The Trick to Making Better Predictions
"It's tough to make predictions, especially about the future."
Last week, I was in San Francisco speaking to 700 financial planners from northern California. The areas that seemed to generate the most feedback and questions were on expert commentary and on risk.
People love forecasts. They shouldn't but it's how they're wired. When it comes to forecasting the future, people have two strange and internally inconsistent perspectives. First, most of us know thatforecasts are folly: Experience teaches that the people who make these guesses about the future aren't very good at it. Economists, strategists and analysts have amassed a track record that is more or less abysmal.
The second aspect of forecasting is even more fascinating: As consumers of predictions, we want them to be bold and precise, and informed by very confident forecasters. If they happened to have gotten a big outlier right, we like them even more.
What makes these circumstances so odd is how little we seem to care about the track record of these prognosticators. To be polite, it is embarrassing how wrong they have been as a group. Yet, that seems irrelevant to the public, who prefer specificity over uncertainty, confidence over humility, accuracy be damned.
This quirk has influenced a generation of soothsayers, who are often wrong, but never in doubt. Marc Faber has been calling for an imminent, 1987-like crash for several years. Nouriel Roubini had been forecasting a different type of crash for many years before 2008 rolled around -- and for completely different reasons than he had warned. Peter Schiff has been predicting hyperinflation and gold at $5,000 an ounce for almost as long. Yet they are among the most sought after pundits and television guests.
If you like, you can make bold, specific forecasts like these gentlemen. Or, you make better predictions than they do by learning a few simple insights. These aren't tricks, but rather, a framework for understanding the inherently unknowable nature of the future.
Learn these three techniques, and you can outperform the big bold forecasters of today.
1) Be less specific: There is an old joke amongst technicians: Forecast the price level or the date, but never both at once.
Economics and markets are complex realms, filled with a near-infinite number of variables. These all influence each other. Even a modest change in one can cascade across a variety of other variables. This is why precise forecasts are nothing more than guesses. Rather than try to do the impossible, instead, learn to think about the future in terms of multiple variables and possible outcomes.
Years ago, when I first started boating, the process of docking was nerve-racking. Between the currents, winds -- and that brand new expensive yacht next to my slip -- my attempts at docking were awful. An old salt saw my efforts, and told me the secret. Don't aim for impossible precision; instead, get close enough to throw a rope to someone on dock. It worked wonders, both in the marina, and when making forecasts.
2) Understand statistical analysis: You don't need a doctorate in applied mathematics to understand the basics of looking at the world of investments from a quantitative prospective. You only need a basic familiarity with a spreadsheet, and a willingness to roll up your sleeves.
A recent Morningstar study provides a perfect example of how to use this approach. They reviewed thousands of funds, and discovered if they controlled for just one factor -- cost -- they could forecast quintiles of outperformance. The cheapest fifth of funds outperformed the second fifth, which then outperformed the third fifth, etc.
You don't need to find a Svengali to select the top-performing fund; rather, you need only to learn the quantitative characteristics of funds to avoid the under-performers.
3) Probability theory can help determine probable outcomes: Many investors are using probability theory and may not even know it. Value investors are buying the lowest-cost equities based on their valuations. It states that cheaper is better than more expensive. This is a simple probability forecast that more expensive stocks will perform worse than cheap stocks. It is based on previous returns of those stocks bought at a discount to fair value.
There is no guarantee that this will happen, only a past correlation -- cheap has outperformed expensive -- and a likely reversion to the mean.
Investing would be easy if you could predict the future, but the reality is you can't with any degree of consistency or accuracy. No one can. Instead, focus on the numbers, using probability to determine likely outcomes.
This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.
To contact the author on this story:
Barry L Ritholtz at firstname.lastname@example.org
To contact the editor on this story:
James Greiff at email@example.com