Add overview

docs/source/envs/overview.rst

@@ -2,3 +2,50 @@
 Overview
 =================================================
 
+In recent years, financial reinforcement learning (FinRL) with deep reinforcement learning (DRL)
+has shown huge potentials in obtaining profitable trading strategies in complex market conditions.
+The inherently noisy and complex real-world markets makes DRL a compelling approach to adapt
+and learn the optimal trading strategy. However, to effectively learn from the noisy market data,
+market environments are essential to provide a close-to-real simulation of complex markets.
+
+Despite the importance, many open-source market environments have remained outdated and in-
+consistently maintained. Differences in data sources and API design hinder reproducibility and
+makes it difficult to integrate and test agents across environments. Moreover, community initiatives
+such as FinRL Contests require standardized market environments to ensure fair and reproducible
+evaluation of agents. The challenges highlights the demand to organize current open-source market
+environments
+
+
+.. image:: ../image/Market_Environments.png
+   :width: 100%
+   :align: center
+
+.. [1] AminHP. *gym-anytrading: A collection of OpenAI Gym environments for reinforcement learning-based trading.*, 2025.  
+   URL https://github.com/AminHP/gym-anytrading. GitHub repository, MIT license. Accessed 2025-08-20.
+
+.. [2] ClementPerroud. *Gym-Trading-Env: A simple, easy, customizable Gymnasium environment for trading.*, GitHub repository, 2025.  
+   URL https://github.com/ClementPerroud/Gym-Trading-Env. Accessed 2025-08-20.
+
+.. [3] drlove2002. *"reliance.csv" from TradingGym repository: Stock price dataset for reinforcement learning environment.*, GitHub file in drlove2002/TradingGym repository, 2025.  
+   URL https://github.com/drlove2002/TradingGym/blob/master/trading_gym/data/reliance.csv. Accessed 2025-08-20.
+
+.. [4] kh-kim. *stock_market_reinforcement_learning: Stock trading environment using OpenAI Gym with deep Q-learning and policy gradient.*, GitHub repository, 2025.  
+   URL https://github.com/kh-kim/stock_market_reinforcement_learning. Accessed 2025-08-20.
+
+.. [5] Xiao-Yang Liu, Ziyi Xia, Jingyang Rui, Jiechao Gao, Hongyang Yang, Ming Zhu, Christina Wang, Zhaoran Wang, and Jian Guo.  
+   *FinRL-Meta: Market environments and benchmarks for data-driven financial reinforcement learning.*  
+   *Advances in Neural Information Processing Systems*, volume 35, pages 1835–1849, 2022.
+
+.. [6] Shuo Sun, Molei Qin, Wentao Zhang, Haochong Xia, Chuqiao Zong, Jie Ying, Yonggang Xie, Lingxuan Zhao, Xinrun Wang, and Bo An.  
+   *TradeMaster: A holistic quantitative trading platform empowered by reinforcement learning.*  
+   *Advances in Neural Information Processing Systems*, volume 36, pages 59047–59061, 2023.
+
+.. [7] tensortrade-org. *TensorTrade: An open-source reinforcement learning framework for trading.*, GitHub repository, 2021.  
+   URL https://github.com/tensortrade-org/tensortrade/tree/master. Accessed 2025-08-20.
+
+.. [8] Xiao Yang, Weiqing Liu, Dong Zhou, Jiang Bian, and Tie-Yan Liu.  
+   *Qlib: An AI-oriented quantitative investment platform.*, 2020.  
+   URL https://arxiv.org/abs/2009.11189.
+
+
+

docs/source/envs/stock/gymanytrading.rst

@@ -40,8 +40,8 @@ At each timestep, the environment performs the following updates:
 1. **Decode Actions (Discrete)**
 
    The action for each asset is a discrete flag:
-   ``0`` = **SELL/close long**, ``1`` = **BUY/open (or add to) long**.
-   Current position flags are kept in ``self.position`` (``0`` = flat, ``1`` = long).
+   ``0`` = **SELL/close long**, ``1`` = **BUY/open long**.
+   Current position flags are kept in ``self.position`` (``0`` = short, ``1`` = long).
 
    .. code-block:: python
 

docs/source/index.rst

@@ -43,7 +43,7 @@ FinRL Contest 2023: `Contest Website <https://open-finance-lab.github.io/finrl-c
    intro/agent
 
 .. toctree::
-   :caption: Environments
+   :caption: Market Environments
    :hidden:
    :maxdepth: 2