SGX Full OrderBook Tick Data Trading Strategy

Modeling High-Frequency Limit Order Book Dynamics Using Machine Learning

• Framework to capture the dynamics of high-frequency limit order books.

  <img src="./Graph/pipline.png" width="650">

Overview

In this project I used machine learning methods to capture the high-frequency limit order book dynamics and simple trading strategy to get the P&L outcomes.

• Feature Extractor

  • Rise Ratio

    <img src="./Graph/Price_B1A1.png" width="650">

  • Depth Ratio

    <img src="./Graph/depth.png" width="650">

    [Note] : [Feature_Selection] (Feature_Selection)

• Learning Model Trainer

  • RandomForestClassifier
  • ExtraTreesClassifier
  • AdaBoostClassifier
  • GradientBoostingClassifier
  • SVM

• Use best model to predict next 10 seconds

  <img src="./Graph/CV_Best_Model.png" width="650">

• Prediction outcome

  <img src="./Graph/prediction.png" width="650">

• Profit & Loss

  <img src="./Graph/P_L.png" width="650">

  [Note] : [Model_Selection] (Model_Selection)