AI4Finance-Foundation / AI4Finance-Foundation/FinRL

Predictons same with StockTradingEnv, using 1 stock

未关闭
#886 7 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看
bug
主要语言
Jupyter Notebook
星标
16.3k
派生
3.5k
PR 合并指标
30 天内没有已合并 PR

描述

**Describe the bug**
I am using your your colab guide here https://github.com/AI4Finance-Foundation/FinRL/blob/master/Stock_NeurIPS2018_SB3.ipynb

I am training the model with PPO agent and using DRLAgent.DRL_prediction to make predictions. Also I am using only 2 stocks ['ADS.DE', 'ALV.DE']

I didn't change nothing inside StockTradingEnv. I just removed from the Environment instantiation for the test the turbolance and risk params
![image](https://user-images.githubusercontent.com/21694707/218275559-fe9d95be-4c9b-49aa-a35b-e4e9065b72c8.png)
These two.

The rest is all the same.

I added a print in DRLAgent.Prediction to print the action while predicting on test data.

This is the result: https://nekobin.com/hapodiniba

Why most of the time the predictions are same? And why if I only use one stock predictions are always same and never change ?

贡献指南

这个仓库没有索引到贡献指南

评估

这个 Issue 还没有评估数据。

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。