-
Notifications
You must be signed in to change notification settings - Fork 19
/
Copy pathdemo.py
99 lines (85 loc) · 4.08 KB
/
demo.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
from config import IR_threshold, gp_start, gp_end, gp_predPeriod, gp_dataDir_holc, \
gp_dataDir_h5, gp_us191, gp_us101, gp_pickalphaDir
import sys
sys.path.append(gp_pickalphaDir)
import os
import numpy as np
import pandas as pd
from pick_alpha import alphaTest
from utilities import function_set, feature_names, transform_label, stdization
from cubicalGenetic import genetic, fitness, functions
from calculate_alpha import calculateAlpha
def genBetterFormula(fitResult, threshold):
'''
输入fit后的est_gp和fitness的阈值，从最后一代选出超过阈值的公式
'''
if fitResult.metric.greater_is_better:
better_fitness = [ f.__str__() for f in fitResult._programs[-1] if f.fitness_ > threshold]
else:
better_fitness = [ f.__str__() for f in fitResult._programs[-1] if f.fitness_ < threshold]
return better_fitness
def IR(y, y_pred, w):
alp = pd.DataFrame(y_pred, index=alphaFitter.univ, columns=alphaFitter.dateRange)
coef, _, _, _, _ = alphaFitter.singleAlpRegression(alp)
if coef.isnull().sum() / len(coef) > 0.8 or coef.std() == 0 :
return 0
else:
IR = coef.mean() / coef.std() * np.sqrt(252 / gp_predPeriod)
if np.isnan(IR):
return 0
else:
return abs(IR)
def ICIR(y, y_pred, w):
alp = pd.DataFrame(y_pred, index=alphaFitter.univ, columns=alphaFitter.dateRange)
_, _, _, _, IC = alphaFitter.singleAlpRegression(alp)
ICIR = IC.mean() / IC.std()
if np.isnan(ICIR):
return 0
else:
return abs(ICIR)
if __name__ == '__main__':
alphaFitter = alphaTest(startDate=gp_start, endDate=gp_end, predPeriod=gp_predPeriod, dataDir=gp_dataDir_h5)
#################################
openPx = pd.read_csv(os.path.join(gp_dataDir_holc,'openprice_adj.csv'), index_col=0).reindex(index=alphaFitter.univ, columns=alphaFitter.dateRange)
highPx = pd.read_csv(os.path.join(gp_dataDir_holc,'highprice_adj.csv'), index_col=0).reindex(index=alphaFitter.univ, columns=alphaFitter.dateRange)
lowPx = pd.read_csv(os.path.join(gp_dataDir_holc,'lowprice_adj.csv'), index_col=0).reindex(index=alphaFitter.univ, columns=alphaFitter.dateRange)
closePx = pd.read_csv(os.path.join(gp_dataDir_holc,'closeprice_adj.csv'), index_col=0).reindex(index=alphaFitter.univ, columns=alphaFitter.dateRange)
volume = pd.read_csv(os.path.join(gp_dataDir_holc,'volume.csv'), index_col=0).reindex(index=alphaFitter.univ, columns=alphaFitter.dateRange)
vwap = pd.read_csv(os.path.join(gp_dataDir_holc,'vwap.csv'), index_col=0).reindex(index=alphaFitter.univ, columns=alphaFitter.dateRange)
industry = pd.read_csv(os.path.join(gp_dataDir_holc,'industry.csv'), index_col=0)
functions.ind_label = transform_label(industry).reindex(index=alphaFitter.univ)
dataArray = np.concatenate([
np.expand_dims(openPx.fillna(0), axis=2),
np.expand_dims(highPx.fillna(0), axis=2),
np.expand_dims(lowPx.fillna(0), axis=2),
np.expand_dims(closePx.fillna(0), axis=2),
np.expand_dims(volume.fillna(0), axis=2),
np.expand_dims(vwap.fillna(0), axis=2),
], axis=2)
################ UPDATE DATA#################
est_gp = genetic.SymbolicRegressor(
population_size = 30,
generations = 2,
tournament_size = 10,
init_depth = (1, 3),
stopping_criteria = 500,
max_samples = 0.9,
low_memory = True,
feature_names = feature_names,
function_set = function_set,
metric = fitness.make_fitness(IR, True),
verbose = 1,
const_range = (3,7),
init_method = 'grow'
)
#gp_us101, _ = stdization(gp_us101)
#gp_us191, _ = stdization(gp_us191)
#gp_us = gp_us101 + gp_us191
# est_gp.fit(X=dataArray, y=np.squeeze(alphaFitter.regDataset['regRet']), us=gp_us)
est_gp.fit(X=dataArray, y=np.squeeze(alphaFitter.regDataset['regRet']))
#获取符合要求的goodAlpha
alpha = genBetterFormula(est_gp, IR_threshold)
# 去重
alpha = list(set(alpha))
#完成alpha更新
calculateAlpha(alpha=alpha, factor_value=False)