MultiDiscreteActions

Discrete scheme, which calculates the trade size as a fraction of the total balance for each instrument provided.

The trade type is determined by action % len(TradeType), and the trade size is determined by the multiplicity of the action. For example, 0 = HOLD, 1 = LIMIT_BUY|0.25, 2 = MARKET_BUY|0.25, 5 = HOLD, 6 = LIMIT_BUY|0.5, 7 = MARKET_BUY|0.5, etc.

Key Variables

  • _instruments
    • The exchange symbols of the instruments being traded.
  • _actions_per_instrument
    • The number of bins to divide the total balance by. Defaults to 20 (i.e. 1/20, 2/20, …, 20/20).
  • _max_allowed_slippage_percent
    • The maximum size above the current price the scheme will pay for an instrument. Defaults to 1.0 (i.e. 1%).

Setters & Properties

Each property and property setter.

  • dtype
    • A type or str corresponding to the dtype of the action_space.
  • exchange
    • The exchange being used by the current trading environment.
    • This will be set by the trading environment upon initialization. Setting the exchange causes the scheme to reset.
  • action_space
    • The shape of the actions produced by the scheme. This takes in a gym.space and is different for each given scheme.

Functions

  • reset
    • Optionally implementable method for resetting stateful schemes.
  • get_trade
    • Get the trade to be executed on the exchange based on the action provided.
    • Usually this is the way we distill the information generated from the action_space.

Use Cases

from tensortrade.actions import MultiDiscreteActions

action_scheme = MultiDiscreteActions(n_actions=20, instrument='BTC')

This discrete action scheme uses 20 discrete actions, which equates to 4 discrete sizes for each of the 5 trade types (market buy/sell, limit buy/sell, and hold). E.g. [0,5,10,15]=hold, 1=market buy 25%, 2=market sell 25%, 3=limit buy 25%, 4=limit sell 25%, 6=market buy 50%, 7=market sell 50%, etc…