Stochastic Programming DEA Model of Fundamental Analysis of Public Firms for Portfolio Selection

A stochastic programming (SP) extension to the traditional Data Envelopment Analysis (DEA) is developed when input/output parameters are random variables. The SPDEA framework yields a robust performance metric for the underlying firms by controlling for outliers and data uncertainty. Using accounting data, SPDEA determines a relative financial strength (RFS) metric that is strongly correlated with stock returns of public firms. In contrast, the traditional DEA model overestimates actual firm strengths. The methodology is applied to public firms covering all major U.S. market sectors using their quarterly financial statement data. RFSbased portfolios yield superior out-of-sample performance relative to sector-based ETF portfolios or broader market index.