Concentrated Bets are Risky

Bloomberg has an article about how some star investment managers are having really bad returns over the past year. John Wood's SRM Global Fund lost about 70 percent through March 31. That's a drawdown. Pirate Capital had $2B under management, and 50% of its holdings in only 4 firms. Now, clearly some of these guys were simply lucky in the past, and taking large, undiversified bets that scored big made them appear like geniuses. Others, got more money than they could invest, and had to change their investment focus from looking at certain subtleties of the business model to industry bets, which are such different approaches they require a separate skill--one that even fewer people have.

Indeed, I worked for a convertible bond desk, and was struck by the fact that for a large convert book, say over $1B long, you have so many positions that stock selection alpha is necessarily very small. The only way to outperform in a big way is to take industry bets, and these are, like macro forecasting, invariably alpha-less. You can still have edge managing a convert desk, but for any large book, looking at individual tilts, you can't add more than a percent to you returns this way. A better, scalable focus, is simply to be a patient trader, willing to act as a liquidity provider, so that you are usually buying at the bid and selling at the offer. When your I-bank calls, and wants to dump $50MM in an issue, you are there, but at a price discount. The edge there is more certain, and larger, than picking the better companies. Thus, an analyst at a desk may be great at picking a handful of converts, but a lousy porrtfolio manager because he does not understand this, while a portfolio manager clueless about relative value, but a disciplined trader and tough negotiator, may be a great portfolio manager (though invariably he markets himself as a convert selector, making for a marketing cluster-puck).

Many of these ex-superstars, I think, are merely having bad luck. Lampert's ESL Investments Inc. dropped 27 percent last year and an additional 1.3 percent in the first three months of 2008. Sounds bad, but Lampert has reportedly produced an average annual return of almost 30 percent since 1988. If that's true, he's got alpha. But it appears his skill is at looking at specific turnaround deals, and so, his edge is necessarily part of a volatile strategy. There is simply no way to take a dozen or so insights, and lower the vol to the 10-12% that most hedge fund investors market. I don't see a problem with that. Not every hedge fund will have an edge in trading systems, like Renaissance or DE Shaw, and able to produce a minor edge via their massive technology investment, which is necessarily a well-diversified strategy. Most investors with alpha, in fact, are like Eddie Lampert. They can identify stocks based on some qualitative insight, and such insights can only be made over a handful of companies because to read the annual reports, listen the executives describe their business models, know the set of competitors, or even get involved with management, and this involves a lot of time, which is a limited resource no matter how rich you are. This does not scale, so one should accept the fact that big losses are going to happen.

Nevertheless, a loss is still a bad signal, and given a Bayesian prior where one puts some weight on the probability that a manager no longer has, if might have never had, alpha, a bad return should imply some rational investment outflows.