PEERING INTO A FUND'S PAST CAN'T TELL YOU ALL YOU NEED TO KNOW ABOUT ITS FUTURE
By JASON ZWEIG

(MONEY Magazine) – There I was on vacation, tilting back in my rocking chair with some light summer reading--the July issue of the Journal of Finance--when it hit me. No, it didn't occur to me that I should be zipping through a Tom Clancy thriller instead of squinting at this stupefyingly complex review of investment theory. What struck me was that, yet again, someone had tried to show that you can use past mutual fund returns as a foolproof guide to the future--and had failed.

"Persistence of performance," the idea that yesterday's winning funds will be tomorrow's too, is an almost irresistibly appealing notion. After all, just as some baseball players hit more regularly than others, some fund managers must be better at picking stocks than others. Plus, you read all the time, in MONEY and elsewhere, about funds that have beaten the market for long periods. If you buy funds that have shown great returns in the past, shouldn't you be assured of great future returns?

Martin Gruber, the distinguished chairman of the finance department at New York University's business school, sets out that case again in the Journal of Finance. "The surprising thing about persistence is not that it exists," he writes, "but rather how strong it appears to be." Gruber looked at 227 diversified U.S. stock funds starting in January 1985. He sliced them into 10 tiers, from best- to worst-performing, and compared each tier's returns with that of an index he customized to capture the movement and risk of the entire market. Gruber followed these 10 tiers and found that if each year you had bought the 10% of funds that finished the previous year with the best returns, you would have beaten the market index over the course of the decade he analyzed.

In other words, fund history repeats itself. A logical conclusion: You should buy last year's hottest funds, hold them for one year, then sell them; repeat until wealthy.

Not so fast. There's less here than meets the eye. First of all, Gruber studied only one period: 1985 to 1994. His technique worked for most of the years in that one 10-year span. But, he admits, it might well fail in any other decade--like, say, 1996 to 2005. Two similar studies, by Burton Malkiel of Princeton and Nobel prizewinner William Sharpe of Stanford, suggest that fund performance is becoming less persistent than it may have been in a few past periods. "Do winners repeat?" asks Sharpe. "The evidence is far from conclusive." In fact, years of fund watching have convinced me that, just like baseball players, even the best fund managers are sure to seem like chumps eventually--often right after they seem like champs.

In any case, let's see how much you could really gain from Gruber's guidelines. If you had bought his "winners" over the decade ended in December 1994, you'd have earned 0.2892 percentage points a year more than Gruber's market index. But to get that extra eyelash of return, you'd have to keep picking each year's top 10% of funds solely from your original list of 227, even if better choices became available later. You'd have to own 22 funds at once; although minimum account balances vary, you'd need at least $22,000, and probably far more, to build this portfolio. Over the decade, you'd have to trade your funds as many as 440 times. After 10 years of frenzy, you would have turned $22,000 into $86,475--just $2,156 more than you'd have made buying Gruber's index and approximating the return of Standard & Poor's 500.

And after all that effort, you wouldn't even get to keep the $2,156. Sales loads would cost you well over $3,000--and if you avoided them by sticking only to the no-loads Gruber tested, your return over the index would drop to $1,633. But your accountant could easily charge that much over 10 years just to unsnarl all your extra paperwork. And by flipping funds instead of holding on for the long haul, you'd pay high taxes each year instead of deferring most of the tax bill into the future. Even if your brokerage and accounting fees were zero, in a few years the extra taxes would devour your entire "gain." In short, fuhgeddaboutit.

Gruber himself doesn't think people should invest like this. "It would be very costly and time consuming to invest in funds this way," he told me. (The point of his study was not to prove that flipping funds is a good idea but to see why all investors don't just buy index funds.) Even so, I'll bet that newsletter writers, market timers and other people who want a piece of your wallet will soon be citing Gruber's study as "proof" that trading hot funds is the best way to wealth.

My advice: Ignore them. Here are four time-tested ways to get great results. 1) Buy index funds, whose managers can't go into a slump, since all they do is mechanically mimic the market's results. 2) If you don't want to index, stick to funds with slightly above-average past returns. If persistency of performance does exist--and no one, not even Gruber, has proved that--then it's more likely to be found among slight outperformers than huge ones (see my column in September's MONEY). 3) Never buy a fund with high annual expenses; many good U.S. stock funds charge 1% or less. 4) Hold on for five years or more; that way you, not the tax man, will make more money.