CryptoEnv

class crypto_env.core.CryptoEnv(max_sell, max_buy, min_sell, min_buy, dataloader: crypto_env.dataloader.dataloader.DataLoader, recorder: crypto_env.recorder.Recorder)

Bases: gym.core.Env, abc.ABC

This is the core module of CrytoEnv. It provide environment for agents to perform buy and sell actions and provide market states.

__init__(max_sell, max_buy, min_sell, min_buy, dataloader: crypto_env.dataloader.dataloader.DataLoader, recorder: crypto_env.recorder.Recorder)

Parameters

• max_sell (float) – maximum crypto to sell

• max_buy (float) – maximum crypto to buy

• min_sell (float) – minimum crypto to sell

• min_buy (float) – minimum crypto to buy

• dataloader (DataLoader) – the crypto_env.dataloader.dataloader.DataLoader instance

• recorder (Recorder) – the Recorder instance

buy(value, verbose=0)

The agent buy some amount of crypto.

Parameters

• value (float) – number of crypto to buy

• verbose (int, optional) – whether to print out debug info. Defaults to 0.

Returns

same return as step()

first_observation()

Return the first observation

Returns

return a dictionary structured dict(features, index)

Return type

dict

abstract classmethod get_reward()

Returns

the reward for agent after taking an action

Return type

float

hold(verbose=0)

The agent does not want to do anything in this step

Parameters

verbose (int, optional) – whether to print out debug info. Defaults to 0.

Returns

same return as step()

meta()

Return the meta information of the environment

Returns

the meta of the env

Return type

dict

render(mode='human')

Placeholder. Not implemented yet.

Parameters

mode (str, optional) – Defaults to “human”.

reset()

Reset the environment to prepare for a new episode

Returns

Return type

CryptoEnv

sell(value, verbose=0)

The agent sell some amount of crypto.

Parameters

• value (float) – number of crypto to sell

• verbose (int, optional) – whether to print out debug info. Defaults to 0.

Returns

same return as step()

step(action=None)

Parameters

action (dict, optional) – action to take. Defaults to None.

Returns

agent’s observation after taking the action (numpy array), reward of the action (float), whether the episode is to the end (bool), and diagnostic information for debugging (any).