Transforming PiotroskiÕs (binary) F-score into a real one

Abstract

Since the 1960s, accounting researchers have attempted to assess the benefit of historical financial statements to capital market investors (Bunting & Barnard, 2015) by developing an indeterminate number of predictive signals designed to Òbeat the marketÓ. One such stock screen, the Piotroski F-score (Piotroski, 2000), attempted to reverse the trend of increasingly complicated predictive algorithms by applying a simple binary calculation method to nine signals extracted from widely accessible historical accounting data. While Piotroski found the F-score to be effective in separating winners from losers, subsequent studies delivered mixed results. This research sought to interrogate the relevance of the F-scoreÕs nine signals on the JSE and to assess a revised calculation methodology, based on a ranked scale of companies. The results of the research were also used to assess the ongoing relevance of accounting-based fundamental analysis in light of the challenges posed by behavioural finance research and technological advances. The results show that six of the F-scoreÕs nine signals were found to be relevant to the JSE and the ranked scale calculation approach was successful in separating winners from losers. Analysis of the F-scoreÕs predictive ability over time showed that accounting-based fundamental analysis remains relevant in the context of the JSE.