Previsibilidade em modelos de ações / Predictability in stock models




In this paper, we check the predictability in the Brazilian stock market. Differently from many studies made in this area that deal with the original time series, we base our study in synthetic series constructed from linear models of stocks. According to Burgess (1999), we use the stepwise regression methodology in order to generate each stocks models. Then, we use the variance ratio profile and a Monte Carlo simulation for the selection of the models with potential predictability. Unlike Burgess (1999), we perform a Whites Reality Check to validate the predictability in the out of sample period. We used the strategies proposed by Sullivan, Timmermann and White (1999) and Hsu and Kuan (2005), with a total of 26410 different simulated rules. Finally, we use a bootstrap methodology with 1000 simulations and found strong evidences that the stocks models have significant predictability including transaction costs


reality check análise técnica reality check bootstrap technical analysis previsibilidade predictability bootstrap monte carlo simulation perfil de razão de variância variance ratio profile simulação de monte carlo economia

Documentos Relacionados