Gambler's Fallacy and the Stock Market: Mean-Reversion or Dilution

On an annual basis, the Dow Jones Industrial Average rose 66% of the time from 1896 to the present regardless of whether the market was up 20% the prior year, rose the prior year, or fell the prior year. (Marketwatch: Here are the odds that U.S. stocks will rise in 2016). From year to year, the stock market appears to follow a random walk. Over time, this random process dilutes the deviations from the long-run average. The Gambler's Fallacy basically says that if something happens more frequently than normal during a given period of time, it will happen less frequently in subsequent periods. Likewise, if something happens less frequently than normal during a given period of time, it will happen more frequently in subsequent periods. If stock market returns are completely random, then it would be false to believe that if the market was up one, two, three, etc. years in a row, the market would be due for a down year.

In contrast, the stock market's value may randomly wander around its intrinsic value. By reverting to the mean over time, the process is not entirely random. One could expect to take advantage of periods where valuations over or under shoot the mean. Models that attempt to take advantage of mean reversion such as Research Affiliates' Asset Class Expected Returns, GMO's 7-Year Asset Class Forecasts, and John Hussman's Market Cap/GVA typically use a time frame of 7-12 years.

Avoid the gambler's fallacy in the short-run, but keep in mind periods of overvaluation may lead to periods of under performance and vice versa. While there is no way to predict turning points based on valuation, it can used to manage risk, provide courage to buy when others are selling, and sell when others are waiting for one last rally.