At the close of the last post in the series, we defined what we meant by beating the market, that is, achieving materially higher returns than the S&P 500, with no more risk, over a typical investor’s time horizon. We also specified that we couldn’t cheat. Today, I want to take a closer look at what cheating might mean.
Some specific examples are easy. We could mess with time frames. If you look at pretty much any strategy from 1995 to 1999, from 2004 to 2007, or (for that matter) from 2009 to present, they look reasonably good. In a bull market, lots of things work that blow up in a bear market. To avoid this form of cheating, we will specifically have to crash test any strategy through at least one major bear market.
The other easy example is leverage, and this one works especially well with time frame cheating. If you borrow money to invest and the market goes up, you look like a genius! If you bought the S&P 500 in 2009, for example, and leveraged it three times, you would have had annual returns just under 30 percent per year—which certainly beats the market. The problem? At various points, you would have been down more than 60 percent (a big number). And if we got another 1987, you would have been bankrupt. Borrowing money can work, but it also substantially magnifies your risk.
More generally, cheating involves doing something that, for whatever reason, isn’t repeatable. Bull market returns don’t continue indefinitely. The returns from leverage are dependent on there being no major downturns, which is certainly not sustainable indefinitely. Any real strategy has to work in all environments or, at a minimum, be designed to survive adverse conditions.
A more subtle example of cheating is “investing” in stocks that are not really tradeable. There are a number of stock screens floating around on the internet that have phenomenal returns. But they are based on small, very illiquid stocks that would almost certainly not have allowed meaningful investment at those prices. If you couldn’t really do it, you can’t call it beating the market.
Which brings us to back testing. You will never see a bad back test, as the saying goes, because if it is bad, no one will talk about it. At the same time, we have to be able to go back and test strategies. What to do? How do we keep ourselves from being fooled?
The first line of defense in back testing is what we have already discussed. Is this a full market cycle, peak to peak or trough to trough? Preferably more than one? Does this strategy include leverage? If so, does it account for the costs? Does it actually use liquid investments, which could have been traded at the stated prices? If so, you might have something. Beyond that, asking whether the strategy actually makes sense at a fundamental level is a surprisingly useful tool because many do not once you start digging.
Regular readers will know these are essentially the same problems I faced when I started the low-drawdown program I outline in my book, Crash-Test Investing, and that’s no accident. That research program was designed around risk, but it faced the same challenges and had to meet the same metrics. This project, focused on returns, is different—but fundamentally the same.
This framework is how I plan to be honest with the objective, with myself, and, not least, with you as we work through this challenge.