Factor Timing, Should You Try?

These are our two cents on whether you, or we, or anyone else can pick the perfect time to increase or decrease exposure to a portfolio of factors like value, quality, or momentum.  

Here we are not talking about any individual security and whether it looks good on any one of these metrics; but instead we are discussing a diversified portfolio of many securities – so diversified that there is little-to-no idiosyncratic risk.    

There is no doubt that, historically, there have been reasonably long periods where value or quality didn’t work.  These episodes typically followed (or preceded) stages where value or quality worked extremely well.  Below we see approximations of what happened to a portfolio that was long quality (and short anti-quality, or “junk”) after the Nifty Fifty peaked in the summer of 1973; and we also show a portfolio long value and short growth during the internet bubble (yes, efficient-market friends, we are calling it a bubble).                

Value and quality of course don’t always work, and sometimes they don’t work for an inconveniently long time.  Many of you already know that we believe that timing the market is an exercise in futility.  We feel similarly about other factors (aka “spreads” in momentum, value, or quality factors, e.g. “returns of top decile quality names minus returns of bottom decile quality names”).  

Our view is that their levels vs. history aren’t necessarily helpful in predicting future factor crashes or rallies, particularly in the short or intermediate terms.  

In August of 2007, the momentum factor famously crashed. [1]  In July of ‘07, the momentum factor wasn’t screaming “short me” any more than the value factor was screaming “short me” in March of ‘99.  In fact, the value factor was actually screaming “buy me” in March of ‘99.  

Back then, I was fresh off a summer internship at Fidelity and a second year MBA student at the (now Booth) School of Business, preparing an analysis of the behavior and valuation of internet stocks (we called them “dot coms” back in the day) for Richard Thaler.  I’ve created a chart (using Ken French’s publicly available data) showing just how crazy cheap the “value spread” became by June of 1999. [2]

In June of 1999, value stocks were being destroyed in favour of very glamorous and sexy internet names.  If you added a “.com” suffix to your name, your stock could double overnight (not kidding).  Starting with the commencement of the bull market in August of 1982, and judging by the ratio of value between the top decile and bottom decile of price-to-book metrics, value stocks were in the bottom decile of their 17 year bull market history by the summer of 1999, and one prone to factor timing may have surmised that this was potentially a great time to get long value names, and to get short the dot-commers.

Not such a good idea.  The “value spread” subsequently decided to create a new bottom decile.

Over the long, long term, we actually do believe that value investing (and tenets of value investing) works, and that quality investing works, especially when the two of them are combined (in fact we believe that quality investing works over time only within the framework of value, particularly if you throw in a little momentum).  

In a related point, for those that bet on factor returns specifically, these “smart-beta” products (for lack of a better word) just made things a lot tougher for those masquerading as diversified-yet-purportedly-idiosyncratic stock-pickers, and a lot tougher to justify charging big fees for exposure to value, momentum and/or quality per se.

Anyway, in summary, if we didn’t already, we now know that value didn’t work in the late-1990s, nor did quality work in the mid-1970s.  There are also many other shorter, albeit less painful, periods where it counterintuitively would have been better to own already-expensive junk than already-cheap quality.  This stuff is volatile.  The same holds true for all factors, including market risk, betting against beta, and momentum – and they all are generally unpredictable.  

If market-timing is flipping silver dollars, then value and momentum timing is flipping quarters, quality timing is flipping dimes, and BAB timing is flipping nickels.[3]  

And flipping coins probably isn’t a sustainable strategy.