A LOCAL CHARITY

has convinced Warren Buffett to fly into town to offer a seminar on value investing. Tickets are $100. You're a regular reader, and I'm a nice guy. So I give you one which you promptly lose.

That evening you're neither richer nor poorer. But you're probably unhappy. It's not rational, but people dislike losses more than they value gains. So the discomfort you felt when you couldn't find the ticket is greater than your satisfaction when you received it. This is a simplified example of what psychologists call loss aversion, which turns out to have major implications for investing.

Economists are just beginning to understand how behavior affects markets, and the relationships are anything but obvious. This month, for example, I will tell you about new academic research that uses this irrational tendency to avoid losses as the basis of a clever indicator that predicts short-term performance of individual stocks. Along the way I'll have some nice things to say about technical analysis (a subject that gets more than its share of bad press), and I'll combine some traditional behavioral tools to prepare a table of companies that look like solid long-term bargains.

What intrigues me and has triggered much research is a form of loss aversion called the disposition effect. That's the technical term for the tendency to sell winners instead of losers, as investors opt for the satisfaction of a gain rather than the pain of a loss. There's plenty of evidence that this really happens, even in situations where logic would dictate doing just the opposite. Sell a winner and you pay capital gains; sell a loser and you get a tax deduction.

Still, psychology prevails. In a six-year study of 10,000 accounts at a discount brokerage firm, Terry Odean of the University of California at Berkeley looked at how investors behave when they need cash. As predicted, they were more likely to unload stocks that had gains than those that had losses by something like 50 percent. Other studies turned up similar behavior among individuals in Finland, Israel and even among Chicago futures traders.

I'll get back to the disposition effect, but let's shift gears and consider a phenomenon called momentum. In the early '90s researchers documented what Wall Street tape-watchers had long suspected: The trend is your friend. Over moderate time periods stocks that have gone up keep going up, while falling stocks keep going down. Suppose that every six months you buy the previous six months' winners and sell the losers short. According to a classic 1993 study, you'll earn a neat 10 percent annually if you aren't swamped by trading costs.

Momentum puzzles economists. They can't explain why past returns predict future performance, even in this limited way. And looking back over longer time periods only adds to the confusion. Then losers beat winners. Some other landmark research shows that stocks that have declined the most over several years can be great buys. Academics call this reversal. Three-year losers, for example, outperform three-year winners by something like 10 percent annually.

Why would the market reward trend followers short term and contrarians long term? New research on the disposition effect provides at least a partial answer and it comes from an impressive source. Bing Han grew up in China, graduated from a math institute there, then earned a Ph.D. in math from the University of Chicago. Now he's getting a second Ph.D. in finance at UCLA and being recruited by 14 schools from Berkeley to MIT.

One reason Han is so popular is his dissertation, co-authored with his UCLA adviser, Mark Grinblatt. Finance professors believe that the market creates a "correct" price for every stock. But Grinblatt and Han (G&H) believe psychological factors create market imbalances that can be exploited by savvy investors. They argue that when recent purchasers sell winners and hold losers, the correct price equilibrium becomes distorted. Winners become too cheap, and losers become too expensive. Other investors scramble to set things right, bidding winners up and pushing losers down. Over time the result is what's called momentum.

Rather than following every stock that appears to have momentum, however, G&H believe that an alternative way to earn momentum profits is to buy stocks whose investors have the largest built-in gains and sell stocks with the largest built-in losses. To find those stocks G&H create what they call a reference price for each company. The math is complex, but the idea is simple: A number that approximates what the average shareholder paid for his stake.

As predicted, stocks whose market price is farthest above G&H's reference price (winners) show the best subsequent six-month gains, while those whose market price is farthest below the reference price (losers) show the worst declines. But here's the clincher: The G&H approach does a much better job of predicting momentum than simply focusing on prior gainers and losers.

This has important implications. G&H's research indicates that momentum makes sense in terms of psychology and economics. And in the real world, reference prices have significant potential for stock pickers. Though G&H didn't test this directly, their results indicate that trading costs are far lower using reference prices to capture momentum profits than using conventional buy-what-went-up and sell-what-went-down strategies. If this catches on, as I think it might, reference prices could become common tools for investors, like P/E ratios.

Which brings me to technical analysis. One way to think of G&H's reference price is as a statistically souped-up version of a 200-day moving average, which is among the most popular technical tools around. Technicians are happy when a stock's price is above that number, and they worry when it drops below. Economists used to laugh, but attitudes are changing. A handful of pioneering studies reveal that several technical tools (including the 200-day average) produce valuable information.

How do I apply all this? It would be neat if I could create a table based on reference prices. But the calculations are daunting. Instead, I'm using a new version of Zacks Research Wizard to try for a behavioral twofer. I want to capture momentum by finding stocks that have moved up smartly over the past six months. But I'm also looking for potential reversals, stocks that are poised for gains because they are long-term losers. I focused on companies with market values between $500 million and $2.5 billion a universe of 1,375 stocks. That's where momentum effects are most pronounced (perhaps because small stocks tend to be owned by individual investors, who are more likely to be influenced by the disposition effect). To find long-term laggards and short-term winners, I looked for stocks in the bottom 20 percent of the pack based on three-year price performance and in the top 20 percent over six months.

Only 27 companies meet both requirements. I created the accompanying table by adding two value criteria: positive cash flow and less debt than equity a cushion in a dicey economy. Since I chose these stocks based on price trends rather than fundamentals, I'll behave like a chart maven and avoid discussing each company in detail though I am pleased to have a diversified roster.

My finalists are into everything from overnight delivery (Airborne) to embedded operating systems (Wind River). But they share a common performance history. On average, shares have fallen nearly 40 percent since the end of 1998. But they're up 16 percent since last July. If I'm reading the research correctly, that could be a powerful combination enough, I'm sure, to offset any unhappiness caused by losing a hypothetical lecture ticket.

Stocks reward trend followers short term and contrarians long term. Here are eight companies poised to do both.