Predicting extreme performers on the JSE securities exchange

Abstract

In this context, this thesis builds on the prior literature on extreme performance by Reinganum (1988), Glickman, DiRienzo and Ochman (2001), O'Neil (2002) and Dong, Duan and Jang (2003), where an extreme winner (loser) is a stock which at least doubles (halves) in a twelve month period. The research is conducted on the JSE Securities Exchange over the ten year period from January 1995 until December 2004. The dataset employed contains monthly data for 213 companies listed on this exchange, incorporating 7807 (5397) unique company months of extreme gain (loss). The data are adjusted for look-ahead bias but not survivorship bias.