<h1 align="center">MMLab Trading: An Algorithmic Trading Framework</h1> <p align="center"> <b>MMLab Trading is still in Beta, meaning it should be used very cautiously if used in production, as it may contain bugs.</b> </p> <p align="center"> <img width="400" align="center" src="Image/white_logo.png" alt="Logo MMLab"/> </p> <p> <img alt="Version" src="https://img.shields.io/badge/version-1.0.0-blue.svg?cacheSeconds=2592000" /> <a href="https://twitter.com/manhkhanhad" target="_blank"> <img alt="Twitter: manhkhanhad" src="https://img.shields.io/twitter/follow/manhkhanhad.svg?style=social" /> </a> </p>

MMLab Trading is an open source Python framework apply reinforcement learning for automatic stock trading and Linear Programming for portflio optimization.

🔧 Install

1. Clone repository:

git clone https://github.com/manhkhanhad/AlgorithmicTrading

2. Install dependency

pip install -r requirement.txt

✨ Getting Started

Portfolio Optimization

1. Download stock historical prices and Market Index (i.e VNIndex) into Data folder
2. Setting the hyperparameter in PortfolioOptimization/config_LP.yaml

Evaluate portfolio optimization

python LinearOptimization.py

<p align="center"> <img width="600" align="center" src="Image/evaluation.png" alt="Evaluate portfolio optimization"/> </p>

Running

4. Train the Senario Classifier for determining whether the current is good time to start inverst

python SenarioClassification.py

5. Optimize the stock portfolio

python app.py

Stock trading

0. Setting the hyperparameter in config.yaml
1. Download stock historical prices and Market Index (i.e VNIndex) into Data folder
2. Preprocess data

python data_processer.py

3. Training

python train.py

4. Testing

python test.py

The strcuture of result folder will be organize as follow

Result folder  
│
└───1st Period
│   ├── result.png # Portpolio value of all agent during trading period
│   ├── result.html
│   └─── Agent 1 
│       ├── sell_buy.csv  #Trading action of agent
│       ├── account_value.csv #Portpolio value during trading
│       └─── trading action
│   		├── html
|			|	├── stock_1.html #Visualize trading action of each stock
|			|   ...
|    		└── png
|				├── stock_1.png
|				....
...

Backtesting

Backtesting is an evaluation technique that simulates the use of an agent in a live trading environment. First, we train all agents for a certain period, and then we divide the remaining time into short periods with specified window size. We have all agents trade on a small-time period and save their actions and portfolio values. We then update the training set with the previous trading period and retrain all agents. This process is repeated until there are no trading periods left.

<p align="center"> <img width="600" align="center" src="Image/backtesting.png" alt="Backtesting"/> </p>

👤 Author

Manh-Khanh Ngo Huu manhkhanhad

🔰 References

Acknowledgement

Give a ⭐️ if this project helped you!