LIKE MOST AMERICANS

I'm inexplicably drawn to a good British scolding. I get my fix from television talent contests, the best of which feature a cranky Briton calling some hopeful singer or chef a train wreck or a donkey. The fascination, I'm pretty sure, has something to do with how predictably nice our reviews have become in these United States of Awesome.

Take analyst stock recommendations. In theory they should tell us whether or not to buy a particular stock. Two ratings would suffice: buy and sell, or buy and don't buy, or brilliant and donkey. Whatever. Instead, analysts soften the criticism with a hold recommendation. Last year 48 percent of recommendations were holds, but just 7 percent were sells, according to Bloomberg. That raises the question of whether a hold is a de facto sell, an endorsement (you can't hold a stock you don't buy) or just analyst indecision.

Given the ambiguity you might think it best to ignore hold-rated stocks altogether and focus purely on buys and sells. But that would be a mistake. Lurking in that pile of ho-hum hold ratings are some of the market's most attractive stocks. A new study suggests a way to extract them and stomp the broad market's returns.

Analyst recommendations aren't terribly useful for their intended purpose but are handy for other things. Decades of studies on the predictive power of analysts offer conflicting results, but a few clear themes show up. Analysts tend to beat the market during bull periods but underperform it during bear ones, perhaps because they focus largely on popular growth stocks, which thrive on market momentum and struggle without it. Sell recommendations tend to be far more prescient than buys. Considering the scarcity of sells, it's safe to assume that an analyst who issues one is truly convinced. Finally, changes in recommendations are more informative for investors than are the level of recommendations. In other words, better to search for stocks with recent upgrades than for those with plenty of buy ratings. Upgrades, after all, represent fresh information.

One last use for recommendations: When screening for value stocks, I sometimes purge the results of buys. Value stocks, I figure, should be unloved. How else are these companies going to win higher valuations by gaining fans on Wall Street once they turn things around?

A pair of finance professors recently set out to clarify what hold recommendations really mean. James Wahlen of Indiana University and Matthew Wieland of the University of Georgia created their own ranking system, using a mix of financial characteristics that have been shown in past studies to predict earnings increases. They started by observing that the return companies generate on their assets tends to "mean revert" over time. That is, extraordinarily profitable companies tend to attract competitors and see their returns on assets shrink, while companies with relatively low ROAs often take steps to improve them. So the two assigned high marks for low ROAs but also for signs that they're on the rise. These include sales that are growing faster than manufacturing and operating costs and that are also outpacing any buildup of assets. The professors also looked for companies whose cash profits were quietly topping the paper earnings that investors obsess over, something that bodes particularly well for future earnings.

For more SmartMoney Magazine features, turn to the July issue.

The study focused on the 12 years ended 2005, a mostly booming period during which buys made up 69 percent of recommendations and sells just 3 percent. Wahlen and Wieland sorted holds into five groups ranked according to their financial metrics at the start of each year. The top quintile indeed proved most likely to increase earnings. The two also created a long/short portfolio, designed to isolate the performance of the strategy and show how it did beyond the broad market's return. They simulated buying the top-ranked quintile of holds at the start of each year while selling short the bottom-ranked. The result was a 16.4 percent yearly return. Favored holds even beat buys in the study period.

Don't Hang Up When on Hold

Stocks with a "hold" recommendation are sometimes "buys" in disguise. Here are some examples: AGCO (AG) Landry's Restaurants (LNY) Multi-Fineline Electronix (MFLX) Owens & Minor (OMI) Sauer-Danfoss (SHS) Steelcase (SCS)

It's perhaps too early to draw sweeping conclusions from the study's results. Wahlen and Wieland's research is still in the early stages of the peer-review process, and further study seems likely. I'm curious to know how well the strategy works over a much longer time period or in bull markets versus bear ones. But the results make sense on an intuitive level. Unlike buy-rated stocks, holds don't carry the burden of high expectations. They also don't get touted by the brokerage arm of analysts' firms. That might help keep them cheap and thus likely to outperform if you can identify the most promising ones.

Readers won't be able to reproduce the study results exactly without access to an expensive research database and customizable sorting software, but they can come close with screening tools available on the Internet. I produced the list of eight stocks above by searching among holds for companies that simply demonstrate the characteristics mentioned above.

I recommended Agco shares at $15 and change in 2003 and $18 in 2005 in my SmartMoney.com column. Today they go for more than $50 but, remarkably, still look affordable at around 16 times this year's earnings forecast. The company sells tractors, combines, sprayers and other farm machines. Soaring demand for food and biofuel means its profits are projected to swell 33 percent this year and 25 percent next year. AGCO's margins are still sharply discounted to those of John Deere, but its shares are discounted even more relative to sales. Steelcase holds a 20 percent share of the office-furniture market. Two things give me pause on the stock right away: It has lost two-thirds of its value over the past decade, suggesting serious mismanagement. Also, there are two share classes with different voting rights. I'm eternally prejudiced against companies that use that ploy to preserve voting power for a privileged class of stockholders, because it can keep bad managers on the job too long. That said, there are promising signs for investors. Steelcase has slashed its manufacturing base, cut its workforce by half and focused on designing less-complex products. I'd wait for the earnings growth to materialize before buying, but the stock is at least priced for low expectations, at 12 times this year's forecasts.

Landry's Restaurants operates Rainforest Cafe, Saltgrass Steak House, Landry's Seafood House and other dining chains. Like most restaurant stocks it has been shelled over the past six months, reflecting tighter consumer spending and runaway food prices. Landry's generates about 20 percent of the sales of Darden Restaurants, owner of the Olive Garden and Red Lobster, and the two firms have comparable operating margins, yet Landry's has just 5 percent of Darden's market value. That has proved too tempting a deal for Landry's chief executive. He has made two offers to take the company private this year, most recently for $21 a share in April. Owens & Minor supplies doctors with things they use once and throw away: gauze, gloves, stitches, needles, gowns and the coverings that go over surgical scopes when they're inserted into uncomfortable places. It's a low-margin business, but the company is gradually grabbing bigger profits by gobbling competitors, eliminating unprofitable accounts and pushing more-lucrative private-label items. Profits are seen jumping 29 percent this year, making the stock's price/earnings ratio of close to 20 seem reasonable.

I recommended Multi-Fineline Electronix at $19 and change in an online column in March 2005. It soared to $65 over the following year. Now it's down to $18. Such is the boom-and-bust nature of manufacturing circuit boards for electronic devices and of a company that once depended on Motorola for four-fifths of its sales. MFLX seems to be diversifying its business: Motorola makes up less than 25 percent of sales; Sony Ericsson contributes 40 percent; BlackBerry maker RIM and Apple hover around 10 percent. But gross margin has soared of late, and the company is expanding production in China. Analysts reckon earnings per share will multiply fivefold, to $1.70, this fiscal year ending September. Finally, Sauer-Danfoss is rated a hold by the lone analyst who covers the stock. The company makes hydraulic and electronic systems used in mobile machines like backhoes, cranes and, crucial to recent results, tractors. Agco is a customer. Profits are on pace to jump 59 percent this year and are seen swelling another 29 percent next year.