Performance measurement of portfolios constructed by the single index model with historical and adjusted betas

Abstract

This study investigates the performance of portfolios constructed by single index model with historical (least squares regression) betas and estimated future betas by Vasicek's Bayesian Estimation Technique. The performances of the portfolios are measured by Sharpe's reward to variability ratio. The portfolios constructed by the simple criteria for portfolio selection of Elton, Gruber, Padberg have outperformed the market, but not at a statistically significant level. This is caused by the low market volume and high volatility of the market. The correlation between the market and the stocks turned out to be very low. Also the betas of the stocks were very volatile. Previous studies have shown that Vasicek's adjusted beta outperforms the historical one, but in this study, this could not be shown for the Istanbul Stock Exchange.