(MONEY Magazine) – Fish gotta swim, birds gotta fly--and Wall Street can't let a day go by without spewing out a fresh supply of nonsense. But smart investors know that there's usually a tiny grain of wisdom hiding in even the worst absurdity.

Take stock splits. In a typical "two-for-one" split, a company gives you a second share for each one you own but cuts the price in half. In theory, that leaves you exactly where you started--but, in practice, the announcement of an upcoming split has often been enough to make a stock jump, particularly in 1999 and 2000. Hyped by a potent mix of their own press releases, the resulting frenzy in online chat rooms, and day-traders whose pagers beeped on news of the latest splits, companies like CMGI and JDS Uniphase split their shares three times in just 12 months; Sycamore Networks even split only a few weeks after it went public. But then the dotcom bubble burst last spring, and most of the overheated stocks that had split went splat.

There's more to this story, however, than the death of the latest get-rich-quick delusion. There's a real puzzle here: Why do investors like stock splits, anyway? Imagine that I asked you to give me a dime, then handed you back two nickels and said, "Don't you feel richer now?" You'd probably think I was insulting your intelligence. Yet that's exactly what happens when a company splits its stock. Let's say you start with 100 shares of Whangdoodle Corp. trading at $100 apiece, and the stock splits two for one; now you have 200 shares, each worth $50, leaving you no better off than before. And the split has zero effect on Whangdoodle's earnings or assets. "A split is an empty transaction," says David Ikenberry, a finance professor at Rice University.

Well, not quite empty. In a fascinating, not-yet-published research paper, Ikenberry has studied more than 12,000 stock splits back to 1930. In every period, stocks that split outperformed those that didn't. Over the most recent decade that Ikenberry analyzed--1988 through 1997--companies that split returned an annual average of 23.3% vs. 18% for Standard & Poor's 500-stock index. Considering the essential emptiness of stock splits, Ikenberry's long-term findings are nothing short of amazing. What's going on here?

Let's do the split

Vanderbilt University historian Peter Rousseau says stock splits date back at least to the Boston Stock Exchange in the 1850s. In those days most people could afford to buy only one share at a time, since stocks traded at a percentage of their so-called par value--which typically ranged from $100 to $1,000 (after 150 years of inflation, that would be as much as $20,000 today). For instance, Salisbury Mills split 10 for one in 1857, carving its $1,000 stock into 10 slices at $100 each. These early splits, argues Rousseau, made it easier for America's pioneering manufacturers to raise capital from small investors.

But nowadays, almost every investor can afford more than a single share at a time, and there's no rational reason why a company would rather have, say, 20 million shares priced at $25 instead of 10 million shares at $50. Either way, its stock has a total market value of $500 million. So why do companies split their shares?

One reason, says Ikenberry, is that splitting has come to be regarded as a sign of good corporate health: "A split is a way for managers to say, 'We have confidence that bad news is not around the corner.'" Indeed, he notes, earnings at the typical splitting company have risen an impressive 20% and the stock an average of 50% in the preceding year.

Finance professors James Angel of Georgetown University and Michael Brennan of UCLA offer another insight. Stockbrokers make much of their money by pocketing the "spread" between what buyers are willing to pay for a stock and what sellers are willing to accept. A typical spread runs roughly from 12.5[cents] to 25[cents] a share. If you trade 100 shares of a $50 stock, your broker might pocket a quick, risk-free $25 in spread. But if that stock split two for one, and the spread stayed constant (as it often does), then your broker would double his take, since he'd get to skim the same spread off twice as many shares.

What's more, some full-service brokerages collect a higher commission on stocks with lower prices. A broker at one major firm tells MONEY he would charge $265 to trade 1,000 shares of a $10 stock but just $105 on 100 shares of a $100 stock--even though each trade totals $10,000. So Wall Street has its own giant incentives to goad companies into splitting.

Why one plus one equals three

Of course, the main reason firms split their shares is that investors want them to. It's bewildering that we can't outthink Yogi Berra--who, when asked whether he wanted his pizza cut into four or eight slices, replied, "Four. I don't think I can eat eight." Why is it so hard for us to realize that it's the size of the entire investment pie, not how many slices it's cut into, that really matters?

It's partly because our own minds play tricks on us. Many investors view a two-for-one split as if it were a windfall--congratulating themselves on having twice as many shares as before, while failing to see that their total stake is worth the same as ever. Casino gamblers use the same kind of "mental accounting" by viewing their winnings as the "house's money," not their own.

There's another mind game going on here too. Let's say you originally bought 100 shares at $10 apiece and the stock then runs up to $50 and splits two for one. Since you've already ridden the stock up to $50 once, it's easy to imagine you'll get there again. Now you're engaged in what psychologists call anchoring: fixating on a previous high-water mark as a way to measure your likelihood of future success. In reality, of course, there's no guarantee that your stock will ever hit $50 again.

Another oddity of splits is that they can create the illusion that a stock has become cheap all over again. A $50 stock may lack bargain-basement appeal. But if it splits two for one, it can suddenly strike you as undervalued, since $25 is a much lower price. After stocks rise, it takes much larger price declines to make them look cheap again. Scientists even have a name for this; it's called Weber's Law, after a German physiologist who noted that people have difficulty perceiving small changes to large quantities.

Betting against Buffett

Can you make money off something as dopey as stock splits? Yes and no. As Rice professor Ikenberry's numbers show, stocks that split have beaten the market by roughly five percentage points a year for decades. Unfortunately, explains Ikenberry, you could have captured those returns only by buying shares in every single one of the thousands of stocks that have ever split and holding them for one year. But nobody invests that way, so Ikenberry also tested what you'd have earned with a more active split-trading strategy. That cut the outperformance nearly in half, meaning that you'd be left with only modest profits after paying taxes and commissions.

Ikenberry also hastens to add that his research stops in 1998, before serial splitters like Amazon, CMGI and DoubleClick came crashing down; updated through 2000, Ikenberry's numbers would lose still more of their gloss.

In the end, Ikenberry advises against buying stocks solely on the basis of splits. However, he does have two suggestions. First, if you're already thinking of buying a stock, and it splits, you can take that as a fairly strong signal that the stock may be a good buy. And, he adds, if you're considering two similar stocks, and one of them splits, that's probably the one to go with.

But while it makes sense to use splits as one of the factors you look at before buying a stock, it's worth remembering that Warren Buffett has never split Berkshire Hathaway's shares (which now cost about $67,000 each). Buffett, the greatest investor alive, has explained that he doesn't want to attract "inferior" buyers who'd "downgrade the quality" of his current shareholders. That's his polite way of saying that people who buy stocks on the basis of splits alone are bozos. Jason Zweig can be reached at fundamentalist@moneymail.com.