When Garry Kasparov, the world chess champion, was beaten by a computer in 1997, he complained that the computer never became demoralised.

Jeremy Grantham, who heads the investment group Grantham Mayo Otterloo, tells this story to point out why computers have an edge over humans when making decisions about which stocks or bonds to buy.

Quantitative investing, which uses computer models to make trading decisions, is growing at twice the rate of the rest of the money management industry and has propelled forward a clutch of new names: Bridgewater, with $150bn under management; LSV, with $65bn; AQR, with $25bn; Intech, with $51bn, and First Quadrant, with $25bn, to name a few.

Sean Healey, the chief executive of Affiliated Managers Group, which has stakes in two such groups, says: "Quant is less dependent on an individual. You don't have to worry about losing the genius manager whose investing strategy cannot be replicated. Quant strategies can be replicated, and are highly scaleable."

A quantitative strategy based on a computer model can, unlike humans, invest $10bn almost as easily as $1bn.

As computing power becomes cheaper and more sophisticated, traditional investing has become increasingly quantitative, with the use of screening techniques to find, for example, the biggest cash flow generators. Increasingly, some firms are taking the next step and eliminating the human fund manager.

Critics of the trend argue that the quants fail to take into account human factors such as a dynamic chief executive who might drive up a company's share price.

Quant models, most of which assume that valuations will revert to a mean, are heavily based on historical data and so not equipped to deal with sudden unexpected events.

A recent Citigroup strategy report said: "Back-testing is vulnerable to extraordinary periods in investment history . . . Many investors dismiss the rise and fall of the TMT [telecommunications media technology] bubble as an aberration, but then base much of their quant strategy on data that will have been distorted by that aberration.

"Buying a quant strategy now might not be the cleverest thing to do after six years of value outperformance," said the report, which suggested that the rise of the quants has helped to price out many of the valuation opportunities in equity markets.

However, Bob Jones, the head of Goldman's quant group, says: "I've been doing this since 1985, and ever since then people have been saying, 'if everyone is using these anomalies, then won't they go away?' There are new anomalies all the time."

The Citigroup report continues: "Are investors over-valuing short-term cash flows [which are very easy to model] at the expense of long-term cash flows [which are very hard to model]? This might distort the market in favour of overly risk-averse companies that are starving their long-term investment plans to boost their short-term cash flows and dividend payments.

"Just as thematic investors encouraged TMT investors to over-invest five or six years ago, are value and quant investors driving corporates to under-invest right now?"

Ron Kahn, a former physics professor who heads the quantitative research team at Barclays Global Investors, says that computer models are becoming more sophisticated all the time.

"Long term, less tangible things are harder to model . . . but quant models in future will look more and more at these intangible things, because everyone will find out the simple things to model. We're always looking at obscure sources of data," Mr Kahn says.

"The way we pick stocks is the way any traditional person would pick stocks. What we do is simply provide academic rigour to the process," he says.

However, most quant managers agree that the strategies to date have had a bias towards value investing, which helps explain part of their outperformance in recent years.