Ben Graham Was a Quant: Raising the IQ of the Intelligent Investor by Steven P. Greiner (Wiley, 2011) is, as intimated by the subtitle, a dense book. Unless the reader is familiar with the principles of quantitative analysis he will have to expend both time and mental energy even though the book contains virtually no math. But the effort is well worth it for anyone who wants to manage his own portfolio or who aspires to become a professional portfolio manager.
At its core Greiner’s book analyzes factors, how to build models from factors, and how to build portfolios from models. Leading up to this analysis are discussions of alpha and beta, risk, and modeling pitfalls. The two final chapters contain ruminations on assorted conceptual topics and a look back at the past and into the future.
As is my wont, I’m going to pull out a few ideas I considered particularly thought-provoking.
Greiner accepts that (1) the stock market is a complex chaotic system, (2) Soros’s reflexivity paradigm is essentially correct, and (3) the GARCH model fits the basic return distribution particularly well. If these three statements are true, the notion of mean reversion is probably, well, meaningless. “One often hears people in finance and asset managers speak of regression to the mean or mean reversion when discussing stock valuation or bond spreads. Unfortunately, in congruence with chaos theory, the notion of stock returns is more accurately termed antipersistent, meaning that there really is no mean to return to, but that, after moving in one direction, the process will soon revert. Additionally, one can readily decipher stock returns as changing direction more easily than one can say where it will revert to. … Investors do not and cannot know the actual average value of any given stock.” (p. 263)
Likewise, quants cannot know when random extinction-level events (ELE), a term Greiner prefers over “black swans,” will occur. “Quants generally are very aware of the unpredictable nature of random unforeseen events, and they do not waste their time trying to predict them. They spend their time building models to predict normal events with known causes that are predictable.” (p. 36)
Along the same lines, Greiner maintains that asset allocation did not fail the investor during the financial crisis. “There are two main categories of risk: market risk and security risk, also called systemic and idiosyncratic risk. Proper asset allocation and diversification results in minimizing idiosyncratic risk because only this kind of risk is diversifiable. … [I]n the credit crisis, when all stocks correlated highly as they do in down markets, … the total risk composed of market and security-specific risk changed their contribution percentages, so that market risk increased and security-specific risk, which is diversifiable, decreased. However, the asset allocation did its job by mitigating the security-specific idiosyncratic risk.” (p. 257)
Finally, let’s look at a topic dear to every investor’s heart: the search for alpha. Greiner outlines eight rules of thumb for deciding when an alpha signal is truly being generated by some factor. Here are three: (1) It must come from real economic variables, (2) The signal must be strong enough to overcome trading costs, and (3) It should not be misconstrued as a risk factor. What does it mean to misconstrue an alpha factor as a risk factor? Once again, let me quote Greiner: “It is safe to say that if a factor is not a risk factor, and it explains the return or the variance of return to some extent, it must be an alpha factor. … A great example is a 12-month past-earnings growth. Though returns are colinear with this earnings growth, they are poor predictors of future returns. This is because many investors will have bid up the pricing of stocks that have shown historically good earnings growth concurrently with earnings announcements. However, 12-month past-earnings growth fails miserably at being a forecaster of future return. … In this case, 12-month past-earnings growth is probably a risk factor because it regresses well with future return statistically, but offers little in the way of alpha.” (pp. 23-24)
Ben Graham Was a Quant may not be a page turner, but I think it would be an important addition to the library of anyone who is interested in building portfolio models—and not just value portfolio models.