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)
def __init__(self,
df,
reward_func='sortino',
initialFunds=10000,
commission=0.0025,
forecastLength=10,
confidenceInterval=0.95,
scaler=preprocessing.MinMaxScaler()):
"""
:param df:
:param reward_func: Risk-adjusted return metric to be used. # Obligatory argument for gym.Env
:param initialFunds: Initial resources the agent holds in USD.
:param commission: Commission the "middleman" takes per transaction.
:param forecastLength:
:param confidenceInterval:
:param scaler: Function to rescale data, default=minMaxScaler to normalize
"""
# Set constant values, and functions to use throughout
self.initialFunds = initialFunds
self.commission = commission
self.reward_func = reward_func # ['sortino', 'calmar', 'omega', 'dummy']
self.forecastLength = forecastLength
self.confidenceInterval = confidenceInterval
self.scaler = scaler
self.strategies = {
'sortino': sortino_ratio,
'calmar': calmar_ratio,
'omega': omega_ratio
}
# Prepare datasets
self.df = df.fillna(method='bfill').reset_index(
) # Treat NaN (fill with next !NaN val) and reset index accordingly
self.stacionaryDf = log_and_difference(
self.df, ['Open', 'High', 'Low', 'Close'])
# Agent spaces definition
self.action_space = spaces.MultiDiscrete(
[3, 10]
) # spaces.Discrete(12) # fixme - not being done by the current model on github
self.obsShape = (1, 5 + len(self.df.columns) - 2 +
(self.forecastLength * 3)) # fixme - why the 5 -2 ...
# print(self.obsShape)
# Observes the OHCLV values, net worth, and trade history.
# A box in R^n with an identical bound for each dimension
self.observation_space = spaces.Box(low=0,
high=1,
shape=self.obsShape,
dtype=np.float16)
def __init__(self,
data: pd.DataFrame,
feature_columns: List[str],
initial_balance: float = 10000,
commission: float = 0.0025,
reward_function: str = 'sortino',
returns_lookback: int = 100,
trade_on_open: bool = True) -> None:
"""
:param data: pandas DataFrame, containing data for the simulation,
'open', 'high', 'low', 'close', 'volume' columns should be present
:param feature_columns: initial trading balance
:param initial_balance: initial trading balance
:param commission: commission to be applied on trading
:param reward_function: type of reward function, calmar, sortino and omega allowed
:param returns_lookback: last values in portfolio to be used when computing the reward
:param trade_on_open: Use next entry open price as price to open/close positions
"""
super(TradingEnv, self).__init__()
self._logger = get_logger(self.__class__.__name__)
self._feature_cols = feature_columns
self._check_initial_data(data, cols=self._feature_cols)
self._data = data
self._scaled_data = log_and_difference(data, feature_columns)
self._initial_balance = initial_balance
self._commission = commission
self._reward_function = reward_function
self._returns_lookback = returns_lookback
self._trade_on_open = trade_on_open
# state and action spaces
self._obs_shape = (1, len(self._feature_cols)
) # todo: add account info here
self.observation_space = spaces.Box(low=0,
high=1,
shape=self._obs_shape,
dtype=np.float32)
self.action_space = spaces.Discrete(n=3)
# placeholders
self.current_step = 0
self.cash = self._initial_balance
self.position = None
self.position_history = [None]
self.portfolio = []
self.viewer = None
def __init__(self, df, initial_balance=10000, commission=0.0003, 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')
self.df = add_indicators(self.df.reset_index())
self.stationary_df = log_and_difference(
self.df, ['Open', 'High', 'Low', 'Close', 'Volume BTC', 'Volume USD'])
self.n_forecasts = kwargs.get('n_forecasts', 10)
self.confidence_interval = kwargs.get('confidence_interval', 0.95)
self.obs_shape = (1, 5 + len(self.df.columns) -
2 + (self.n_forecasts * 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)