-
Notifications
You must be signed in to change notification settings - Fork 0
/
calc_roto.py
366 lines (253 loc) · 13.2 KB
/
calc_roto.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
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
"""
Calculate roto league valuations for the rest of season
"""
import argparse
import itertools
from collections import namedtuple
import numpy
import pandas
from scipy.optimize import minimize_scalar
from scipy.stats import linregress
from utils import load_projections, load_recent_playing_time, load_rosters
from config import MY_TEAM_ID, N_TEAMS, ROSTER_SIZE, TOP_N
COUNTING_STATS = ['3pm', 'pts', 'treb', 'ast', 'stl', 'blk', 'to']
RATIO_STATS = ['fg%', 'ft%']
RATIO_STATS_PARTS = ['fga', 'fgm', 'fta', 'ftm']
IMPORTANT_CATS = ['treb', 'ast', 'stl', 'blk', 'to', 'fg%', 'ft%']
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--optimize", action="store_true", help="run roster optimizer")
args = parser.parse_args()
ros = combine_projections()
# add team id to projections
rosters = load_rosters()
ros = ros.merge(rosters[['yahoo_id', 'team_id']], on='yahoo_id', how='left')
standings = load_standings()
final_standings = calc_final_standings(standings, ros)
print(final_standings[['team_id'] + COUNTING_STATS + RATIO_STATS + RATIO_STATS_PARTS].to_string(index=False))
final_rank = calc_rankings(final_standings)
print(final_rank)
current_final_rank = final_rank.loc[final_rank['team_id'] == MY_TEAM_ID, 'total'].values[0]
# since fg% and ft% are rate stats, we need to establish value above and below a threshold
base_fga = final_standings[final_standings['team_id'] == MY_TEAM_ID]['fga'].values[0]
base_fta = final_standings[final_standings['team_id'] == MY_TEAM_ID]['fta'].values[0]
spg = calc_spg(final_standings, base_fga, base_fta)
base_ratio_stats = {
'fg%': final_standings['fg%'].min(),
'ft%': final_standings['ft%'].min()
}
ros_values = calc_valuation(spg, base_ratio_stats, ros)
# print team valuations
cols = ['yahoo_name', 'yahoo_id', 'team_id', 'rank', 'gtp', 'p_mpg', 'total_value', 'mod_value', 'pts_value', '3pm_value', 'fg%_value', 'ft%_value', 'treb_value', 'ast_value', 'stl_value', 'blk_value', 'to_value']
print(ros_values[ros_values['team_id'] == MY_TEAM_ID][cols])
# print best available free agents
print(ros_values[ros_values['team_id'].isna()].head(15)[cols])
# output valuations
ros_values[cols].to_csv("ros_values.csv", encoding='utf8', index=False)
ros_by_team = ros_values.groupby('team_id').sum().reset_index()
ros_by_team.sort_values('total_value', ascending=False, inplace=True)
if args.optimize:
optimize_roster(ros_values, standings)
def combine_projections():
"""
Use rate projections from Hashtag Basketball, but games to play from Yahoo
"""
id_mapping = pandas.read_csv('id_mapping.csv')
htb = load_projections()
htb.rename(columns={'name': 'htb_name'}, inplace=True)
ros_rate = htb.merge(id_mapping, on='htb_name', how='left')
# check that all players have been ID mapped
print(ros_rate[ros_rate['yahoo_id'].isna()])
yahoo = pandas.read_csv('yahoo_projections.csv')
ros_rate = ros_rate.merge(yahoo[['yahoo_id', 'gtp', 'rank']], how='left')
# override minutes per game projections with recent playing time
pt = load_recent_playing_time()
ros_rate = ros_rate.merge(pt, on='yahoo_id', how='left')
# weight recent playing time by the number of games played recently
ros_rate['p_mpg'] = ros_rate['mpg'].where(
ros_rate['mpg_recent'].isna(),
(
(ros_rate['mpg'] * (14 - ros_rate['gp_recent'])) +
(ros_rate['mpg_recent'] * ros_rate['gp_recent'])
) / (14)
)
for stat in COUNTING_STATS + RATIO_STATS_PARTS:
ros_rate[stat] = ros_rate[stat] / ros_rate['mpg'] * ros_rate['p_mpg']
# override playing time projections with manual ones if necessary
gtp_manual = pandas.read_csv('gtp_manual.csv', comment='#')
gtp_manual.drop('name', axis=1, inplace=True)
ros_rate = ros_rate.merge(gtp_manual, on='yahoo_id', how='left')
ros_rate['gtp'] = ros_rate['gtp_override'].combine_first(ros_rate['gtp'])
ros = ros_rate[['yahoo_name', 'yahoo_id', 'rank', 'gtp', 'p_mpg', 'fg%', 'ft%']].copy()
# scale rate projections to total rest of season stats
for stat in COUNTING_STATS + RATIO_STATS_PARTS:
ros[stat] = ros_rate[stat] * ros_rate['gtp']
return ros
def load_standings():
"""
Load current roto standings
"""
standings = pandas.read_csv('standings.csv')
standings.rename(columns={
'3ptm': '3pm',
'reb': 'treb',
'st': 'stl',
}, inplace=True)
standings.sort_values('team_id', inplace=True)
standings = standings.reset_index()
return standings
def calc_mult(mult, pct_played, max_games, gtp):
"""
Objective function for minimizer
"""
return numpy.abs(max_games - numpy.sum(numpy.clip(pct_played * mult, 0, 1) * gtp))
def find_weights(pct_played, max_games, gtp):
"""
Optimizing function that finds the % of games each player on the roster will start given a maximum number of games
This assumes that the manager will start higher ranked players as much as possible.
