Topic : Machine Learning for Financial Applications - Stock Market Crash Prediction
Supervisor: Prof. LI Lingfei
Group ID : B
Group Members :
CHEUNG Kwong Tai, 1155142517
CHAN, Tan Fung, 1155158914
LO Yi Chun, 1155142579
This paper examines the effectiveness of machine learning techniques in predicting stock market crises based on various macroeconomic and financial variables. We trained a comprehensive set of market and fundamental predictors from 10 markets to several machine learning models, namely Log Regression (Log-R), Support Vector Machines (SVMs), Extreme Gradient Boosting (XGBoost) and Long Short-Term Memory (LSTM) network. With expanding window forecasting and validation procedures, we predict stock crash for next quarter out-of-sample. We demonstrated that Log Regression, SVM and XGBoost models outperform our benchmark model.