One finds the following "deal" described in Nicholas Dunbar's Inventing Money. The year is 1997; the page is 171.

Long-Term Capital Management, before the 1999 meltdown that destroyed it and occasioned Dunbar's book, enjoyed a phenomenal run. It reported gains of over 40% its first two years in existence. Together with the enormous influx of investors encouraged by such high rates of return, these gains pushed LTCM's invested capital up into the multiple billions, leading it to close its fund to new investors. Since LTCM charged especially high fees to its investors, its own returns were even more phenomenal. The value of its partner-owners' stake more than doubled in that two-year period. So, on the one hand, outside investors were clamoring to put money in, and on the other, LTCM's partners were looking for ways to get money out without paying huge tax bills.

LTCM recruited the Union Bank of Switzerland -- still smarting from its decision not to invest in LTCM two years prior and now frantically looking for a piece of the action -- to take part in one of the crazed option-swap schemes by which LTCM made its money. In essence, LTCM wanted to change its money from "income" into an "investment." Since investment is uncertain and capital flow makes the wheels of capitalism spin and investors should be encouraged to put their capital to productive use and blah blah blah blah blah, the United States taxes capital gains (profits from risk-bearing investments, such as stocks) at a lower rate than income (money made from predictable, "risk-free" activities, like laboring by the sweat of your brow or making a loan). Thus, LTCM asked UBS to help it launder some money. It was a classic "please hold onto my money long enough to fool the Feds" trick with some serious plot twists. Here's what they did:

UBS issued an $800 million option on LTCM (equal to about an eighth of LTCM's total value). If, at the end of seven years, LTCM's value had risen, then the option could be exercised to buy a share in LTCM at the older, cheaper, price, which could then be sold at the more recent, more expensive price. If LTCM's value had fallen in that time, the option would be worthless.

So far, so good. Except that UBS sold this option back to LTCM. And simultaneously bought an $800 million stake in LTCM to cover the option. Any increase in LTCM's value would affect the stake and the option equally, so that UBS's position in LTCM was neither positive nor negative. Whatever it gained on the stake it would pay out on the option. Bingo. Squeaky-clean money.

The astute observer may well point out that the above deal leaves UBS well and solidly fucked if LTCM's value declines. Options being voluntary things, options after all, LTCM could just refuse to exercise the option and walk away from the deal, leaving UBS holding a share worth less than the $800 million UBS had paid for it. To cover this eventuality, UBS' could, at its option, convert the stake into a floating-rate loan on normal investment-bank terms. That is, UBS' stake in LTCM was something approaching a pure accounting fiction, since it was either a hedge against having to pay out on the option or a regular old loan. From LTCM's perspective, the stake-option swap converted any future profit into low-tax capital gains, but wiped out the "risk" involved with any real investment.

The icing on the cake is that UBS turned around and invested over 90% of the payment it got for issuing the option directly in LTCM. Which is to say, LTCM's performance was tracked by an option issued by UBS, and that option was held by LTCM, for which it gave $300 million to UBS, which then invested most of that sum in LTCM. That a fair portion portion of LTCM's hedge fund deals flowed through UBS is incidental. One can see in such machinations, if one so chooses, the workings of a rational market on a search-and-destroy mission to eliminate inefficiency, in this case a difference in tax rates.

One might also see the financial markets as institutions for large-scale gambling (certainly, the existence for markets in weather futures does nothing to discourage such thoughts). Ordinary, old-fashioned gamblers, like ordinary, old-fashioned investors, make simple bets: tech stocks will rise, interest rates will fall, Bill's Bunion to place in the 3rd, the Ravens by six. And the arbitrageurs, those connoisseurs of liquidity and volatility, are the bookies: the middlemen who eliminate stupidity by preying on it, who take a small cut of the huge sums that pass through their hands and keep the whole market going. Bookies don't take risks; they run their spreadsheets and set the spread to keep bets on the favorite equal to bets on the underdog.

Which would all be well and good, except that bookies are gamblers, too. They don't so much reduce their risk as push it around, concentrate it into tiny singularities that can be conveniently swept under a corner of the carpet. They never lose money, and when they do, they lose astronomical sums. That's what happens when you move the line on the big game and the margin of victory falls between the old line and the new: you're paying out to both sides. That's what happens when you buy calls and puts on margin and the stock doesn't move even a hair. That's what happens when your martingale burns through your cash reserve and the casino bouncers stop only to remove the cocktail glass from your hand before throwing you out on the street.

And that, in the end, is what happened to LTCM and UBS. While the nation was busy with a sex scandal, Russia defaulted on some loans in a particularly brutal way, herd mentalities on Wall Street magnified small losses, other firms crowded around like sharks tasting the blood of one of their own, and that was all she wrote for LTCM. When LTCM bit the dust, the stake-turned-load UBS held went pennies-on-the-dollar too, and everyone involved was in a world of hurt. The Swiss bankers lost six-sevenths of the money they poured into the LTCM deal. The mere possibility that LTCM might cease to exist was the eventuality no one factored into the "risk-free" option swap.

Five billion dollars in capital made the odds on something like that happening pretty long, but not long enough for the meltdown not to happen. It's not an encouraging thought. The markets are supposed to be wellsprings of rationality, capable of compensating for the human inability to deal with large numbers and small probabilities. And it turns out that it's not so much that they don't get burned as that they get burned much less often and far far worse than the rest of us. Which is why I have some pretty serious concerns with letting these exemplary capitalists get away with the kind of tax shenanigans UBS and LTCM pulled.