Andrew Corn, the CEO of Clear Asset Management, has many reasons to like his money managers. They never whine about their Christmas bonus. They never threaten to leave and start a hedge fund. And best of all, they seem to be terrific stock pickers: five of the six separately managed account portfolios the firm runs are outperforming their benchmarks by at least 8 percent.

Of course, Corn's money managers are all computers. "We are believers that computer technology has evolved to be better than people at certain things," he explains – 'we' presumably meaning the human part of his New York investment startup.

So far, Corn's hunch seems to be right. Although the system has only been live since September, back-testing using the same models results in some extraordinary hypothetical returns, according to company figures. For example, one-year performance of 19.6 percent for large cap growth compared to 6 percent on the S&P/Barra 500 index, and a five year performance of 2.17 percent compared to the same index's negative 6.22 percent return. In large cap value, it beat the S&P 500/Barra benchmark by a narrower margin for one year – 26.19 percent  to 18 percent, but a wider margin for five years: 11.67 percent to 0.33 percent.  

Another sign that Corn, until recently a senior vice president of marketing at TheStreet.com, may be on the money about human money managers nearing their John Henry moment is the decidedly low-tech nature of the venture. The machines that run their proprietary models are just Dell computers loaded with a SQL database. The databases are filled with up-to-date Standard &Poor's Compustat statistical data on the stocks of 10,100 U.S. companies and American depositary receipts of foreign companies that traded in the U.S. markets. Using algorithms his team developed in eight months last year, the machines crunch through all that data and generate lists of their favorite 20 to 25 stocks in each of the six standard style categories, ranging from large cap growth to small cap value. After that, another tool balances and rebalances the holdings at set intervals as their relative values change.

Company executives are trying to get their accounts included in a number of wrap-fee programs, but at the moment the only way for an adviser to use the company's services is through the broker-dealer that holds its clients' stocks, FolioFN of Vienna, Va. Minimums are $25,000, with fees ranging from 80 to 200 basis points. Like more conventional separately managed accounts, Corn says that holdings can be adjusted based on the client's needs and preferences. Alternatively, Clear's research is also available by subscription, for $300 a year per portfolio. 

For all the company's spin about the novelty of computerized trading, such systems are familiar fixtures in the institutional investing world. Roger Ehrenberg, a seed investor in Clear Asset and until last October the director of Deutsche Bank Advisors, a multi-billion dollar trading operation, says that many hedge funds run long-short portfolios along similar lines. "Within the hedge fund community, there is a large contingent of model driven long/short managers who are applying the same discipline but are seeking to do that on both the long and short side," says the New York angel investor.

It was his familiarity with such approaches that attracted Ehrenberg to investing in the startup. "My background is in running proprietary trading businesses with a largely model-driven approach," says Ehrenberg. "I have always been a believer in taking the passion out of investing and really focusing on screening tools to optimize stock selection. So when Andy was putting the company together and described the algorithmic approach that was something that resonated with me as an investor."

In general, the models are designed to search for value, according to portfolio manager Osman Arain. Even the growth portfolio is designed to pick stocks that are fairly conservatively priced. It looks for strong price-to-earnings-to-growth (PEG) ratio, strong earnings and revenue growth, and a conservative financing structure.

Beyond designing the system to screen stocks for these familiar ratios, Corn says they have tried to identify quantitative factors often present in some of the greatest investors' stock choices. He says his team has analyzed the purchasing patterns of such luminaries as Peter Lynch and Warren Buffett and tried to reconstruct the factors present in the stocks they bought when they bought them.

But other factors may be just as key to Clear Asset's success as how its models perform. Ehrenberg says that he believes one of the big challenges the company faces is in helping individual retail investors understand their limitations. "Part of the challenge of having Clear succeed as a company, not simply as a series of models, is getting people to acknowledge that they make poor trading decisions," he says.

After 17 years on Wall Street trading desks, he says, he doesn't trade single stocks but concentrates entirely on asset allocation. "Now if I feel that way – and I know the math of the markets … how does your average retail investor stand a chance over time?" he asks.

Yet getting investors to leave all the control – and fun – of choosing their investments to an overgrown calculator is easier said than done. Ehrenberg says that he likes the way the company sends out a lot of information about the choices the computer makes, information they can then interpret and use in their own way if they choose.

To educate clients about the reasoning behind the computers' moves, Clear Asset e-mails frequent short research notes to clients and subscribers that interpret why the electronic oracles have made a particular choice. In a recent note, for instance, Clear Asset managers explained that the computer recommended selling Sears on October 22, just weeks before Kmart's takeover bid boost the stock's price 17 percent. "If computers could kick themselves, I'm sure ours would be. But being emotionless machines, they don't dwell on past decisions, positive or negative, at all," continued the researcher, before going on to explain that the computer had decided to buy Sears again on Dec. 3, probably because Sears has a price-to-earnings ratio of just 5 compared to the market average for comparable companies that have a P/E of 21.

Such exceptions as the missed opportunity at Sears may be one reason Clear Asset won't settle the debate on the superiority of the computer-driven investing anytime soon. One adviser who still relies on old-fashioned gray matter isn't so sure systems like Clear Asset will ever completely outmode old-fashioned stock-picking. Tad Borek, a financial adviser in San Francisco who picks some of his clients' stocks, argues that while such an automated approach is a good way to sort for value, another part of stock picking would be more difficult to replicate: an understanding of a company's story. "To me, narrative drives the performance of individual stocks, and sometimes, the narrative becomes very important. What's the story on this company? What's going on?"

Apple Computers, Borek says, is a perfect example of a narrative opportunity an algorithmic approach would miss. Although now Apple stock is at $70 thanks to the success of the iPod, a few years ago, it was selling for $11, almost equal to its cash on hand. At that time, the story about the company then was unrelentingly negative. "The gut reaction I had was, well, I think the Apple brand is worth more than a couple bucks; it's got some value to it. And there are some crazy people out there that love these products because they're really well designed and they're neat and all that and these people aren't going away, so I think it's worth more than the money in the bank."

The value of a brand – the value of Steve Jobs – the value of the iPod. Borek argues that all of those factors are important, but none of them can be reduced to an algorithm. That said, Borek has sold all his clients' Apple.  Of course, he was just following his own algorithm: "buy when the narrative sounds bad and sell when the narrative sounds good."