Dumb Alpha: Are Your Forecasts Better Than a Random Walk?
This series spotlights investment techniques that are so simple they seem “dumb” and often surprise us by how well they work in practice. The first article demonstrated the efficacy of a naive equal-weighted asset allocation. One reason this approach is effective is that it does not depend on accurate asset return forecasts. Nailing accurate forecasts is obviously difficult. In fact, it is so difficult that even experts tend to perform worse than chance.
A Physicist’s Dilemma
I used to work as a physicist where the critical lesson you learn is that you can only publish your work if you show the measurement errors of your experiments or the prediction errors of your model. When I switched to finance, I got the distinct impression that I had a better chance of publishing my work if I omitted them! Of course, this may be stretching things somewhat, but while our industry is obsessed with forecasting, it is astonishing how little attention is paid to the accuracy of forecasts and their estimation errors.
Take, for instance, the annual ritual of publishing year-end forecasts for stock market indices, interest rates, or exchange rates. As we enter 2016, I am certain there will be a plethora of surveys among economists and financial analysts by news organizations around the world, each asking the “experts” where the S&P 500 Index or the 10-year Treasury yield will be at the end of the year. And, as usual, many of us will happily provide these forecasts, secretly hoping that nobody checks back a year later to see how they turned out.
Unfortunately, some people actually do look at the accuracy of these forecasts and have found the results wanting.
Some of the most comprehensive studies have been done by Markus Spiwoks and his colleagues. They looked at the three- and 12-month forecasts by professional analysts of stock markets, short- and long-term interest rates, and exchange rates in several countries, and compared these forecasts to a naive forecast constructed on the assumption that financial markets follow a random walk. If they do follow the random walk, then the best forecast one can give is that interest rates, exchange rates, and the stock market will be exactly where they are today one year from now.
The Wisdom and Madness of Crowds
The good news is that this research confirms a “wisdom of the crowds” effect insofar as only a few analysts seem able to consistently outperform the consensus forecast compiled from many different analysts. In a study of US interest rate forecasts, the German researchers found that about 90% of forecasts were less accurate than the consensus forecast. Recent research from Yury Kucheev, Felipe Lopez, and Tomas Sörensson has also shown that there is some persistence in the recommendations of the few analysts that manage to beat the crowd. Analysts who are rated as star analysts in surveys like the ones published by Institutional Investor, the Wall Street Journal, or StarMine tend to outperform non-star analysts in the year after their election — though the outperformance tends to be quite a bit lower than in the year before they are elected star analysts.
But that is where the good news ends. In fact, Spiwoks and his colleages found that none of the predictions for stock markets, interest rates, and exchange rates for four countries were able to beat the naive forecast. For interest rates for example, about 98% of the predictions were more closely correlated to current market circumstances than to market circumstances at the forecasting horizon. In other words, the forecasts reflected the present more than the future. Or as they say in Denmark: “Predictions are difficult, especially when they involve the future.”
If you have to rely on short-term forecasts for whatever reason, the best recommendation is to assume that interest rates, exchange rates, or stock market levels will be exactly where they are today. Of course, for many investment decisions, short-term forecasts over three or 12 months may not be relevant. Instead, one might rather rely on long-term return forecasts. In that case, reversion to the mean should lead to more predictable returns in the long run. But one might be surprised to find that this belief is misplaced, as the next article in this series will show.
For more from Joachim Klement, CFA, don’t miss Risk Profiling and Tolerance: Insights for the Private Wealth Manager, from the CFA Institute Research Foundation, and sign up for his regular commentary at Klement on Investing.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©iStockphoto.com/CSA-Archive