| Resumo: | This work project compares simple linear predictive regressions and regime switching predictive regression, to analyse time varying predictability of the stock returns in XU100 index. The approach is to compare regression statistics in-sample and to compare error statistics and predictive graphs out-of-sample. Findings reveal that employing predictive regime switching models reflect arbitrage opportunities better than predictive linear models in-sample. Out-of-sample analysis, however, provide no evidence for high predictive power for either predictive linear models or regime switching models.
|
|---|