Feb. 17--Every year, three senior academics at the London Business School, Elroy Dimson, Paul Marsh and Mike Staunton, produce an analysis of long-term investment trends covering the world's major stock markets. It was sponsored by ABN Amro but has proved more resilient than the bank, and now comes courtesy of Credit Suisse.

There is nothing quite like it because, while its unrivalled long-term statistics make the case for equity investment in comparison to the returns from other asset classes, the analysis also makes clear just how uncertain equity returns are and how difficult it is to capture them. Indeed, anyone thinking of investing in shares for the first time should read this book before they buy. They may feel the need to lie down, during which time the desire will go away.

In a wealth of data this year, two things seem particularly pertinent. One is their challenge to the belief that if shares have plunged to lows, and if the world is not coming to an end, all one needs to do is buy good companies now and sit tight. That seems to be broadly what Warren Buffett has done over the years, and he is unchallenged as the world's most successful investor. So might it not work for others?

The academics' answer is it might, but don't count on it. They looked at all sorts of markets -- some high, some middling, some low -- and then examined how they performed in subsequent years. For the theory to hold, one would expect shares coming off a low base to do better than those where there had been no heavy prior falls. But it did not work like that. There was no discernible pattern, no statistical support for the theory over any reasonable time horizon. Cheap shares can stay cheap for a very long time.

Just how long is also startling and, it must be said, depressing. The FTSE 100 index hit its all-time high of 6930 on the last day of 1999. The academics calculate that the index has just a one-in-two chance of hitting a new high by 2019, some 10 years hence.

Even if the index is calculated with all dividend income reinvested, which has the effect of increasing its value, there is still only a 50-50 chance it will reach new high ground by 2014. Of course, for individuals this latter calculation does not work because tax and charges make it impossible to reinvest all the dividend income.

What this means is that if someone was sucked into the stock market at the time of the tech bubble around the millennium, they have only a 50-50 chance of getting their money back inside 20 years. But we should not actually be surprised by this.