"""
result = minimize_scalar(calc_mult, bounds=(1, 20), method='bounded', args=(pct_played[:len(gtp)], max_games, gtp))
return numpy.clip(pct_played[:len(gtp)] * result.x, 0, 1)
def sum_team(x, pct_played, max_games):
"""
Given how often each player on the roster starts, sum the total rest of season stats
"""
w = find_weights(pct_played, max_games, x['gtp'])
return x[COUNTING_STATS + RATIO_STATS_PARTS].multiply(w, axis='index').sum()
def calc_team_projections(projections):
"""
Calculate the total rest of season stats for every team
"""
pct_played = numpy.array((100, 100, 100, 100, 100, 100, 100, 80, 70, 50, 40, 30, 10, 10, 5, 5, 0)) / 100
ros_by_team = projections.groupby('team_id').apply(sum_team, pct_played, projections['gtp'].max())
return ros_by_team.reset_index()
def calc_final_standings(standings, projections):
"""
Given current standings and rest of season player projections, calculate the final standings
This uses a weighted average of the players on the roster.
"""
ros_by_team = calc_team_projections(projections)
final_standings = pandas.DataFrame()
final_standings['team_id'] = standings['team_id']
for stat in COUNTING_STATS + RATIO_STATS_PARTS:
final_standings[stat] = standings[stat] + ros_by_team[stat]
final_standings['fg%'] = final_standings['fgm'] / final_standings['fga']
final_standings['ft%'] = final_standings['ftm'] / final_standings['fta']
return final_standings
def calc_spg(standings, base_fga, base_fta):
"""
Calculate standing points gained based on given standings
Uses a linear regression between the stat's value and the stat's standing points. The worst team has standing points = 1 and the best team has standing points = N_TEAMS + 1.
"""
spg = {}
for stat in COUNTING_STATS + RATIO_STATS:
r = linregress(list(range(1, N_TEAMS + 1)), standings[stat].sort_values())
spg[stat] = r.slope
spg['to'] = -1 * spg['to']
spg['fgp'] = spg['fg%'] * base_fga
spg['ftp'] = spg['ft%'] * base_fta
return spg
def calc_valuation(spg, base_ratio_stats, projections):
"""
Based on standing points gained, calculate player total value.
Mod value is the valuation if punting certain categories.
"""
projections['total_value'] = 0
for stat in COUNTING_STATS:
projections[f"{stat}_value"] = projections[stat] / spg[stat]
projections['total_value'] += projections[f"{stat}_value"]
projections["fg%_value"] = (projections['fg%'] - base_ratio_stats['fg%']) * projections['fga'] / spg['fgp']
projections["ft%_value"] = (projections['ft%'] - base_ratio_stats['ft%']) * projections['fta'] / spg['ftp']
projections['total_value'] += projections["fg%_value"]
projections['total_value'] += projections["ft%_value"]
projections['mod_value'] = projections['total_value'] - projections['pts_value'] - projections['3pm_value']
projections.sort_values('mod_value', ascending=False, inplace=True)
return projections
def calc_rankings(standings):
"""
Based on the given standings, calculate the standing points (rankings)
"""
rankings = standings.rank()
rankings['to'] = standings['to'].rank(ascending=False)
rankings.drop(['fga', 'fgm', 'fta', 'ftm'], axis=1, inplace=True)
rankings['total'] = rankings.sum(axis=1) - rankings['team_id']
rankings.sort_values('total', ascending=False, inplace=True)
return rankings
def calc_buffer(standings):
"""
For each category, what % ahead is my team ahead of the next team?
