-
Notifications
You must be signed in to change notification settings - Fork 729
/
fundamental_portfolio_drl.py
128 lines (94 loc) · 4.25 KB
/
fundamental_portfolio_drl.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
import warnings
warnings.filterwarnings("ignore")
import pandas as pd
import numpy as np
from pypfopt.efficient_frontier import EfficientFrontier
from pypfopt import risk_models
from pypfopt.risk_models import CovarianceShrinkage
from pypfopt import expected_returns
from datetime import datetime
from pandas.tseries.offsets import BDay
from finrl.agents.stablebaselines3.models import DRLAgent
from finrl.meta.env_portfolio_allocation.env_portfolio import StockPortfolioEnv
from finrl.meta.preprocessor.preprocessors import FeatureEngineer
from finrl.meta.preprocessor.preprocessors import data_split
from finrl import config
import pickle
from rl_model import run_models
df_price = pd.read_csv("/home/wenbiaolin/20221117/sp500_price_19960101_20221021.csv")
df_price['adjcp'] = df_price['prccd'] / df_price['ajexdi']
df_price['date'] = df_price['datadate']
df_price['open'] = df_price['prcod']
df_price['close'] = df_price['prccd']
df_price['high'] = df_price['prchd']
df_price['low'] = df_price['prcld']
df_price['volume'] =df_price['cshtrd']
df = df_price[['date', 'open', 'close', 'high', 'low','adjcp','volume', 'gvkey']]
df['tic'] = df_price['gvkey']
df['date'] = pd.to_datetime(df['date'], format='%Y%m%d')
df['day'] = [x.weekday() for x in df['date']]
df.drop_duplicates(['gvkey', 'date'], inplace=True)
selected_stock = pd.read_csv("stock_selected_rf.csv")
trade_date=selected_stock.trade_date.unique()
with open('all_return_table.pickle', 'rb') as handle:
all_return_table = pickle.load(handle)
with open('all_stocks_info.pickle', 'rb') as handle:
all_stocks_info = pickle.load(handle)
df_dict = {'trade_date':[], 'gvkey':[], 'weights':[]}
testing_window = pd.Timedelta(np.timedelta64(1,'Y'))
max_rolling_window = pd.Timedelta(np.timedelta64(10, 'Y'))
for idx in range(1, len(trade_date)):
p1_alldata=all_stocks_info[trade_date[idx-1]]
p1_alldata=p1_alldata.sort_values('gvkey')
p1_alldata = p1_alldata.reset_index()
del p1_alldata['index']
p1_stock = p1_alldata.gvkey
earliest_date = pd.to_datetime(trade_date[idx-1]) - max_rolling_window
df_ = df[df['tic'].isin(p1_stock) & (df['date'] >= earliest_date) & (df['date'] < trade_date[idx])]
print(df_)
fe = FeatureEngineer(
use_technical_indicator=True,
use_turbulence=False,
user_defined_feature = False)
df_ = fe.preprocess_data(df_)
df_=df_.sort_values(['date','tic'],ignore_index=True)
df_.index = df_.date.factorize()[0]
cov_list = []
return_list = []
# look back is one year
lookback=252
for i in range(lookback,len(df_.index.unique())):
data_lookback = df_.loc[i-lookback:i,:]
price_lookback=data_lookback.pivot_table(index = 'date',columns = 'tic', values = 'close')
return_lookback = price_lookback.pct_change().dropna()
return_list.append(return_lookback)
covs = return_lookback.cov().values
cov_list.append(covs)
df_cov = pd.DataFrame({'date':df_.date.unique()[lookback:],'cov_list':cov_list,'return_list':return_list})
df_ = df_.merge(df_cov, on='date')
df_ = df_.sort_values(['date','tic']).reset_index(drop=True)
stock_dimension = len(df_.tic.unique())
state_space = stock_dimension
env_kwargs = {
"hmax": 100,
"initial_amount": 1000000,
"transaction_cost_pct": 0.001,
"state_space": state_space,
"stock_dim": stock_dimension,
"tech_indicator_list": config.INDICATORS,
"action_space": stock_dimension,
"reward_scaling": 1e-4
}
a2c_model,ppo_model,ddpg_model,td3_model,sac_model,best_model = run_models(df_, "date", pd.to_datetime(trade_date[idx-1]), env_kwargs,testing_window, max_rolling_window)
trade = data_split(df_, pd.to_datetime(trade_date[idx-1]), pd.to_datetime(trade_date[idx]))
e_trade_gym = StockPortfolioEnv(df = trade, **env_kwargs)
df_daily_return, df_actions = DRLAgent.DRL_prediction(
model=a2c_model, environment=e_trade_gym
)
for i in range(len(df_actions)):
for j in df_actions.columns:
df_dict['trade_date'].append(df_actions.index[i])
df_dict['gvkey'].append(j)
df_dict['weights'].append(df_actions.loc[df_actions.index[i], j])
df_rl = pd.DataFrame(df_dict)
df_rl.to_csv("drl_weight.csv")