Modelo de Seleção de Carteiras Baseado em Erros de Predição

Abstract

This work presents a new prediction errors-based portfoliooptimization model that cap-tures short-term investment opportunities. We used autoregressive moving references neuralnetwork predictors to predict the stock’s returns and derived a risk measure based on thepredictor’s errors of prediction that maintains the same statistical foundation of the mean-variance model. The efficient diversification effects hold by selecting predictors with lowand complimentary error profiles. A large set of experimentswith real data from the Brazil-ian stock market was employed to evaluate our portfolio optimization model, which includedthe examination of the Normality of the errors of prediction. Our main results showed that itis possible to obtain Normal prediction errors with non-Normal series of stock returns, andthat the prediction errors-based portfolio optimization model better captured the short termopportunities, outperforming the mean-variance model andbeating the market index.