Consider the null hypothesis

Burlington, Vermont is a town populated by bumper stickers. It's just a thing here. Last week I was driving to work and I saw one that said "Consider the null hypothesis." I thought to myself "It's 7:45 AM and I have not yet considered the null hypothesis today. In fact, I have no idea what that even means."

So I looked it up, and lo and behold, I learned I should be considering it more often. The null hypothesis is something from statistics that says random chance, or luck, alone results in an outcome.

Nobel Laureate Daniel Kahneman and Amos Tversky, Israeli psychologists, discovered decades ago that the Israeli army had no ability to predict who should be promoted. Despite many tests and feedback systems, their decisions were as good as random chance.

In individual stock picking, most success is random chance. By success, I don't mean gains. I mean gains above a comparison to the appropriate index, after taxes are paid (1).

About fifteen years ago I set up a series of test cells to see if I could outperform the indexing model. Each cell had a defined set of parameters for stock selection, and the parameters were largely based on principles I had learned from operating businesses. My theory was that people who worked on Wall St often had little operating experience, and without it they were at a disadvantage.

One of these cells (just one), 15 years later, did beat its small-cap index comparison. With one exception, every security roughly quadrupled. They were all buy and hold scenarios - no active trading. I told my wife about it, thinking she would be impressed. She said "You got lucky. Stick with the indexes."

Two observations: first, once the tax effect of the gain was subtracted from the returns, the out-performance was marginal. Second, the outperformance was marginal in part because I owned one stock that had flatlined. Even one weighting like this hampers returns, and it's almost impossible to avoid.

So that's why this idea died, and never found its way to One Day In July for others. I could never convince myself that even in the best case scenario it was more than just luck. We strongly believe investing is not and should not be entertainment, so without performance based on something other than luck, the idea had to die.

In other words, the null hypothesis was alive and well.

Dan Cunningham

(1) If you want a research paper backing this idea, I suggest drinking a couple of Red Bull's and settling down with this tome from Eugene Fama and Ken French. Or just peruse the summary.

Return to Articles