|
207
|
|
|
'ALGORITHM_PERIOD_ALPHA': { |
|
208
|
|
|
'Monthly': DataIndex('Sim Period', 'AT', 23, 34), |
|
209
|
|
|
'3-Month': DataIndex('Sim Period', 'AU', 25, 34), |
|
210
|
|
|
'6-month': DataIndex('Sim Period', 'AV', 28, 34), |
|
211
|
|
|
'year': DataIndex('Sim Period', 'AW', 34, 34), |
|
212
|
|
|
}, |
|
213
|
|
|
|
|
214
|
|
|
'ALGORITHM_PERIOD_BENCHMARK_VARIANCE': { |
|
215
|
|
|
'Monthly': DataIndex('Sim Period', 'BJ', 23, 34), |
|
216
|
|
|
'3-Month': DataIndex('Sim Period', 'BK', 25, 34), |
|
217
|
|
|
'6-month': DataIndex('Sim Period', 'BL', 28, 34), |
|
218
|
|
|
'year': DataIndex('Sim Period', 'BM', 34, 34), |
|
219
|
|
|
}, |
|
220
|
|
|
|
|
221
|
|
|
'ALGORITHM_PERIOD_COVARIANCE': { |
|
222
|
|
|
'Monthly': DataIndex('Sim Period', 'BF', 23, 34), |
|
223
|
|
|
'3-Month': DataIndex('Sim Period', 'BG', 25, 34), |
|
224
|
|
|
'6-month': DataIndex('Sim Period', 'BH', 28, 34), |
|
225
|
|
|
'year': DataIndex('Sim Period', 'BI', 34, 34), |
|
226
|
|
|
}, |
|
227
|
|
|
|
|
228
|
|
|
'ALGORITHM_PERIOD_DOWNSIDE_RISK': { |
|
229
|
|
|
'Monthly': DataIndex('Sim Period', 'BN', 23, 34), |
|
230
|
|
|
'3-Month': DataIndex('Sim Period', 'BO', 25, 34), |
|
231
|
|
|
'6-month': DataIndex('Sim Period', 'BP', 28, 34), |
|
232
|
|
|
'year': DataIndex('Sim Period', 'BQ', 34, 34), |
|
233
|
|
|
}, |
|
234
|
|
|
|
|
235
|
|
|
'ALGORITHM_PERIOD_SORTINO': { |
|
236
|
|
|
'Monthly': DataIndex('Sim Period', 'BR', 23, 34), |
|
237
|
|
|
'3-Month': DataIndex('Sim Period', 'BS', 25, 34), |
|
238
|
|
|
'6-month': DataIndex('Sim Period', 'BT', 28, 34), |
|
239
|
|
|
'year': DataIndex('Sim Period', 'BU', 34, 34), |
|
240
|
|
|
}, |
|
241
|
|
|
|
|
242
|
|
|
'ALGORITHM_RETURN_VALUES': DataIndex( |
|
243
|
|
|
'Sim Cumulative', 'D', 4, 254), |
|
244
|
|
|
|
|
245
|
|
|
'ALGORITHM_CUMULATIVE_VOLATILITY': DataIndex( |
|
246
|
|
|
'Sim Cumulative', 'P', 4, 254), |
|
247
|
|
|
|
|
248
|
|
|
'ALGORITHM_CUMULATIVE_SHARPE': DataIndex( |
|
249
|
|
|
'Sim Cumulative', 'R', 4, 254), |
|
250
|
|
|
|
|
251
|
|
|
'CUMULATIVE_DOWNSIDE_RISK': DataIndex( |
|
252
|
|
|
'Sim Cumulative', 'U', 4, 254), |
|
253
|
|
|
|
|
254
|
|
|
'CUMULATIVE_SORTINO': DataIndex( |
|
255
|
|
|
'Sim Cumulative', 'V', 4, 254), |
|
256
|
|
|
|
|
257
|
|
|
'CUMULATIVE_INFORMATION': DataIndex( |
|
258
|
|
|
'Sim Cumulative', 'AA', 4, 254), |
|
259
|
|
|
|
|
260
|
|
|
'CUMULATIVE_BETA': DataIndex( |
|
261
|
|
|
'Sim Cumulative', 'AD', 4, 254), |
|
262
|
|
|
|
|
263
|
|
|
'CUMULATIVE_ALPHA': DataIndex( |
|
264
|
|
|
'Sim Cumulative', 'AE', 4, 254), |
|
265
|
|
|
|
|
266
|
|
|
'CUMULATIVE_MAX_DRAWDOWN': DataIndex( |
|
267
|
|
|
'Sim Cumulative', 'AH', 4, 254), |
|
268
|
|
|
|
|
269
|
|
|
} |
|
270
|
|
|
|
|
271
|
|
|
def __init__(self): |
|
272
|
|
|
self.workbook = xlrd.open_workbook(ANSWER_KEY_PATH) |
|
273
|
|
|
|
|
274
|
|
|
self.sheets = {} |
|
275
|
|
|
self.sheets['Sim Period'] = self.workbook.sheet_by_name('Sim Period') |
|
276
|
|
|
self.sheets['Sim Cumulative'] = self.workbook.sheet_by_name( |
