WHEN I FIRST

started writing about the stock market, the juiciest business story around was eerily similar to today's Enron scandal. A fast-growing company called Equity Funding collapsed overnight. It had complex products a mix of insurance and mutual funds and managers were cooking the books. Investors lost hundreds of millions of dollars, 22 people were indicted, and some went to jail. A Wall Street analyst was at the center of the whirlwind.

I'm fascinated by the shifting roles analysts played in these dramas. Today analysts are scorned as villains. Perhaps for good reason. While Enron stock slumped, they kept recommending that investors buy it, and after it collapsed, they were hauled before Congress to explain why fat investment fees hadn't colored their judgment. But back in 1973, an analyst was a hero in the Equity Funding case. His name was Ray Dirks, and he discovered the fraud through a tip from a former employee. The mistake he made was alerting his clients first instead of the SEC, which got him in hot water with regulators. He was eventually cleared by the U.S. Supreme Court.

Why have analysts fallen so far from grace? Think about the way they get paid. In the '70s, brokerage firms made their money from fixed commissions on securities sales. Research was a cost of doing business, paid for by big traders. Now everyone can buy and sell stocks for pennies. Wall Street profits come from dealmaking, and an analyst can do more for the firm by talking up deals than by kicking a company's tires. I'd be as "shocked" as Claude Rains in Casablanca if analysts didn't give companies that generate investment fees special treatment. Instead of worrying about conflicts of interest, I'd rather squeeze the most value out of the Wall Street research I do see. Analysts often know as much as anyone about what's going on inside a company. And thanks to technology, one important window on their thinking is just a mouse-click away for individual investors.

I'm talking about up-to-date consensus recommendations, compiled by vendors such as Zacks Investment Research and Multex. These numbers (once available only to pros) now appear on SmartMoney.com and other Web sites. The consensus is an average of the opinions of all analysts who cover a company. Typically, there's a five-point scale 1 means a strong buy; 3, a hold; and 5, a strong sell. Changes show up as a percentage increase/decrease, measured over the past four weeks or three months.

How can you translate these numbers into practical investment decisions? New research by Narasimhan Jegadeesh at the University of Illinois and Charles Lee at Cornell offers some surprising conclusions about when to follow analysts and when to bet against them. Academics have long been intrigued by Wall Street research, and in a variety of studies, it gets high marks. Over time the stocks analysts pick beat those they pan by at least 6% annually. The performance gap is much wider (maybe 20%) if you focus on just the extremes strong buys versus strong sells. That's reassuring, if not useful. For one thing, sell recommendations are rarer than tech stocks with dividends. And there are occasional oddball periods when sells go up and buys go down.

Enter our two professors, whom I'll call JL. In a massive number-crunching effort (helped by co-authors Joonghyuk Kim and Susan Krische), they tracked nearly 55,000 analyst recommendations made from 1985 through 1998. They classified each of those stocks according to momentum benchmarks (price and earnings gains), growth benchmarks (historical and projected growth rates) and value benchmarks (price/earnings and price/book ratios). The idea: to find out what kinds of companies analysts like and when their picks add value.

Perhaps the single most useful discovery is this: When considering analyst opinion, you're better off looking at the change in the consensus rather than at the absolute number. A major weakness of analysts is a reluctance to downgrade favorite stocks no surprise to anyone who owned Enron. When recommendations do move, however, revisions are likely to be in the right direction. So it may be smarter to own stocks rated sell when the consensus is moving up than buys when it is moving down.

Analysts are also trend followers. They like stocks most when share prices have moved up, sales are growing and earnings are rising. Such companies tend to be investment-banking customers. A touch of glamour also makes them easier for brokers to sell. Even here analysts add value. Over time they are consistently able to pick the best growth stocks and avoid the also-rans.

The trouble comes when growth stocks tumble. That happened in 2000 and analysts looked like incredible dunces. Companies with the highest analyst ratings went down 30%, and those with the lowest ratings went up nearly 50%. But while those numbers were way off the mark, some analysts were gradually upgrading value and downgrading growth. So it was a great time to concentrate on changes in the consensus opinion, rather than the absolute numbers.

Now for the good news. The best returns in JL's research come from portfolios with the oddball mix of high ratings, good value scores and bad growth scores. When analysts make that rare contrarian call, the odds that they have a winner are unusually high.

Which leads to this month's screen. I looked for stocks with good scores on JL's value benchmarks, bad scores on growth benchmarks and above-average analyst ratings. Limiting my search to companies with a market value over $1 billion, I zeroed in on the bottom half of the pack in terms of price/earnings and price/book ratios. Next, to get the "bad" growth component, I looked for stocks with below-average estimates for future earnings growth and a history of slower-than-average sales growth. I added two final hurdles. I wanted companies with better-than-average analyst ratings, and I wanted the ratings to be moving up. My finalists compose an intriguing roster. These are relatively cheap, out-of-favor stocks that certainly wouldn't appeal to growth mavens. But analysts like them.

Maybe that's because several of these companies are in transition, a time when talking to management can be particularly helpful. Union Pacific is rebounding after its acquisition of troubled Southern Pacific. Toys "R" Us is beginning to see gains from its expensive store redesigns. And Quebecor and Valero are implementing big mergers.

AutoNation, meanwhile, has great potential if only it were as well managed as competitors. BorgWarner's fuel economy and emissions systems could be a bonanza, helping Detroit meet tighter clean-air standards. You get the point: Something could go right and make a big difference at each of these companies. Analysts are optimistic. And since the stocks are cheap, their ratings ought to have extra value.

.