Data science project using machine learning and historical data from the Tokyo Stock Exchange to predict and maximize future returns. This project is based on the Kaggle competition: IPX-Tokyo-Stock-Exchange-Prediction. In this repository you will find a Jupyter notebook containing code and explainations for the entirety of this project including exploratory data analysis, feature engineering, machine learning model training and validation, prediction results etc.
Problem Description: In contrast to traditional stock trading, quantitative stock trading relies on trained machine learning models to predict the future performance of stocks. These predictions are used to formulate and execute trading strategies to maximize returns. As such, it is important to predict stock performance as accurately as possible. The objective of this project is to build a model that will predict future returns of Japanese stocks, and rank stocks from the Tokyo Stock Exchange (TSE) in order of predicted performance. Predictions will be generated using the LightGBM regressor. The data is from the Kaggle JPX Tokyo Stock Exchange Prediction competition. The dataset is historical data for a variety of Japanese stocks and options. Please see the link below for more information about the Kaggle competition.