Method step

tf2.0/rl_trader.py:174–200

(self, action)

  def step(self, action):
    assert action in self.action_space

    # get current value before performing the action
    prev_val = self._get_val()

    # perform the trade
    self._trade(action)

    # update price, i.e. go to the next day
    self.cur_step += 1
    self.stock_price = self.stock_price_history[self.cur_step]

    # get the new value after taking the action
    cur_val = self._get_val()

    # reward is the increase in porfolio value
    reward = cur_val - prev_val

    # done if we have run out of data
    done = self.cur_step == self.n_step - 1

    # store the current value of the portfolio here
    info = {'cur_val': cur_val}

    # conform to the Gym API
    return self._get_obs(), reward, done, info


  def _get_obs(self):