"""
standings = standings.copy()
standings['to'] = -1 * standings['to']
rankings = standings.rank() # we don't want to reverse turnovers here
behind_values = numpy.sort(numpy.array(standings[IMPORTANT_CATS]), axis=0)[
(rankings[rankings['team_id'] == MY_TEAM_ID][IMPORTANT_CATS].values - 2).flatten().astype(int).tolist(),
tuple(range(7))
]
my_values = standings[standings['team_id'] == MY_TEAM_ID][IMPORTANT_CATS].values
pct_behind = numpy.abs(1 - behind_values / my_values).flatten().tolist()
return dict(zip(IMPORTANT_CATS, pct_behind))
def optimize_roster(projections, standings):
"""
Swap every player on roster for another team's player and see if that improves the team rank
"""
# current roster's standings
final_standings = calc_final_standings(standings, projections)
final_ranks = calc_rankings(final_standings)
team_rank = final_ranks.loc[final_ranks['team_id'] == MY_TEAM_ID, 'total'].values[0]
current_buffer = calc_buffer(final_standings)
tryouts = []
Player = namedtuple('Player', 'yahoo_id team_id rank')
empty_player = Player(yahoo_id=0, team_id=None, rank=None)
for drop_player in itertools.chain(projections[projections['team_id'] == MY_TEAM_ID].itertuples(), [empty_player]):
for add_player in itertools.chain(projections[projections['team_id'].isna()].itertuples(), [empty_player]):
if drop_player != empty_player or add_player != empty_player:
new_final_ranks, new_final_standings = swap_players(drop_player, add_player, projections, standings)
new_team_rank = new_final_ranks.loc[new_final_ranks['team_id'] == MY_TEAM_ID, 'total'].values[0]
changes = (new_final_ranks[new_final_ranks['team_id'] == MY_TEAM_ID] - final_ranks[final_ranks['team_id'] == MY_TEAM_ID]).to_dict('records')[0]
change_string = ','.join([f"{k}:{v}" for k, v in changes.items() if v != 0])
if new_team_rank >= team_rank:
buffer = calc_buffer(new_final_standings)
min_buffer = min(buffer.values())
buffer_change = {k: buffer[k] - current_buffer[k] for k in buffer}
else:
min_buffer = None
buffer_change = {}
tryouts.append(dict({
'drop_player_id': drop_player.yahoo_id,
'add_player_id': add_player.yahoo_id,
'add_player_team_id': add_player.team_id,
'drop_player_rank': drop_player.rank,
'add_player_rank': add_player.rank,
'new_team_rank': new_team_rank,
'change': change_string,
'min_buffer': min_buffer,
}, **buffer_change))
id_mapping = pandas.read_csv('id_mapping.csv')
tryouts = pandas.DataFrame(tryouts)
tryouts = tryouts.merge(id_mapping.rename(columns={'yahoo_name': 'drop_name', 'yahoo_id': 'drop_player_id'})[['drop_player_id', 'drop_name']], on='drop_player_id', how='left')
tryouts = tryouts.merge(id_mapping.rename(columns={'yahoo_name': 'add_name', 'yahoo_id': 'add_player_id'})[['add_player_id', 'add_name']], on='add_player_id', how='left')
tryouts.sort_values(['new_team_rank', 'min_buffer'], ascending=[False, False], inplace=True)
tryouts.to_csv("tryouts.csv", encoding='utf8', index=False, float_format='%.4f')
def swap_players(drop_player, add_player, projections, standings):
"""
Calculate team ranks based on swapping players between teams
"""
if add_player is not None:
projections.loc[projections['yahoo_id'] == add_player.yahoo_id, 'team_id'] = MY_TEAM_ID
if drop_player is not None:
projections.loc[projections['yahoo_id'] == drop_player.yahoo_id, 'team_id'] = add_player.team_id
else: # just dropping a player
projections.loc[projections['yahoo_id'] == drop_player.yahoo_id, 'team_id'] = None
final_standings = calc_final_standings(standings, projections)
final_rank = calc_rankings(final_standings)
# reverse the team changes
if add_player is not None:
projections.loc[projections['yahoo_id'] == add_player.yahoo_id, 'team_id'] = add_player.team_id
if drop_player is not None:
projections.loc[projections['yahoo_id'] == drop_player.yahoo_id, 'team_id'] = MY_TEAM_ID
else:
projections.loc[projections['yahoo_id'] == drop_player.yahoo_id, 'team_id'] = MY_TEAM_ID
return final_rank, final_standings
if __name__ == '__main__':
main()