forked from enarjord/passivbot
-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathpso.py
222 lines (190 loc) · 8.16 KB
/
pso.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
import pyswarms as ps
import asyncio
import aiomultiprocess
from multiprocessing import shared_memory, Lock
from collections import OrderedDict
from backtest import backtest
from plotting import plot_fills
from downloader import Downloader, prep_config
from pure_funcs import denumpyize, numpyize, get_template_live_config, candidate_to_live_config, calc_spans, \
get_template_live_config, unpack_config, pack_config, analyze_fills, ts_to_date, denanify
from procedures import dump_live_config, load_live_config, make_get_filepath, add_argparse_args
from time import time
from optimize import iter_slices, iter_slices_full_first, objective_function, get_expanded_ranges, single_sliding_window_run
import os
import sys
import argparse
import pprint
import matplotlib.pyplot as plt
import json
import pandas as pd
import numpy as np
import asyncio
lock = Lock()
BEST_OBJECTIVE = 0.0
async def aiowrap(func, args=()):
result = await aiomultiprocess.Worker(target=func, args=args)
return result
class Worker:
def __init__(self, data, config):
self.data = data
self.config = config
self.expanded_ranges = get_expanded_ranges(config)
for k in list(self.expanded_ranges):
if self.expanded_ranges[k][0] == self.expanded_ranges[k][1]:
del self.expanded_ranges[k]
self.bounds = get_bounds(self.expanded_ranges)
self.result = None
self.done = True
self.idle = True
async def run_task(self, candidate):
self.done = False
self.idle = False
config = self.candidate_to_config(candidate)
self.result = single_sliding_window_run(config, self.data)
self.done = True
def process_result(self):
result = self.result.copy()
self.result = None
return result
def config_to_xs(self, config):
xs = np.zeros(len(self.bounds[0]))
unpacked = unpack_config(config)
for i, k in enumerate(self.expanded_ranges):
xs[i] = unpacked[k]
return xs
def xs_to_config(self, xs):
config = self.config.copy()
for i, k in enumerate(self.expanded_ranges):
config[k] = xs[i]
return numpyize(denanify(pack_config(config)))
def rf(self, xss):
return np.array([self.single_rf(xs) for xs in xss])
def single_rf(self, xs):
config = self.xs_to_config(xs)
objective, analyses = single_sliding_window_run(config, self.data)
return -objective
def dump_result(result):
pass
def custom_pso(data, config):
workers = {i: Worker(data, config) for i in range(config['num_cpus'])}
results = []
k = 0
while True:
if k >= config['iters']:
break
for worker in workers:
if worker.idle:
asyncio.create_task(worker.run_task(candidate))
pass
elif worker.done:
result = worker.process_result()
results.append(result)
if result['objective'] > best_objective:
best_objective = result['objective']
best_result = result
dump_result(result)
k += 1
worker.idle = True
await asyncio.sleep(0.01)
def get_bounds(ranges: dict) -> tuple:
return (np.array([float(v[0]) for k, v in ranges.items()]),
np.array([float(v[1]) for k, v in ranges.items()]))
class BacktestPSO:
def __init__(self, data, config):
self.data = data
self.config = config
self.expanded_ranges = get_expanded_ranges(config)
for k in list(self.expanded_ranges):
if self.expanded_ranges[k][0] == self.expanded_ranges[k][1]:
del self.expanded_ranges[k]
self.bounds = get_bounds(self.expanded_ranges)
def config_to_xs(self, config):
xs = np.zeros(len(self.bounds[0]))
unpacked = unpack_config(config)
for i, k in enumerate(self.expanded_ranges):
xs[i] = unpacked[k]
return xs
def xs_to_config(self, xs):
config = self.config.copy()
for i, k in enumerate(self.expanded_ranges):
config[k] = xs[i]
return numpyize(denanify(pack_config(config)))
def rf(self, xss):
return np.array([self.single_rf(xs) for xs in xss])
def single_rf(self, xs):
config = self.xs_to_config(xs)
objective, analyses = single_sliding_window_run(config, self.data)
global lock, BEST_OBJECTIVE
if analyses:
try:
lock.acquire()
to_dump = {}
for k in ['average_daily_gain', 'score']:
to_dump[k] = np.mean([e[k] for e in analyses])
for k in ['lowest_eqbal_ratio', 'closest_bkr']:
to_dump[k] = np.min([e[k] for e in analyses])
for k in ['max_hrs_no_fills', 'max_hrs_no_fills_same_side']:
to_dump[k] = np.max([e[k] for e in analyses])
to_dump['objective'] = objective
to_dump.update(candidate_to_live_config(config))
with open(self.config['optimize_dirpath'] + 'intermediate_results.txt', 'a') as f:
f.write(json.dumps(to_dump) + '\n')
if objective > BEST_OBJECTIVE:
if analyses:
config['average_daily_gain'] = np.mean([e['average_daily_gain'] for e in analyses])
dump_live_config({**config, **{'objective': objective}}, self.config['optimize_dirpath'] + 'intermediate_best_results.json')
BEST_OBJECTIVE = objective
finally:
lock.release()
return -objective
async def main():
parser = argparse.ArgumentParser(prog='Optimize', description='Optimize passivbot config.')
parser = add_argparse_args(parser)
parser.add_argument('-t', '--start', type=str, required=False, dest='starting_configs',
default=None,
help='start with given live configs. single json file or dir with multiple json files')
args = parser.parse_args()
config = await prep_config(args)
try:
template_live_config = get_template_live_config(config['n_spans'])
config = {**template_live_config, **config}
dl = Downloader(config)
data = await dl.get_data()
shms = [shared_memory.SharedMemory(create=True, size=d.nbytes) for d in data]
shdata = [np.ndarray(d.shape, dtype=d.dtype, buffer=shms[i].buf) for i, d in enumerate(data)]
for i in range(len(data)):
shdata[i][:] = data[i][:]
del data
config['n_days'] = (shdata[2][-1] - shdata[2][0]) / (1000 * 60 * 60 * 24)
config['optimize_dirpath'] = make_get_filepath(os.path.join(config['optimize_dirpath'],
ts_to_date(time())[:19].replace(':', ''), ''))
print()
for k in (keys := ['exchange', 'symbol', 'starting_balance', 'start_date', 'end_date', 'latency_simulation_ms',
'do_long', 'do_shrt', 'minimum_bankruptcy_distance', 'maximum_hrs_no_fills',
'maximum_hrs_no_fills_same_side', 'iters', 'n_particles', 'sliding_window_size',
'n_spans']):
if k in config:
print(f"{k: <{max(map(len, keys)) + 2}} {config[k]}")
print()
bpso = BacktestPSO(tuple(shdata), config)
optimizer = ps.single.GlobalBestPSO(n_particles=24, dimensions=len(bpso.bounds[0]), options=config['options'],
bounds=bpso.bounds, init_pos=None)
# todo: implement starting configs
cost, pos = optimizer.optimize(bpso.rf, iters=config['iters'], n_processes=config['num_cpus'])
print(cost, pos)
best_candidate = bpso.xs_to_config(pos)
print('best candidate', best_candidate)
'''
conf = bpso.xs_to_config(xs)
print('starting...')
objective = bpso.rf(xs)
print(objective)
'''
finally:
del shdata
for shm in shms:
shm.close()
shm.unlink()
if __name__ == '__main__':
asyncio.run(main())