Machine Learning For Market Volatility Prediction

Abstract

Market volatility prediction is one of the most commonly used terms in the trading market today. Price movements, market volatility, and trading risks are all represented by realized volatility. A small change in volatility affects the expected return on all assets. In this research, To predict volatility, we will use the dataset provided by the Kaggle platform. Optiver is a leading global electronic market maker and is committed to continuously improving financial markets. improving access and prices for options. Exchange traded funds(ETFs), On numerous exchanges around the world, cash equities, bonds, and foreign currencies are traded. The prediction models we used in our study are LightGBM and XGBoost and CatBoost and Linear Regression, and we concluded some related works on forecasting volatility. Then we ran our models. The results show that the LightGBM model is the best among these models, as it achieved the lowest root mean square error percentage (RMSPE) score of: 0.286, And the highest score in the coefficient of determination

is: 0.817, and the RMSPE for other models: XGBoost, CatBoost and Linear regression is, respectively: 0.303, 0.302 and 0.347, and the score of

is also: 0.791, 0.784, 0.766.