Book Review: Quantitative Value
What do quant great Edward O. Thorp, behavioralist James Montier, and value investing legends Benjamin Graham and Warren Buffett have in common? These investment practitioners all make a seemingly incongruous appearance together in Quantitative Value, a new book by Wesley Gray and Tobias Carlisle. The authors’ objective — not clearly stated, alas, until the tenth chapter — is “to develop a sensible quantitative value investment strategy that will deliver returns in the real world.” The long-only strategy espoused in the book rests on a checklist of sophisticated stock screens, strict adherence to which delivers on the book’s subtitle: “A Practitioner’s Guide to Automating Intelligent Investment and Eliminating Behavioral Errors.”
In the opening chapter, Carlisle, a portfolio manager, and Gray, a hedge fund manager and assistant finance professor at Drexel University, present an elderly, quantitative Benjamin Graham. This version of Graham had succumbed to what I call the slim pickings hypothesis (SPH) of Paul Samuelson. At this point in his life, Graham had lost faith in the elaborate techniques of security analysis. His own approach had devolved into a quantitative one for creating a portfolio of 30 stocks consisting of three elements:
- a price-to-earnings ratio below 10;
- a debt-to-equity ratio less than 50%; and
- a discipline of holding stocks for two years or until they gain 50%, whichever came first.
Practically speaking, investors who mirror this approach must be able to stomach the resulting volatility. They must also possess the fortitude to let the quantitative model “do its thing” without the emotional interference that behavioral finance experts see as every investor’s worst enemy. Consistent with the views of James Montier, Gray and Carlisle argue that the best use of Kahnemman and Tversky is to computerize, thereby taking emotions off the table.
In the same chapter, Buffett and Thorp are compared in their use of imperfect information at bridge, poker, and investing. Buffett’s insistence on buying a dollar of value for fifty cents and Thorp’s focus on multiplying a tiny edge (epsilon) into hedge fund profits both rest on their ability to process imperfect information more craftily than those who are trained to see prices as right. (Unfortunately, after the first chapter, Thorp and his legendary hedge fund approach make no further appearances.)
In their second chapter, Gray and Carlisle introduce their theme and a base case — the “Magic Formula” popularized by Joel Greenblatt in The Little Book That Beats the Market. Their focus on best quality at best price builds on Greenblatt’s combination of the highest return on capital (earnings before interest and taxes (EBIT)/capital) for quality and highest earnings yield (EBIT/Total Enterprise Value (TEV)) for price. Moving beyond Greenblatt, Gray and Carlisle’s quantitative checklist of screens culminates in a 50/50 weighting on franchise power and franchise strength for quality and EBIT/TEV for price. The remainder of their book is an academically rigorous approach to beating Greenblatt’s Magic Formula. In this effort, they studiously avoid survivorship, small-sample, and look-ahead biases. In their quantitative checklist, the authors introduce research-based formulas to detect earnings manipulation and fraud, and they check their results against a five-factor pricing model that includes momentum and liquidity. Essentially, Gray and Carlisle present an excellent approach to quantitative financial statement analysis (QFSA).
The authors’ road to quantitative value and their story of how quantitative value beats Greenblatt’s Magic Formula make for very interesting reading. I particularly enjoyed the observation that “experts” can wind up underperforming the model because of emotional interference. Better to follow rules than to avoid buying low because you know too much! (Nonetheless, there is something odd about a free lunch from magic or quantitative formulas.)
Rather than thinking about such magic, one is left to wonder about the first chapter’s orientation of Buffett and Thorp processing imperfect information more craftily than the efficient markets investor. One also wonders about the transactions costs omitted from the analysis. Still, the table of selected quantitative value portfolio holdings from 1974 to 2011 (see pages 250 to 258) is quite fascinating in its own right and does show the model’s ability to unearth household names in the making. Buffett may have more turnover than implied by his target holding period of forever, but his expert approach seems quite different from this quantitative value approach with the transactions costs implied from its annual rebalancing. I would still bet on Buffett at bridge or Thorp at poker — even if their opponents had computer models to help them play their hands.
Please note that the content of this site should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute.
You did not say if it is worth a buy for individual investors like me. I read reviews to help me make decision. You went round and round but did not reach a decision.
Thanks Girish, A fair criticism. I should have mentioned that I think Gray and Carlisle’s Quantitative Value (QV) is a definite buy for individual as well as institutional investors. Each chapter contains investing gems, and I highly recommend the book. Recognizing that there is no magic bullet, I would supplement QV with the Fisher Investment series on Sectors as well as with the CFA Institute industry series: http://www.cfapubs.org/toc/ind/2013/2013/1. Best, Dennis
Putting this into practice yourself would probably not be realistic for someone without some programming chops, and a good data set. I am not yet complete with the book, but it seems as though you will need a data set comprising of 8 years of data ( anything over 4 isn’t free). In theory one could create their own data base, but this is impractical for the non-programmer. To apply these screens require advanced knowledge of some data-science-y libraries ( in python I’ll likely have to use pandas, or a language like R).
In short, it seems to target those with some background and inclination to mechanically implement these screens as they see fit. And as someone who develops software, it seems like an involved project with a good set of challenges, especially if you want to (1) test ideas for yourself and (2) tune parameters.
good luck