An old but much-loved joke in the theory of finance concerns a professor and his student. One day, as they were walking across campus, the student spotted a £20 note on the floor. Being a bright young chap, he said, Let's stop for a minute and pick up that note. We'll be £20 better off". "No," said the professor, "there's no point - if it were there someone would have picked it up already!"

In its own way this teaches the first lesson learned by budding index managers: that the major markets are efficient. In other words, so much information is available to investors at such a speed that attempts to trade on it are ultimately futile. Before we have the chance to use public information the markets have already processed it and built it into the price. This is the key argument advanced by index managers against traditional investment management.

An impressive array of evidence supports market efficiency. Academic studies and the reports of independent performance measurers consistently show a propensity for traditional managers to underperform. Median manager performance, with few exceptions, is below that of the index. Worse still, there is little evidence of continued good performance even among the best managers. Not surprisingly, studies of this kind have spurred the growth of indexation, in which managers concentrate on reproducing market performance at the least cost and risk to their clients.

It may come as a surprise then to learn that a growing number of quantitative managers are trying to beat the market. And some of them are succeeding. At first sight this looks paradoxical: why should those managers most swayed by academic thought and empirical data apparently reject them?

The answer, I believe, goes back to the early work on market efficiency. Much of this examined the behaviour of the largest companies in the major stock markets and concluded that these markets behave as though their participants quickly reach a consensus about fair stock prices. The precise details of the mechanism that led to this consensus were not usually discussed.

The studies of market efficiency raise two important questions, namely "What is this price-setting mechanism?" and, "How quickly does it work?". In trying to answer these hard questions quantitative managers have found ways to adopt an active approach to investment.

The first question deals with the way that an agreement about prices is reached and the information that is pertinent. Although this is still contentious much of the analysis suggests that certain factors - such as measures of value, size and recent performance can give useful predictions of future prices. Of course, these forecasts are very imperfect and would be dangerous to use as a way of picking individual stocks; but they can be worthwhile if they are used in a disciplined, risk-conscious way.

The second question - how quickly prices adjust - becomes of real interest when investing across markets or in under-researched areas. For example, although we can calculate a theoretically fair price for a stock option, this price is often not the one at which trading is taking place. If the discrepancy between theoretical and actual prices becomes wide enough there may be opportunities to make a profit without incurring risk. Such discrepancies may only exist for a few hours or minutes, but with today's technology this can be long enough to find, analyse and act on them.

Does all this mean that the advocates of efficient markets were wrong from the beginning? No, but the situation is, as always, more complex than it first appeared. Markets may be very efficient but they are not perfectly efficient; one consequence, I believe, is that there is a need to be careful about what we infer from performance surveys. The conclusion need not be that it is impossible to consistently beat the index, but that it is extremely difficult for managers, using traditional investment management methods, to do so reliably. In the major markets we might also doubt that outperformance can be simultaneously large and consistent.

Much of the activity in modern quantitative research is concerned with how opportunities for better performance can be found and used. These developments, with a widening appreciation of quantitative risk - and cost-control - should bring interesting times for all investment managers. The sharp boundaries between active and passive management are already starting to blur: we may soon find that they become increasingly irrelevant.

Carl Moss is director, quantitative fund management and research, at Kleinwort Benson Investment Management"