TensorforceTradingStrategy

A trading strategy capable of self tuning, training, and evaluating with Tensorforce.

Class Parameters

• environment
  • A TradingEnvironment instance for the agent to trade within.
• agent
  • A Tensorforce agent or agent specification.
• model
  • The runner will automatically save the best agent
• policy
  • The RL policy to train the agent’s model with. Defaults to ‘MlpPolicy’.
• _model_kwargs

Properties and Setters

• environment
  • A TradingEnvironment instance for the agent to trade within.

Functions

• restore_agent
  • Deserialize the strategy’s learning agent from a file.
  • parameters:
    • path
      • The str path of the file the agent specification is stored in.
• save_agent
  • Serialize the learning agent to a file for restoring later.
  • parameters:
    • path
      • The str path of the file to store the agent specification in.
• tune
  • Function NotImplemented
• run
  • Runs all of the episodes specified.
  • parameters:
    • steps
    • episodes
    • episode_callback

Use Cases

Use Case #1: Run a Strategy

from stable_baselines import PPO2
from tensortrade.strategies import TensorforceTradingStrategy


agent_spec = {
    "type": "ppo_agent",
    "step_optimizer": {
        "type": "adam",
        "learning_rate": 1e-4
    },
    "discount": 0.99,
    "likelihood_ratio_clipping": 0.2,
}

network_spec = [
    dict(type='dense', size=64, activation="tanh"),
    dict(type='dense', size=32, activation="tanh")
]

a_strategy = TensorforceTradingStrategy(environment=environment,
                                        agent_spec=agent_spec,
                                        network_spec=network_spec)
a_strategy.run(episodes=10)

Use Case #2: Run a Live Strategy

import ccxt
from tensortrade.environments import TradingEnvironment
from tensortrade.strategies import StableBaselinesTradingStrategy
from tensortrade.exchanges.live import CCXTExchange
coinbase = ccxt.coinbasepro(...) # your credentials go here in dictionary form
exchange = CCXTExchange(exchange=coinbase,
                        timeframe='1h',
                        base_instrument='USD',
                        feature_pipeline=feature_pipeline)

environment = TradingEnvironment(exchange=exchange,
                                 action_strategy=action_strategy,
                                 reward_strategy=reward_strategy)

agent_spec = {
    "type": "ppo_agent",
    "step_optimizer": {
        "type": "adam",
        "learning_rate": 1e-4
    },
    "discount": 0.99,
    "likelihood_ratio_clipping": 0.2,
}

network_spec = [
    dict(type='dense', size=64, activation="tanh"),
    dict(type='dense', size=32, activation="tanh")
]

a_strategy = TensorforceTradingStrategy(environment=environment,
                                        agent_spec=agent_spec,
                                        network_spec=network_spec)

strategy.environment = environment
strategy.restore_agent(path="../agents/ppo_btc/1h")
live_performance = strategy.run(steps=0, trade_callback=episode_cb)