US Quantitative Modeling
Our firm has decades of experience in developing quantitative models for equity stock selection, with two of our principals, Michael Goldstein and Brian Cho, having worked together on these issues for the last 20 years. Empirical’s modeling philosophy is rooted in behavioral finance and seeks to capitalize on recurring aspects of human nature. Part of the process of building a quantitative framework is deciding what is recurring. Our models take advantage of biases in investor behavior as manifest in valuation relationships and in their reaction to fundamentals in the equity market. We also seek to exploit the actions of corporate managements by monitoring how they deploy and finance capital, and the psychology both reveal. Measures of the quality of earnings complete the picture. These building blocks of our approach are computed both cross- and intra-sectorally, yielding a scheme without obvious systematic biases. They then are combined in a dynamic fashion, emphasizing those that offer the highest expected returns. We believe this dynamic weighting approach is a source of significant out-performance over a market cycle.
We believe the most important criteria in evaluating a partner for quantitative modeling are the quality and depth of thinking. Many organizations can process data, but in the end what counts is judgment. We believe the right way to build a quantitative stock selection model is to start with ideas and test them. Opinions and judgment are the critical parts of the process, which is itself open-ended in nature. Creativity, combined with analytic discipline, is the key to success. Our firm has done a large body of work on what characterizes failed stocks and how to avoid them. We believe that loss avoidance is particularly critical when choosing among either high-growth or mega-cap stocks. We’ve also done a great deal of work on the character of factor returns over multi-year holding periods and optimizing returns over investment, rather than trading horizons. We also believe we have unmatched expertise in combining conflicting factors into a coherent approach. Our models are the day-to-day application of what we’ve learned so far. They’re also sector specific, which has historically maximized the likelihood of finding winners. Our models have produced consistent returns over time and across sectors. In addition to our general stock selection models, we offer failure models that highlight potential big losers as well as a separate model gauging the attractiveness of big top-line growers. Finally, our model outputs can be tailored to all public companies listed domestically as well as ADRs.