Clarifications on Frequency of Algo trades about gym-anytrading HOT 1 CLOSED

1. You can call the render method of the training env, and it will plot the training positions. Moreover, you can get the required information from the history attribute of the env.

2. This plot shows positions, not actions. The position at 379 is Long, so it means the agent made a Buy action at 378.

3. I can't say if it is a best practice or not. But I think it is better to trade based on the exact timestamps in the training dataframe. For example, if your training dataframe has hourly data of these 9:30, 10:30, 11:30 points, you should trade at these timestamps in the test data.
   You can give a negative reward based on the difference between _last_trade_tick and _current_tick attributes.

from gym-anytrading.