-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsignal_confirmation.py
138 lines (114 loc) · 4.83 KB
/
signal_confirmation.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
# ---
# jupyter:
# jupytext:
# formats: ipynb,py:percent
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.6.0
# kernelspec:
# display_name: Python 3
# language: python
# name: python3
# ---
# %%
# %load_ext autoreload
# %autoreload 2
# %matplotlib inline
from pprint import pprint
from IPython.display import display
# %%
import pandas as pd
import numpy as np
import matplotlib.pylab as plt
# %%
from model_logistic_regression import y, X, idx, num_test
from model_logistic_regression import scaler, model
from turnpt_analysis import get_daily_TXF_OHLC
# %%
X_scaled = scaler.transform(X)
X_, y_, idx_ = X_scaled[-num_test:], y[-num_test:], idx[-num_test:]
# %%
signal = model.predict(X_)
signal
# %%
def get_position(df, holding_days=5):
df['exit'] = 0
df['exit_at'] = np.nan
df['exit_at'] = np.where((df.enter == 1) & (df.p0 > df.p5), 'p5', df.exit_at)# NOTE: order is important
df['exit_at'] = np.where((df.enter == 1) & (df.p0 > df.p4), 'p4', df.exit_at)
df['exit_at'] = np.where((df.enter == 1) & (df.p0 > df.p3), 'p3', df.exit_at)
df['exit_at'] = np.where((df.enter == 1) & (df.p0 > df.p2), 'p2', df.exit_at)
df['exit_at'] = np.where((df.enter == 1) & (df.p0 > df.p1), 'p1', 'p_holding')
assert df.exit_at.isna().sum() == 0
df['prc_exit'] = np.nan
df['prc_exit'] = np.where(df.exit_at == 'p5', df.p5, df.prc_exit)
df['prc_exit'] = np.where(df.exit_at == 'p4', df.p4, df.prc_exit)
df['prc_exit'] = np.where(df.exit_at == 'p3', df.p3, df.prc_exit)
df['prc_exit'] = np.where(df.exit_at == 'p2', df.p2, df.prc_exit)
df['prc_exit'] = np.where(df.exit_at == 'p1', df.p1, df.prc_exit)
df['prc_exit'] = np.where(df.exit_at == 'p_holding', df.p0.shift(-holding_days), df.prc_exit)
days_enter = df.query('enter == 1').index
nrows = df.shape[0]
for idx_row in range(nrows):
dt, row = df.index[idx_row], df.iloc[idx_row]
if dt in days_enter:
if row.exit_at == 'p_holding':
idx_exit = idx_row+holding_days
elif row.exit_at == 'p1':
idx_exit = idx_row+1
elif row.exit_at == 'p2':
idx_exit = idx_row+2
elif row.exit_at == 'p3':
idx_exit = idx_row+3
elif row.exit_at == 'p4':
idx_exit = idx_row+4
elif row.exit_at == 'p5':
idx_exit = idx_row+5
else:
raise ValueError(f'There is no such exit_at code: {row.exit_at}')
if idx_exit <= nrows-1:
dt_exit = df.index[idx_exit]
df.loc[dt_exit, 'exit'] += 1
buy = df.enter.cumsum()
sell = df.exit.cumsum()
df['position'] = buy - sell
return df
def get_df_action(dt_includes, signal, holding_days):
df = (
get_daily_TXF_OHLC().
set_index('tx_datetime').loc[:, ['close']].
query('index in @idx_').
assign(signal = signal,
signal_true = y_).
assign(p0 = lambda x: x.close,
p_1 = lambda x: x.close.shift(1), p_2 = lambda x: x.close.shift(2), p_3 = lambda x: x.close.shift(3), p_4 = lambda x: x.close.shift(4), p_5 = lambda x: x.close.shift(5),
p1 = lambda x: x.close.shift(-1), p2 = lambda x: x.close.shift(-2), p3 = lambda x: x.close.shift(-3), p4 = lambda x: x.close.shift(-4), p5 = lambda x: x.close.shift(-5),).
dropna().
assign(enter = lambda x: np.where(x[['p0', 'p_1', 'p_2', 'p_3', 'p_4', 'p_5']].idxmin(axis=1) == 'p0', x.signal, 0)).
pipe(get_position, holding_days = holding_days).
assign(action = lambda x: x.position.diff().fillna(x.iloc[0].position))
)
return df
# %%
if __name__ == '__main__':
holding_days = 5
df_action = get_df_action(idx_, signal, holding_days)
display(df_action.head(15).loc[:, ['signal_true', 'signal', 'enter', 'exit_at', 'position', 'action']])
# correlation
df_signal_confirmed = (
df_action.
assign(signal_enter = lambda x: x.enter == 1,
signal_t1 = lambda x: ((x.enter == 1) & (~x.exit_at.isin(['p1']))).astype(int),
signal_t2 = lambda x: ((x.enter == 1) & (~x.exit_at.isin(['p1', 'p2']))).astype(int),
signal_t3 = lambda x: ((x.enter == 1) & (~x.exit_at.isin(['p1', 'p2', 'p3']))).astype(int),
signal_t4 = lambda x: ((x.enter == 1) & (~x.exit_at.isin(['p1', 'p2', 'p3', 'p4']))).astype(int),
signal_t5 = lambda x: ((x.enter == 1) & (~x.exit_at.isin(['p1', 'p2', 'p3', 'p4', 'p5']))).astype(int)).
loc[:, ['signal_true', 'signal', 'signal_enter', 'signal_t1', 'signal_t2', 'signal_t3', 'signal_t4', 'signal_t5']]
)
display(round(df_signal_confirmed.corr(), 2))
# %%
# %%
# %%
# %%