Modeling the S&P 500 index using the Kalman filter and the LagLasso

This article introduces amethod to predict upward and downward monthly variations of the Standard & Poor´s 500 (S&P 500) index by using a pool of macro-economic and financial explanatory variables. The method is based on the combination of a denoising step, performed by Kalman filtering, with a variable selection step, performed by a Lasso-type procedure. In particular, we propose an implementation of the Lasso method called LagLasso which includes selection of lags for individual factors. We provide promising backtesting results of the prediction model based on a naive trading rule.