def __init__(self,
df,
initial_balance=10000,
commission=0.0025,
reward_func='sortino',
**kwargs):
super(BitcoinTradingEnv, self).__init__()
self.initial_balance = initial_balance
self.commission = commission
self.reward_func = reward_func
self.df = df.fillna(method='bfill').reset_index()
self.stationary_df = log_and_difference(
self.df,
['Open', 'High', 'Low', 'Close', 'Volume BTC', 'Volume USD'])
benchmarks = kwargs.get('benchmarks', [])
self.benchmarks = [
{
'label':
'Buy and HODL',
'values':
buy_and_hodl(self.df['Close'], initial_balance, commission)
},
{
'label':
'RSI Divergence',
'values':
rsi_divergence(self.df['Close'], initial_balance, commission)
},
{
'label':
'SMA Crossover',
'values':
sma_crossover(self.df['Close'], initial_balance, commission)
},
*benchmarks,
]
self.forecast_len = kwargs.get('forecast_len', 10)
self.confidence_interval = kwargs.get('confidence_interval', 0.95)
self.obs_shape = (1, 5 + len(self.df.columns) - 2 +
(self.forecast_len * 3))
# Actions of the format Buy 1/4, Sell 3/4, Hold (amount ignored), etc.
self.action_space = spaces.Discrete(12)
# Observes the price action, indicators, account action, price forecasts
self.observation_space = spaces.Box(low=0,
high=1,
shape=self.obs_shape,
dtype=np.float16)
from util.benchmarks import buy_and_hodl, rsi_divergence, sma_crossover
df = pd.read_csv('./data/wdo_small.csv')
# df = df.drop(['Symbol'], axis=1)
df = df.sort_values(['Date'])
test_len = int(len(df) * 0.2)
train_len = int(len(df)) - test_len
test_df = df[train_len:]
test_df = test_df.fillna(method='bfill').reset_index()
initial_balance = 10000
commission = 0.0025
buy_and_hodl_net_worths = buy_and_hodl(test_df['Close'], initial_balance,
commission)
rsi_divergence_net_worths = rsi_divergence(test_df['Close'], initial_balance,
commission)
sma_crossover_net_worths = sma_crossover(test_df['Close'], initial_balance,
commission)
with open('./research/results/buy_and_hodl_net_worths.pkl', 'wb') as handle:
pickle.dump(buy_and_hodl_net_worths, handle)
with open('./research/results/rsi_divergence_net_worths.pkl', 'wb') as handle:
pickle.dump(rsi_divergence_net_worths, handle)
with open('./research/results/sma_crossover_net_worths.pkl', 'wb') as handle:
pickle.dump(sma_crossover_net_worths, handle)
def __init__(self,
df,
initial_balance=1,
commission=0.002,
close_key='close_price',
date_key='timestamp_ms',
annualization=365 * 24 * 12,
max_leverage=4,
training=True,
max_steps=None,
is_cnn=True,
verbose=False,
**kwargs):
super(MarginTradingEnv, self).__init__()
self.initial_balance = initial_balance
self.commission = commission
self.annualization = annualization
self.max_leverage = max_leverage
self.close_key = close_key
self.date_key = date_key
self.position = 0.1
self.training = training
self.max_steps = max_steps
self.is_cnn = is_cnn
self.verbose = verbose
self.reward_func = kwargs.get('reward_func', 'sortino')
self.window_size = kwargs.get('window_size', 20)
self.account_history_size = kwargs.get('account_history_size', 3)
self.action_type = kwargs.get('action_type', 'discrete_nl_3')
self.obs_type = kwargs.get('obs_type', 2)
# Actions of the format -1=Short 1=Long
if self.action_type == "continuous":
self.action_space = spaces.Box(low=-1,
high=1,
shape=(1, ),
dtype=np.float32)
elif self.action_type == "discrete_20":
self.action_space = spaces.Discrete(20)
elif self.action_type == "discrete_5":
self.action_space = spaces.Discrete(5)
elif self.action_type == "discrete_3":
self.action_space = spaces.Discrete(3)
elif self.action_type == "discrete_nl_3":
self.action_space = spaces.Discrete(3)
try:
self.df = df[['close_price', 'high_price',
'all_volume']].reset_index()
self.stationary_df = df.drop(
columns=['close_price', 'high_price',
'all_volume'], axis=1).reset_index()
except Exception as e:
print(df.columns)
print(e)
benchmarks = kwargs.get('benchmarks', [])
self.benchmarks = [
{
'label':
'Buy and HODL',
'values':
buy_and_hodl(self.df[self.close_key], initial_balance,
commission)
},
{
'label':
'RSI Divergence',
'values':
rsi_divergence(self.df[self.close_key], initial_balance,
commission)
},
{
'label':
'SMA Crossover',
'values':
sma_crossover(self.df[self.close_key], initial_balance,
commission)
},
*benchmarks,
]
self.obs_shape = (1, (8 * self.account_history_size) +
(len(self.stationary_df.columns) * self.window_size))
# Observes the price action, indicators, account action, price forecasts
self.observation_space = spaces.Box(low=-1,
high=1,
shape=self.obs_shape,
dtype=np.float16)
self.scaler = preprocessing.MinMaxScaler(feature_range=(-1, 1))