In the fall of 2004, Bloomberg added a small feature to its formidable arsenal of analytical tools. Called the Bloomberg Absolute Return Rank, or BARR, the feature allows you to rank the analysts who cover any given stock by the return you would have earned had you followed their recommendations to buy, sell, and hold.

Other rankings of analysts—like the Institutional Investor poll—take into account squishy stuff, such as the analyst's personal charm! And some rank analysts by the accuracy of their earnings estimates. But the Bloomberg method is, in many ways, the most nuts and bolts: After all, the essence of an analyst's job is to make investors money.

Bloomberg hasn't turned this into a broader ranking of analysts or research firms. Bloomberg doesn't rank analysts or firms in any scientific fashion, but a survey reveals a surprising thing. A firm you've probably never heard of is ranked No. 1 on a large number of stocks. The firm, Matrix USA, covers 1,027 stocks that Bloomberg ranks analysts on. Its managing director, Ivan Feinseth, says that on 580 of those, Matrix ranks No. 1. The firm holds one of the top three spots on 870 stocks.

Matrix USA, which began publishing its ratings in the fall of 2004, is one of a crop of independent research firms that sprung up in the wake of New York Attorney General Elliot Spitzer's gigantic settlement with Wall Street. Unlike most of Wall Street, Matrix has no investment banking business to create a conflict of interest. But it's different from most of its rivals in other ways too. For example, Matrix doesn't produce earnings estimates. ("EPS is just accounting profit," says Feinseth.) Instead, the firm bases its analysis on a valuation methodology called EVA, or economic value added, which was pioneered by consulting firm Stern Stewart. In fact, Matrix has used Stern Stewart's intellectual capital to develop its research product.

Economic value added is a simple concept: It's a company's return on capital less the cost of that capital, and as Feinseth says, it comes closer than any other method to measuring the true economic profit of an enterprise. Matrix then uses a stock rating and screening system called PRVit (pronounced "prove it"), which stands for Performance Risk Valuation investment technology. (PRVit is based on a book called "Qwest for Value," which was written by Stern Stewart senior partner Bennett Stewart.) There are 12 data inputs for PRVit including measures of operating performance (such as sales growth), risk (such as leverage), and key valuation multiples for every company. Then, the system calculates each company's risk-adjusted performance versus its actual market value, and ranks each company relative to its peers and the rest of the Russell 3000.

This can lead to some contrarian results. For example, Feinseth loves to point out that under Matrix's methodology, Enron would have been rated a sell before its bankruptcy—a time when almost every Wall Street analyst rated the stock a strong buy. Twenty-seven analysts cover videogame maker Activision (Research). Only one of those—Matrix—rates the stock a sell. Investors following Matrix's ratings on Activision would have earned a 52.6% return over the past year by following Matrix's ratings on Activision.

Or take neutraceutical maker Martek Biosciences (Research). Last spring, when the stock was at $65, Matrix slapped a sell on it. It has since fallen more than 50%, to just under $27.

On top of that, Matrix also creates a Focus List, a more limited list of 12 to 25 stocks in which the firm's analysts have added human research and analysis to the quantitative model to come up with stocks to buy. The price of stocks on that list rose 20.5% in 2004 and 32.6% in 2005, versus 11% and 3% returns for the S&P 500 in each of those years. Thus far in 2006, the list is up 6.3%.

(There are no sure things in investing, though. Even Matrix has its mishaps.Krispy Kreme (Research) was on the Focus List in both 2004 and 2005, because its return on capital was sky high due to the inflated franchise fees it charged its franchisees.)

You might think, given this track record, that there would be huge demand for Matrix's service. In fact, Feinseth says that it is an "uphill climb." One very large, well-known mutual fund, which shall go unnamed, refused to use Matrix's research because the firm doesn't provide earnings estimates. However, given that research becomes less valuable the more it becomes conventional wisdom, the rigidity of some big investors leaves more opportunity for the rest of us.

See today's earnings.