generated from Advestis/complex
-
Notifications
You must be signed in to change notification settings - Fork 0
/
portfolio_optimisation.py
165 lines (139 loc) · 5.8 KB
/
portfolio_optimisation.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
# Useful refeences, number refer to my note
# calculate temperature for current epoch
# [1] http: // what - when - how.com / artificial - intelligence / a - comparison - of - cooling - schedules -
# for -simulated - annealing - artificial - intelligence /
# [2] http://www.scielo.org.mx/pdf/cys/v21n3/1405-5546-cys-21-03-00493.pdf
# [3] https://www.researchgate.net/publication/227061666_Computing_the_Initial_Temperature_of_Simulated_Annealing/link/
# 543f88a20cf2e76f02246e49/download
# [4] https://nathanrooy.github.io/posts/2020-05-14/simulated-annealing-with-python/
import os
os.environ["OMP_NUM_THREADS"] = "1"
os.environ["OPENBLAS_NUM_THREADS"] = "1"
os.environ["MKL_NUM_THREADS"] = "1"
os.environ["VECLIB_MAXIMUM_THREADS"] = "1"
os.environ["NUMEXPR_NUM_THREADS"] = "1"
from adutils import setup_logger
import numpy as np
setup_logger()
from pathlib import Path
import logging
import argparse
from adlearn.engine import Engine
from time import time
import matplotlib.pyplot as plt
import pandas as pd
engine = Engine(kind="multiproc")
from adannealing import Annealer
from profiling.financial import (
load_financial_configurations,
LossPortfolioMeanVar,
)
Annealer.set_cpu_limit(1)
logger = logging.getLogger(__name__)
(
path_save_images,
date,
common_fee,
overall_risk_coeff,
overall_sparse_coeff,
overall_norm_coeff,
sparsity,
desired_norm,
continous_window,
n_iterations,
step_size,
alpha,
all_prices,
) = load_financial_configurations("profiling/run_configs.json")
def run(number_isins, verbose=True):
logger.info("")
logger.info(f"Starting annealing profiler with {number_isins} isins...")
limits = tuple([(-1, 1) for _ in range(number_isins)])
selected_prices = all_prices.dropna(how="any", axis=1)
chosen_isins = selected_prices.columns[:number_isins]
# selected_prices is dense
selected_prices = selected_prices[chosen_isins]
selected_returns = selected_prices.pct_change()
selected_cov = selected_returns.cov()
# test loss evaluation at some dates
# startin equi-w
weights_day_before = pd.DataFrame(
data=np.full(shape=(len(chosen_isins), 1), fill_value=(1.0 / number_isins)), index=[chosen_isins]
)
fees = pd.DataFrame(data=np.full(shape=(number_isins, 1), fill_value=common_fee), index=[chosen_isins])
portfolio_opt_constraints = LossPortfolioMeanVar(
wt_1_np=weights_day_before.to_numpy(),
r_np=selected_returns.loc[date].to_numpy().reshape((number_isins, 1)),
lambda_risk=overall_risk_coeff,
lambda_sparse=overall_sparse_coeff,
lambda_norm=overall_norm_coeff,
fees=fees,
cov_risk=selected_cov.to_numpy(),
sparsity_target=sparsity,
constraints=limits,
sum_w_target=desired_norm,
continous_window=continous_window,
n=len(chosen_isins),
)
bounds_min = np.full(shape=(1, number_isins), fill_value=-1.0)
bounds_max = np.full(shape=(1, number_isins), fill_value=+1.0)
bounds = np.concatenate([bounds_min, bounds_max]).T
# Using custom start temp.
t0 = time()
hpath = Path(path_save_images) / f"history_{number_isins}"
if not hpath.is_dir():
hpath.mkdir()
ann = Annealer(
loss=portfolio_opt_constraints,
weights_step_size=step_size,
bounds=bounds,
alpha=alpha,
iterations=n_iterations,
verbose=verbose,
history_path=str(hpath),
logger_level="INFO",
# TODO: test more this experimental feature
optimal_step_size=True, # experimental
)
numerical_solution, val_at_best, _, hist, final_hist, _ = ann.fit(
alpha=alpha, stopping_limit=0.001, npoints=1, stop_at_first_found=True
)
tf = time() - t0
ann.plot(hpath, step_size=10, weights_names=chosen_isins, do_3d=True)
logger.info(f"date : {date}")
logger.info(f"Numerical loss : {val_at_best}")
logger.info(f"Annealing time: {tf} s")
if __name__ == "__main__":
parser = argparse.ArgumentParser(
description="AdAnnealing Profiler",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
parser.add_argument("-n", "--nisins", type=int, default=5, help="Number of isins to use")
parser.add_argument("-p", "--plot", action="store_true", help="Do plot if nisins <= 5")
parser.add_argument("-s", "--start", type=int, default=5, help="Initial number of isins to use with 'profile'")
parser.add_argument("-S", "--step", type=int, default=1, help="Steps in number of isins to use with 'profile'")
parser.add_argument("-e", "--end", type=int, default=40, help="Final number of isins to use with 'profile'")
parser.add_argument("-P", "--profile", action="store_true", help="Do profiling")
parser.add_argument("-m", "--multiproc", action="store_true", help="Do profiling in parallel")
args = parser.parse_args()
if args.profile:
if args.end != -1:
isins = list(range(args.start, args.end + 1, args.step))
else:
isins = list(range(args.start, len(all_prices.columns), args.step))
if args.multiproc:
errors_norms_times = engine(run, isins, do_plot=False, verbose=False)
else:
errors_norms_times = [run(i, False) for i in isins]
isins, errors, times = zip(*errors_norms_times)
fig, axes = plt.subplots(2, 1, figsize=(10, 7))
axes[1].set_xlabel("# Isins", fontsize=15)
axes[0].set_ylabel("Errors (%)", fontsize=15)
axes[1].set_ylabel("Annealing time (s)", fontsize=15)
axes[0].grid(True, ls="--", lw=0.2, alpha=0.5)
axes[1].grid(True, ls="--", lw=0.2, alpha=0.5)
axes[0].scatter(isins, errors)
axes[1].scatter(isins, times)
fig.savefig(str(Path(path_save_images) / f"profile_{args.start}_{args.end}_{args.step}.pdf"))
else:
run(args.nisins, args.plot)