forked from bobcolner/pandas-polygon
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathbar_labels.py
180 lines (146 loc) · 7.32 KB
/
bar_labels.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
import statsmodels.api as sm
import pandas as pd
def get_trend_outcome(label_prices: pd.DataFrame) -> dict:
if len(label_prices) < 30:
return {}
df = label_prices.copy()
df['const'] = 1
df = df.reset_index()
model = sm.OLS(endog=df['price'], exog=df[['const', 'index']])
results = model.fit()
trend = {
'label_trend_slope': results.params[1],
'label_trend_tvalue': results.tvalues[0],
'label_trend_r2': results.rsquared,
}
return trend
def get_tb_outcome(reward_ratio: float, risk_level: float, side: str, label_prices: pd.DataFrame,
goal: str='profit', price_col: str='jma') -> dict:
first_price = label_prices['price'].values[0]
if side=='long':
if goal=='profit':
target_price = first_price + (risk_level * reward_ratio)
target_at = label_prices[label_prices[price_col] >= target_price].min()['utc_dt']
elif goal=='stop':
target_price = first_price - risk_level
target_at = label_prices[label_prices[price_col] < target_price].min()['utc_dt']
reward_ratio = -1
elif side=='short':
if goal=='profit':
target_price = first_price - (risk_level * reward_ratio)
target_at = label_prices[label_prices[price_col] <= target_price].min()['utc_dt']
elif goal=='stop':
target_price = first_price + risk_level
target_at = label_prices[label_prices[price_col] > target_price].min()['utc_dt']
reward_ratio = -1
reward_ratio = reward_ratio * -1
outcome = {
'label_side': side,
'label_outcome': goal,
'label_rrr': reward_ratio,
'label_outcome_at': target_at,
}
return outcome
def triple_barrier_outcomes(label_prices: pd.DataFrame, risk_level: float, reward_ratios: list) -> list:
first_price = label_prices['jma'].values[0]
tb_outcomes = []
for side in ['long', 'short']:
stop_outcome = get_tb_outcome(None, risk_level, side, label_prices, goal='stop')
tb_outcomes.append(stop_outcome)
for reward in reward_ratios:
profit_outcome = get_tb_outcome(reward, risk_level, side, label_prices, goal='profit')
tb_outcomes.append(profit_outcome)
tb_df = pd.DataFrame(tb_outcomes).sort_values('label_outcome_at')
return tb_df
def signed_outcomes_to_label(outcomes: pd.DataFrame, label_end_at: pd._libs.tslibs.timestamps.Timestamp) -> dict:
outcomes = outcomes.dropna()
if outcomes.shape[0] == 0: # no outcomes
# print('neutral')
label = [{
'label_side': 'neutral',
'label_outcome': 'neutral',
'label_rrr': 0,
'label_outcome_at': label_end_at,
}]
elif outcomes[outcomes['label_outcome']=='stop'].shape[0] > 0: # stop-loss found
idx = outcomes[outcomes['label_outcome']=='stop'].index.values[0] # index of stop
if idx == 0: # stop-loss is first outcome
# print('stop')
label = outcomes.head(1).to_dict(orient='records')
elif idx > 0:
# print('profit before stop')
multi_label = outcomes[outcomes.index < idx] # profit outcomes before stop-out
label = multi_label[multi_label['label_rrr'].abs()==multi_label['label_rrr'].abs().max()].to_dict(orient='records')
else:
# print('profit')
label = outcomes[outcomes['label_rrr'].abs()==outcomes['label_rrr'].abs().max()].to_dict(orient='records') # get max reward
return label[0]
def outcomes_to_label(outcomes: pd.DataFrame, label_end_at: pd._libs.tslibs.timestamps.Timestamp) -> dict:
long_outcomes = outcomes.loc[outcomes['label_side']=='long'].reset_index(drop=True)
short_outcomes = outcomes.loc[outcomes['label_side']=='short'].reset_index(drop=True)
long_label = signed_outcomes_to_label(long_outcomes, label_end_at)
short_label = signed_outcomes_to_label(short_outcomes, label_end_at)
if (long_label['label_outcome'] == 'profit') and (short_label['label_outcome'] == 'stop'):
label = long_label
elif (short_label['label_outcome'] == 'profit') and (long_label['label_outcome'] == 'stop'):
label = short_label
elif (short_label['label_outcome'] in ['neutral', 'stop']) and (long_label['label_outcome'] in ['neutral', 'stop']):
try:
label_end_at = max(long_label['label_outcome_at'], short_label['label_outcome_at'])
except:
label_end_at = label_end_at
label = {
'label_side': 'neutral',
'label_outcome': 'neutral',
'label_rrr': 0,
'label_outcome_at': label_end_at,
}
else:
label = {'label_outcome': 'unknown'}
return label
def get_concurrent_stats(lbars_df: pd.DataFrame) -> dict:
# from mlfinlab.sampling.bootstrapping import get_ind_matrix, get_ind_mat_average_uniqueness
from mlfinlab.sampling.concurrent import get_av_uniqueness_from_triple_barrier
samples_info_sets = lbars_df[['label_start_at', 'label_outcome_at']]
samples_info_sets = samples_info_sets.set_index('label_start_at')
samples_info_sets.columns = ['t1'] # t1 = label_outcome_at
price_bars = lbars_df[['open_at', 'close_at', 'price_close']]
price_bars = price_bars.set_index('close_at')
label_avg_unq = get_av_uniqueness_from_triple_barrier(samples_info_sets, price_bars, num_threads=1)
# ind_mat = get_ind_matrix(samples_info_sets, price_bars)
# avg_unq_ind_mat = get_ind_mat_average_uniqueness(ind_mat)
results = {
# 'label_avg_unq': label_avg_unq,
'grand_avg_unq': label_avg_unq['tW'].mean(),
# 'ind_mat': ind_mat,
# 'ind_mat_avg_unq': avg_unq_ind_mat
}
return results
def get_label_ticks(ticks_df: pd.DataFrame, label_start_at: pd._libs.tslibs.timestamps.Timestamp,
horizon_mins: int) -> pd.DataFrame:
delayed_label_start_at = label_start_at + pd.Timedelta(value=3, unit='seconds') # inference+network latency compensation
label_end_at = label_start_at + pd.Timedelta(value=horizon_mins, unit='minutes')
label_prices = ticks_df.loc[(ticks_df['utc_dt'] >= delayed_label_start_at) & (ticks_df['utc_dt'] < label_end_at)]
return label_prices, label_end_at
def label_bars(bars: list, ticks_df: pd.DataFrame, risk_level: float, horizon_mins: int,
reward_ratios: list, add_trend_label: bool=False) -> list:
for idx, row in enumerate(bars):
label_prices, label_end_at = get_label_ticks(ticks_df, label_start_at=row['close_at'], horizon_mins=horizon_mins)
label_duration = label_end_at - row['close_at']
if label_duration < pd.Timedelta(minutes=5):
print('Dropping label, less then 5min from bar close_at:', row['close_at'])
continue
if len(label_prices) < 1:
print('Dropping label, only', len(label_prices['price']), 'trades;' 'start at:', row['close_at'])
continue
outcomes = triple_barrier_outcomes(label_prices, risk_level, reward_ratios)
label = outcomes_to_label(outcomes, label_end_at)
label.update({
'label_start_at': row['close_at'],
'label_end_at': label_end_at,
})
bars[idx].update(label)
if add_trend_label:
trend = get_trend_outcome(label_prices)
bars[idx].update(trend)
return bars