A Study of Stock Market Volatility Prediction Based on Traditional Regression and Intelligent Algorithms

Abstract

China's stock market has been turbulent since the reform and opening up, often with the risk of plummeting, and in recent years it has been even more volatile, while the national level's concern for stock market stability is increasing day by day, and research on the stability of China's stock market is imminent. In this paper, on the basis of previous research and traditional macro fundamental analysis, firstly, we use LASSO regression analysis to initially determine the influence indicators that are causally related to stock market stability, and then test the significance of the model coefficients, and amend the original model by using multivariate linear regression, to further screen out the significant influencing factors; secondly, we use support vector machine regression (SVM) and random forest regression to fit the stock market volatility in the machine learning method. regression to fit the stock market volatility, determine the importance ratio of different characteristic variables in the model, and analyse the factors affecting the prediction of stock market volatility; finally, the LASSO regression is combined with machine learning to establish an improved model, and the screened indicator factors are fitted with machine learning models to further deepen the prediction of stock market volatility. Stock market oscillations and volatility are further deepened to measure the stability of the Chinese stock market.