MOST OF US WERE

perusing turkey recipes and pondering pie crusts on the day before Thanksgiving, but Allen Lenzmeier found time to make a $400,000 investment. He bought 50,000 shares of stock in UTStarcom, a California-based cell phone manufacturer that does most of its business in China and is trying to crack the U.S. market. Lenzmeier ought to know what he's doing: He is vice chairman of Best Buy, which sells lots of cell phones, and in March, he became a director of Starcom.

As a corporate insider, Lenzmeier must disclose all his Starcom stock transactions within two working days. On the following Monday, his broker sent the details including Lenzmeier's average purchase price of $7.94 to the Securities and Exchange Commission. The filing was an electronic version of what the SEC calls "Form 4: Statement of Changes in Beneficial Ownership." About 1,500 similar reports flow silently into the SEC's Edgar database every day.

That's when the excitement begins. Insider trading is getting more attention than ever on the web, on television and in print. But I'm not sure all this fast, accurate information is making anyone a better investor. In what follows I'll provide a cautionary tale, update you on the changes and introduce a stock screen that uses insider information in a stodgy, conservative and I hope profitable way.

For years, tracking insider buying (the legal kind that involves corporate officers, directors and large shareholders) was time well spent. Academic studies conducted decades ago revealed that piggybacking on insiders can be a consistent way to make money. Subsequent work highlighted the most promising strategies: Buy the "buys," ignore the "sells." Look for large transactions. Follow the C's: chairmen, chief executives and chief financial officers.

Then, in 2002, Sarbanes-Oxley reforms tightened the rules. Previously, insiders could take up to a month before mailing in their ownership forms, and processing might require another two weeks. Now the two-day reporting requirement makes electronic filing essential. That sparked a profusion of web sites with names like Form4Oracle, InsiderScoop and ScumFish. Smart programmers can write software that combs the Edgar database to create a feed of Form 4 filings.

The result: Go to SecForm4.com and you'll see trades reported instantly for free. Here's how this profusion of data plays out in the real world. Allen Lenzmeier's Starcom purchase was time-stamped by the SEC at 1:19 p.m. Four minutes later, there was a post on the stock's Yahoo bulletin board with a link to a copy of his Form 4. Starcom shares, which had languished below $8.00 since August, zipped from $7.90 to $8.50 in 30 minutes. They closed at $8.40, up 6 percent on volume of 5 million shares more than twice normal activity.

What's happening here is that thousands of small "swing" traders are playing the insider game. The idea is to buy within seconds of seeing a transaction and hold for five or 10 days. You sell sooner if the stock moves by, say, 10 percent up or 5 percent down.

Are these traders making money? Frankly, I don't care. But data vendors estimate that there are perhaps 10,000 of them, and sharp moves like the reaction to Lenzmeier's purchase are now common. This roils the water for the rest of us, and my advice is to stay away. Don't buy or sell based on any insider activity until at least two weeks after the announcement date. That way you're not feeding the sharks.

There's extra confusion about selling, which accounts for nearly 90 percent of all Form 4 transactions. Research indicates that these trades don't predict stock declines. Like the rest of us, insiders sell when they need money. Some economists point out that stocks that insiders sell often perform better than stocks with no insider activity at all. Executives may avoid unloading shares when they see bad news ahead because they're afraid of getting sued.

Selling is a big part of the new fascination with insider activity. And the added interest comes just as a regulatory change makes the numbers less relevant than ever. A five-year-old SEC rule called 10b5-1 gives insiders a defense against lawsuits if they use automatic trading plans. These plans may account for 25 percent of all insider selling now, and that number could double or triple in a few years.

It's a great setup for insiders. Besides the legal protection, the plans allow trades during "blackout" periods (typically around earnings releases) when activity would otherwise be banned. Unfortunately, it is often impossible to tell when a trade is automatic and meaningless as an indicator. Some companies issue press releases when executives set up 10b5-1 plans, and some Form 4s indicate that a trade is made through a plan, but disclosure isn't required.

For all these reasons, I'm cautious these days when I draw conclusions from insider activity. But the numbers are still intriguing. Consider a new study by Jim Hsieh at George Mason University and Lilian Ng and Qinghai Wang at the University of Wisconsin. They examined the relationship between insiders and analysts, looking at results for thousands of firms between 1994 and 2003. The conclusion, in a nutshell: Follow analysts only when they're negative (downgrading), and follow insiders only when they're positive (buying).

I'm also impressed that gains from stocks with significant insider buying have "legs." In Hsieh's study, stocks with net buying, measured not by one or two trades but over periods of one to three months, were beating the market by the equivalent of six percentage points annually more than one year after the purchase. One reason is reversion to the mean. Insiders buy when prices have gone down. In a similar contrarian vein, they also preferred stocks that analysts shunned.

This month's screen is my attempt to capture those long-term gains, which I hope won't be diluted by zany swing traders. The economists' purchase criteria were simple. They tracked net insider purchases two ways based on the dollar value and based on the number of transactions. Results were similar either way. It also made little difference whether the look-back window was 30, 60 or 90 days. I like the longest period because it minimizes the impact of frantic announcement-day activity.

With that in mind, I zeroed in on large companies (market value above $2.5 billion) with net buying (more insider buys than sells) over the most recent three months. That's a small universe only about one company in 15 qualifies. My finalists also meet these common value criteria: five-year average return on equity above 15 percent annually, debt less than 50 percent of equity and a price/earnings-to-growth ratio of less than 2. They pay dividends too, with a yield of at least 1 percent.

This is a diversified roster. Abercrombie, Kimberly-Clark and Mattel are household names, while other companies have lower profiles. City National, which caters to affluent Californians, and North Fork, which has lots of Manhattan branches, are both banks with enviably modest operating costs.

Graco sells paint sprayers and almost everything else connected with liquids and adhesives. Yes, book value is sky-high. But so are its 50 percent-plus profit margin and lavish cash flow. Iowa-based HNI has quietly become the second-largest (and most profitable) office furniture company in the country. Finally, slot-machine maven International Game Technology could hit jackpots from both an uptick in domestic replacements and burgeoning foreign demand.

I'm pleased that most of these stocks are comfortably below 52-week highs, consistent with insiders' contrarian bent. These may not be top picks based on trades made days or hours ago. But I prefer to look for bargains after the dust has settled.