The investment markets are all about what have you done for me lately. Make money for years but fall behind for a couple of quarters, and there will be someone calling for your head and others saying you are a friggin’ muppet who just lucked out in the past. (Fortunately, it works the other way too—you can lead a giant hedge fund over the cliff and still be regarded as brilliant so long as you are making money now.)
Bill Miller beat the S&P for 15 years running but this year is down 10% and way off his pace. So of course people are saying his past performance was just good fortune. More specifically, they are saying that if you had a bunch of monkeys managing money you’d expect at least one would do about as well as Miller.
Wrong, says Paul Kedrosky.
The math is easy: Assuming data independence (i.e., bad years don't influence managers the following year), then fifteen years of market-beating performance at a 0.5 likelihood per year gives us a probability for Miller's performance of 0.003%. Darn unlikely, in other words.
But that's not enough, of course. We need to know how many portfolio managers were running active money back in 1990 when Miller began his beat-down of the S&P. According to data I found elsewhere, there were almost 700 such funds back then (and there are probably twenty times as many now). Given that number of funds, and given the above-mentioned probability, we would expect around 0.02 fund managers to have turned in a Miller-like performance by now given the cohort size from fifteen years ago.
Trouble is, 0.02 of a portfolio manager isn't a very effective portfolio manager (even if it's cheaper), so we can reasonably say that Mr. Miller's performance is highly unlikely, especially if he is really a coin-flipping monkey. Of course, it's not inconceivable that it happened by chance -- and back at the nine-year mark it was perfectly likely -- but a 15-year streak would still have been unusual.