Why We Can't Predict Financial Markets

clipped from blogs.harvardbusiness.org Harvard Business Publishing 4:22 PM Thursday January 22, 2009 by Jeff Stibel Remember the saying that "past performance is not an indicator of future success"? This quote shows up at the end of ads for mutual funds, hedge funds -- really, any promo for investment advice. The irony of that statement is, of course, that it came after you'd just been told how well that investment had done in the past (a wink-wink, nod-nod promise that the future would indeed be very much like the past). Of course, as recent events have confirmed all too well, the past really isn't a good indicator of the future. So how do we accurately predict where the markets are headed? The truth is, we can't. The future, like any complex problem, has far too many variables to be predicted. Quantitative models, historical models, even psychic models have all been tried -- and have all failed. Just imagine predicting something far simpler than the future of the stock market; say, chess. There are an overwhelming 10 to the 120th power possible moves. That's a 1 followed by 120 zeros! As James Hogan explains it in his book Mind Matters, that sum far exceeds the number of atoms in the universe. The best prediction machines known to us are actually our own brains. Many people think of the brain as either a really powerful computer or just something too mysterious to explain. But it turns out that the brain is not mysterious, nor is it a computer; it is instead a damn good prediction machine. That is one of the reasons we do well at games like chess or baseball. While a human brain cannot calculate a mathematical equation as quickly as even the most basic calculator, it can easily determine where a ball in mid-flight will land -- without calculating its precise trajectory or velocity, as a computer would do. Could you imagine trying to instantaneously calculate where a fly ball will land? Of course not. But I bet you could catch it. Our brains are great at what they do because they make educated guesses -- but that also makes us vulnerable to errors in judgment. Nowhere is this more pronounced than when we try to forecast the future. The human brain is great at predictions but horrible at long-range forecasting. This is why we have no problem anticipating that the slithering stick on the ground might bite us (and can jump out of the way in a millisecond) but have so much trouble guessing where the snake will be the next time we go out into the yard (we almost always guess incorrectly and avoid the same spot). This is one of the reasons Nassim Taleb argues in The Black Swan that we are guilty of ascribing far too much predictability to the truly unpredictable. The real rub on Wall Street is that economic and stock models play on our biases. We believe that models that have accurately predicted the future in the past, are likely to predict the future going forward. But that is no more true than believing me when I tell you that a coin will land heads up just because I accurately predicted it would do so the last ten times. Even the most failsafe predictions aren't accurate over the long-term. Remember these? The stock market never declines over periods longer than 10 years; real estate prices never decline nationally; AAA rated securities have always had a lower default rates than A rated securities; I do not have a history of raising taxes so trust me when I say, "read my lips, no new taxes." To be sure, some models do stand the test of time and there are a few smart people out there who are better at predicting the future than the likes of you and me. But in general, we stink at forecasting. Lest I leave you with a bad taste in your mouth, allow me to instead offer you one model that seems to be doing a very good job these days. This model has accurately predicted the stock market over the past 15 years: ups, downs and in-betweens. And the kicker is that it even predicted the steep recession we are all now living with (and is currently indicating a depression, although I suspect a comeback is likely). It is one of the most remarkable models I have seen and it continues to produce mind-numbingly accurate forecasts. And the best thing about the model is that it is free to anyone, easy to understand, and fun to watch. So what is this model? It is called the Tiger Bears and Bulls Index. Using a simple polynomial trend analysis, the model tracks Tiger Woods' performance (tournament placement) against the Dow. The results are striking: the model predicted virtually every peak and trough of the Dow Industrials average since 1993 and continues to do so with amazing accuracy. How could it be that this model does as well as the most ardent stock guru, quant model, or economist? Maybe it has something to do with what I said in the first few paragraphs -- remember -- that predicting the future is really not something human beings are very good at. Other than Tiger Woods, of course. *Special thanks to my brother, Aaron Stibel, who built the Tiger Woods' Bears and Bulls Index.