|
277
|
|
|
'Sim Cumulative') |
|
278
|
|
|
self.sheets['s_p'] = self.workbook.sheet_by_name('s_p') |
|
279
|
|
|
|
|
280
|
|
|
for name, index in self.INDEXES.items(): |
|
281
|
|
|
if isinstance(index, dict): |
|
282
|
|
|
subvalues = {} |
|
283
|
|
|
for subkey, subindex in index.items(): |
|
284
|
|
|
subvalues[subkey] = self.get_values(subindex) |
|
285
|
|
|
setattr(self, name, subvalues) |
|
286
|
|
|
else: |
|
287
|
|
|
setattr(self, name, self.get_values(index)) |
|
288
|
|
|
|
|
289
|
|
|
def parse_date_value(self, value): |
|
290
|
|
|
return xlrd.xldate_as_tuple(value, 0) |
|
291
|
|
|
|
|
292
|
|
|
def parse_float_value(self, value): |
|
293
|
|
|
return value if value != '' else np.nan |
|
294
|
|
|
|
|
295
|
|
|
def get_raw_values(self, data_index): |
|
296
|
|
|
return self.sheets[data_index.sheet_name].col_values( |
|
297
|
|
|
data_index.col_index, |
|
298
|
|
|
data_index.row_start_index, |
|
299
|
|
|
data_index.row_end_index + 1) |
|
300
|
|
|
|
|
301
|
|
|
@property |
|
302
|
|
|
def value_type_to_value_func(self): |
|
303
|
|
|
return { |
|
304
|
|
|
'float': self.parse_float_value, |
|
305
|
|
|
'date': self.parse_date_value, |
|
306
|
|
|
} |
|
307
|
|
|
|
|
308
|
|
|
def get_values(self, data_index): |
|
309
|
|
|
value_parser = self.value_type_to_value_func[data_index.value_type] |
|
310
|
|
|
return [value for value in |
|
311
|
|
|
map(value_parser, self.get_raw_values(data_index))] |
|
312
|
|
|
|
|
313
|
|
|
|
|
314
|
|
|
ANSWER_KEY = AnswerKey() |
|
315
|
|
|
|
|
316
|
|
|
BENCHMARK_DATES = ANSWER_KEY.BENCHMARK['Dates'] |
|
317
|
|
|
BENCHMARK_RETURNS = ANSWER_KEY.BENCHMARK['Returns'] |
|
318
|
|
|
DATES = [datetime.datetime(*x, tzinfo=pytz.UTC) for x in BENCHMARK_DATES] |
|
319
|
|
|
BENCHMARK = pd.Series(dict(zip(DATES, BENCHMARK_RETURNS))) |
|
320
|
|
|
ALGORITHM_RETURNS = pd.Series( |
|
321
|
|
|
dict(zip(DATES, ANSWER_KEY.ALGORITHM_RETURN_VALUES))) |
|
322
|
|
|
RETURNS_DATA = pd.DataFrame({'Benchmark Returns': BENCHMARK, |
|
323
|
|
|
'Algorithm Returns': ALGORITHM_RETURNS}) |
|
324
|
|
|
RISK_CUMULATIVE = pd.DataFrame({ |
|
325
|
|
|
'volatility': pd.Series(dict(zip( |
|
326
|
|
|
DATES, ANSWER_KEY.ALGORITHM_CUMULATIVE_VOLATILITY))), |
|
327
|
|
|
'sharpe': pd.Series(dict(zip( |
|
328
|
|
|
DATES, ANSWER_KEY.ALGORITHM_CUMULATIVE_SHARPE))), |
|
329
|
|
|
'downside_risk': pd.Series(dict(zip( |
|
330
|
|
|
DATES, ANSWER_KEY.CUMULATIVE_DOWNSIDE_RISK))), |
|
331
|
|
|
'sortino': pd.Series(dict(zip( |
|
332
|
|
|
DATES, ANSWER_KEY.CUMULATIVE_SORTINO))), |
|
333
|
|
|
'information': pd.Series(dict(zip( |
|
334
|
|
|
DATES, ANSWER_KEY.CUMULATIVE_INFORMATION))), |
|
335
|
|
|
'alpha': pd.Series(dict(zip( |
|
336
|
|
|
DATES, ANSWER_KEY.CUMULATIVE_ALPHA))), |
|
337
|
|
|
'beta': pd.Series(dict(zip( |
|
338
|
|
|
DATES, ANSWER_KEY.CUMULATIVE_BETA))), |
|
339
|
|
|
'max_drawdown': pd.Series(dict(zip( |
|
340
|
|
|
DATES, ANSWER_KEY.CUMULATIVE_MAX_DRAWDOWN))), |
|
341
|
|
|
}) |
|
342
|
|
|
|
quantopian /
zipline
