Window Dressing, Data Mining, Or Data Errors: A Re-Examination Of The Dogs Of The Dow Theory

Abstract

We re-examine the Dow dividend yield anomaly to ascertain if data errors create the superior returns of the trading rule. Empirical testing, using both parametric and nonparametric methods, suggests that the trading rule outperforms the index. Additionally, data errors are not the drivers of superior trading rule returns. Moreover, the Chow breakpoint test of structural stability suggests that neither window dressing nor data mining explain this phenomenon. Finally, a turn of the year formation date fails to explain superior trading rule returns, further mitigating the data mining explanation.