The Gambler's Ruin, as it is called, is a useful economic proposition. Briefly, if you go to Monte Carlo and ask what is the longest consecutive run of, say, red on a particular roulette wheel that has ever occurred, you will get an answer - for example, nine.

If you then take your available fortune and calculate the most consecutive losses that you can afford to bet on that colour, doubling your bet each time you lose, and still withstand the longest run against it that has ever occurred, you can attempt the following strategy:

First, bet £1 on red. If you lose, bet £2 on red. If you lose, bet £4 on red. If you lose, bet £8 on red - and so on. Eventually, you will win and get back all that you have lost, plus your original pound. At that point, you start again.

There is only one minor, tiny, trifling flaw in this procedure. Eventually, you must infallibly, if you keep it up long enough, lose all your money. The reason is that, soon or later, there will be a longer run against you, the bettor, than any that ever occurred before. At that point, you will be wiped out, irretrievably.

This type of betting is called a progression system. Usually, the progression is much less than two for one. Anyway, that is how it works.

Now, what happened to Long-Term Capital Management (LTCM), the Meriwether hedge fund catastrophe? Using extremely high gearing (on the order of 100 times its original capital), it bet on a series of relationships in different currencies and financial instruments.

Where the divergences were greater than the historical or mathematical norm, the manager expected that they would in due course narrow to their traditional relationships, based on underlying economic values. The problem is that, as every commodity speculator knows, a spread of this sort, like a bet on gold below its cost of production, can go very far and very long against you - for years, sometimes. Gold can be forced down below the cost of production and stay there for years. The spreads can widen instead of narrowing during a period of extreme economic dislocation. The one affecting us now is just such a period.

It is all very well to use your own money to bet on a spread going the way it logically should. If you lose, perhaps you are out 10 or 15 per cent for a while until things rectify themselves. But, if you are betting with 100 times your original capital, even a small move against economic logic will wipe you out before you can say "knife".

That is what happened to Meriwether. It is not that his team necessarily was wrong in its calculations. It is just that, in the rapture of having made huge amounts of money themselves in recent years, they assumed they could do no wrong and exposed themselves to an unforeseen event - which promptly occurred.

I do not have precise religious ideas, but I do observe that God is like that. He observes our proud manoeuvres here below and says to himself (or herself, or itself): "Oh, he thinks that, does he? He says that, does he?" Zap! Hubris really does tend to lead to nemesis.

There is a particular reason why this holds for these complicated derivatives transactions, namely, their very complexity. Readers may recall that, at the end of War and Peace, Tolstoy analyses orders at the Battle of Borodino. Napoleon was far enough away from the scene on the ground so that, according to Tolstoy, none of these orders could be carried out.

Similarly, it is not unusual for these immensely complicated derivative structures to get so tangled that one loses track of them. In other words, complication itself is a high degree of hazard.

I have mentioned that a characteristic of the latter phase of a bull market blow-off is the plausible variation of a margin account at that time. These arrangements are always different enough from the last time so one does not realise what is happening.

To use a very homely analogy, it is like the floor show in a seedy nightclub. A sequence of girls trots on to the scene, first a collection of Apaches, then some ballerinas, then cowgirls and so forth. Only after a while does the bemused spectator realise that, in all cases, they are the same girls in slightly different costumes. Thus, in past bull market blow-offs, we had real margin, Chinese paper, LBOs and the rest of it. This time - and, in fact, on one previous occasion - the so-called hedge fund actually was an excuse for a margin account.

Incidentally, in a particular irony, a number of the proprietors of LTCM borrowed on their own assets to buy out some of their limited partners just at the peak. In other words, not only did they lose the partnership's money, they lost their own money. This disaster should be a useful paradigm of the Gambler's Ruin - or, in this case, the Gambler's Rapture unless something even worse happens. Let's hope not.

John Train is chairman of Montrose Advisers, an investment manager in New York City