Equity Ranking with Stochastic Dominance and Combinatorial Fusion
Equity rating systems have been widely used by investment professionals as equityevaluation tools, however, criticism on the determination of cutoff points betweendifferent ratings remains unresolved. One could argue that it is unfair for equities ofsimilar values to be issued with different ratings. On the other hand, equity rankingcould resolve this issue by precisely sorting candidates from the strongest to weakest,subject to a certain criterion or multiple criteria. This dissertation aims at ranking U.S.publicly-traded equities, utilizing a fusion technique - “Combinatorial Fusion Analysis(CFA)” proposed by Hsu, Chung, and Kristal (2006). We combine and analyze resultsfrom multiple models using a set of regressors including selected U.S. public companies’fundamentals and technical indicators and three-month total return as our responsevariable, and meanwhile improve the identification process for the final ranking listthrough comparing the densities of top equities across all model combination resultsby examining four orders of stochastic dominance. The simulated long-only portfoliothat results from selecting the top 100 equities from the ranking lists significantlyoutperforms the iShares S&P 500 (“The Benchmark”) in both returns and Sharpe ratiowith a realized annualized alpha of 22.298% and a realized annualized beta of 1.205.
Jiang, Nan, "Equity Ranking with Stochastic Dominance and Combinatorial Fusion" (2023). ETD Collection for Fordham University. AAI30492834.