Fooled by Randomness

Fooled by Randomness by Nassim Nicholas Taleb is a great read, especially for anyone who is interested in the profession of investing. My partner, Greg Gretsch, kindly sent me a copy. The book presents strong evidence that all these superstar investment managers are mostly just lucky. Since Greg's investing record looks pretty good, and since he is a competitive type (aren't we all?), this is an interesting gift. However, Greg (along with all good venture capitalists) is rescued at the last moment by a Harvard professor ... read on.

In Fooled by Randomness, Taleb reminds us that humans crave meaning in our lives, for all sorts of reasons, and therefore most patterns we see in life are ex post facto explanations which have no statistical meaning. By this I mean that we impose some kind of meaning or explanation for usually random events that have happened to us, and which, when analyzed, really (and in all probability) did happen just randomly.

Let me retell a fable that Taleb offers about a charlatan, let's call him Charlie, who makes money from those of us fooled by randomness. Charlie finds (or buys) a mailing list for active amateur investors. It contains 10,000 names. He sends 5,000 of those people a letter introducing himself and predicting that IBM stock will rise over the next week. The other 5,000 receive a similar letter predicting that IBM will fall. A week later Charlie throws away the names of the people who received the "wrong" prediction. He then splits the remaining group into two halves and sends two similar letters. Charlie repeats for a total of four times, each time sending subsequent letters only to the people who received the "correct" forecast. After four weeks around 600 people have seen that he is consistently right with his short term forecasts. How does Charlie make money? He now sends the remaining eager 600 investors a letter offering them the next forecast for the very reasonable sum of $10,000.

This sounds very cute, but it turns out to be how most mutual fund (or investment) managers are judged. Imagine that all investors' results are basically random, like Charlie's, rather than the product of any skill. Purely by luck, some number of these investors will have consistently positive results over the first few years of their career. This is the same as tossing a coin one million times and finding a few consecutive sequences of 10 or even 100 heads. If an investor has been successful over several years truly on the basis of skill, then you would expect the following year to also be good. Taleb finds (as do many researchers) that in most cases the results of the following year for previously superstar investors follow the same pattern of randomness as for novices (the same proportion of the successful subset succeed or fail as you find with the novices). Only a very small number of investors are truly superstars and succeed year after year beyond expected randomness.

We treat all "top-performing" investment managers like Charlie, not realizing that we are doing the charlatan's work for him. Instead of Charlie dividing the world of newsletter recipients into winners and losers from week to week, we are dividing the world of investment managers the same way, and then assuming those who are left are great. We are fooled by randomness.

Despite these findings, there is good news for Greg, for Sigma Partners, and for the rest of the top-tier VC world. It turns out that previous good results from a venture capital firm do predict future results. Harvard Business School Professor Paul Gompers et al present supporting evidence of this in their paper Skill vs. Luck in Entrepreneurship and Venture Capital: Evidence from Serial Entrepreneurs. This is not just an academic finding, as noted in one of my recent posts quoting from an investor in VC funds who seems to agree based on historical results.

So the takeaway is that Greg has provided a wonderful read for me, without disturbing our own haven from the winds of randomness.