FYP_CUHK_CSCI4998

Features

• SVM
• SVM + Sliding window (3 labels)
• SVM + Sliding window (5 labels)
• SVM + Sliding window + PCA
• SVM + Sliding window + PCA + Condition

File Structure

1. Collect Data
   • Collect orderbook data: dowload_order_book.py
   • Collect fear and greed index: select_fng.py
   • Calculate threshold: calculate_threshold.py
2. Data preprocessing
   • Split data into windows: interval_split.py
3. Model
   • Select features, labels and construct original SVM model: svm.py
   • Build self-designed model from original SVM model: model_builder.py
4. Backtesting
   • From prediction results to investment positions: generate_pos.py
   • Take actions of positions: trading.py
   • Evaluate performances: backtesting.py (The full version of backtesting is not used in main.py, but write a simplified version in main.py.)

Branch Description

1. Master
2. CPU_Version The version to be deployed locally
3. GPU_Version The version to be deployed on Colab