Forecasting stock market crashes via machine learning
Abstract
This paper uses a comprehensive set of variables from the five largest Eurozone countries to compare the performance of univariate and machine learning-based multivariate models in predicting stock market crashes. In terms of statistical predictive performance, a support vector machine-based prediction model outperforms a random classifier and is superior to the average univariate benchmark as well as a multivariate logistic regression. The ability to forecast stock market crashes out-of-sample translates into large value-added for active investors. Incorporating nonlinear and interactive effects is both imperative and foundation for the predictive performance of support vector machines compared to their linear benchmark. It enables identify and explain the complex relationships in the underlying economic conditions that precede large stock market declines. Our findings can help forecast financial fragility and deteriorating macroeconomic outcomes.