Predictive Modeling: An Optimized and Dynamic Solution Framework for Systematic Value Investing

摘要

This paper defines systematic value investing as an empirical optimizationproblem. Predictive modeling is introduced as a systematic value investingmethodology with dynamic and optimization features. A predictive modelingprocess is demonstrated using financial metrics from Gray & Carlisle andBuffett & Clark. A 31-year portfolio backtest (1985 - 2016) comparesperformance between predictive models and Gray & Carlisle's Quantitative Valuestrategy. A 26-year portfolio backtest (1990 - 2016) uses an expanded set ofpredictor variables to show financial performance improvements. This paperincludes secondary novel contributions. Quantitative definitions are providedfor Buffett & Clark's value investing metrics. The "Sak ratio" is proposed asan extension to the Benjamini-Hochberg procedure for the inferentialidentification of false positive observations.