The classic image of a Wall Street market trader is someone, usually a man, on the trading floor, shouting buy and sell orders, clutching a sheaf of trading tickets. Floor traders must have an instinctive feel for trading and get by on their quick wits. They are now a vanishing species and will soon join Mark Twain's riverboat captains in extinction and myth.

Modern trading is now almost entirely paperless and takes place in the cyberspace of computers and computer networks. The instincts of market traders are being augmented and in some cases replaced by mathematical pricing models. Traders are being drawn from schools like MIT, rather than the City College of New York. A feeling for market dynamics and trends will always be important, but along side these skills modern traders have a command of statistics and probability theory.

John Meriwether gained a measure of fame in Michael Lewis' book Liar's Poker, where he is described by Lewis as a Salomon Brothers Uber-trader and master of Liar's Poker. Meriwether was one of the top bond traders at Salomon Brothers and later became head of the fixed income securities department (which was responsible for mortgage security and bond trading). Meriwether was one of the first people on Wall Street to recruit mathematicians and physicists from schools like MIT and Cal. Tech and turn them into bond traders. Meriwether was a harbinger of the conjunction between Wall Street and the Ivory Tower.

Perhaps to the horror of the old Wall Street operators, academic financial theory provided a framework that allowed markets to function more effectively. One example of this is the Black-Scholes model for pricing stock options. Acceptance of the Black-Scholes model has become so wide spread that Web sites like Yahoo and E*trade that quote stock option prices also list the associated Black-Scholes values.

When Genius Failed, by Roger Lowenstein, is the true story of Wall Street traders, academics and hubris. It is the story of the failure of Long-Term Capital Management, a hedge fund founded by John Meriwether.

In 1991, when John Meriwether was the head of the Salomon Brothers fixed income security desk, a US Treasury bond trader in Meriwether's department at Salomon falsified a US Treasury bill bid. The scandal that ensued when this came to light put Salomon in danger of losing their status as a Treasury bill broker. The head of Salomon, John Gutfreund was forced out and Meriwether went along with him.

This left John Meriwether unemployed and very wealty. But Meriwether, to use Michael Lewis' term, was a Big Swinging Dick, a Master of the Universe, an Uber-Trader. Consignment to the golf course, even a very exclusive golf course, is not as gratifying as being a player in the market. In 1993 Meriwether took the first steps toward founding the Long-Term Capital Management (LTCM) hedge fund. As befits a Big Swinging Dick market trader, LTCM was to be a Big Swinging Dick of a Hedge fund, capitalized with two and a half billion dollars. For the privilege of having their money managed by Masters of the Universe, LTCM would also charge investors over double the usual management fees.

A hedge fund is an investment fund for wealthy individuals and institutions like banks and pension funds. The number of investors in a hedge fund is limited and they are restricted, in theory, to only those who can afford the risks which may be associated with the hedge fund. Unlike mutual funds, hedge funds are unregulated.

Although the story of the failure of LCTM and its subsequent bail-out, organized by the US Federal Reserve is inherently interesting, it is the stories of the people involved that make When Genius Failed difficult to put down. Lowenstein wrote a best selling biography of Warren Buffet and he excels at telling the stories of the people behind the events.

The Genius mentioned in the title refers to Robert Merton and Myron Scholes. Nine months before LTCM failed 1997, Merton and Scholes shared the Nobel prize in economics. Merton, Scholes and Stanford's William Sharp (famous for developing the sharp ratio to measure risk) are some of the founders of modern finance, which attempts to apply quantitative techniques to market analysis. Merton and Scholes jumped at the chance to join LTCM where they could not only apply their theoretical work but make a great deal of money.

The trading cachet of the LTCM trading group, headed by Meriwether and the stellar academic reputations of Merton and Scholes was joined by the final pillar of LTCM's power base: David W. Mullins who was vice chairman of the US Federal Reserve before joining LCTM.

As a Big Swinging Dick hedge fund with the most stellar partners and a huge capital base, LTCM was able to convince banks to lend them money at rates that were not available to lesser mortals (including investment banks like Salomon Brothers). LTCM used this credit to leverage their capital base by a factor of twenty to thirty times. In the first few years this allowed LCTM to make spectacular profits for themselves and their investors.

One of the flaws of When Genius Failed is that Lowenstein sacrifices the details of the investment strategies used by LTCM to tell the story (Nicholas Dunbar's book Inventing Money does a better job explaining the details of the LTCM's financial strategies). Many of the strategies that LTCM used involved derivatives. The term "derivative" has taken on an ominous cast because of the failure of hedge funds like LTCM. Derivatives are more innocent than their sinister reputation. A derivative is a security that derives its value from an underlying asset. A futures contract for corn, for example, derives its value from the underlying market price for corn. A stock option derives its value from the underlying price of a stock or stock index. Derivatives and the strategies traders use to make money on them can be complex and Lowenstein may have felt that these details would make the eyes of many readers glaze over. For example, Lowenstein never fully explains what exactly a "swap" is and incompletely explains LTCM's "volatility" bets. As we will see below, LTCM lost most its money on swaps and volatility bets, so these derivatives play an important part in the story. I've included a footnote here on interest rate swaps and volatility bets.

In the epilogue Lowenstein summarizes LTCM's losses:

This table makes it clear that the last two items, interest swaps and equity volatility bets, account for the majority of the losses. For the first four years, LTCM's trading strategies made huge profits. But no matter how good the trader is and no matter how good the models are, no one wins all the time. There will always be bad market bets. It is impossible to perfectly predict the future. LTCM's losses where large for two reasons:

• LTCM used large amounts of leverage. For every dollar of assets in the fund they borrowed twenty to thirty dollars to place their trading bets. In many cases the loans made to LTCM where the equivalent of signature loans. There were not backed by securities and there were no margin calls when the value of LTCM's assets dropped. Leverage greatly magnified LTCM's profits when they bet correctly. It also greatly magnified their losses when the market turned against them. Although LTCM's returns were impressive when they did well, their risk adjusted returns were not nearly as good.

• Low liquidity. Or to put it simply, there were few buyers when the market turned against LTCM.

  Interest rate swaps are contracts between two parties. They do not trade on an open market the way stocks, bonds or exchange traded options do. The options contracts that LTCM traded in their volatility bets were huge customized option contracts:

  Since long-dated options [LTCM was entering into 5 year options contracts] don't trade on exchanges, Long-Term had to tailor private options contracts, which it sold to big banks such as J.P. Morgan, Salomon Brothers, Morgan Stanley, and Bankers Trust. The market for such arcane contracts was thin, with only a handful of players who traded on a "by appointment" basis.

  When LTCM wanted to sell these contracts when they started taking losses, they could not get out of their positions at a reasonable price since there were few buyers. And the buyers knew they were in distress.

  The huge size of LTCM's leveraged capital base forced them into markets with less liquidity. Investment funds must worry about their trades having a market impact, narrowing or obliterating their expected profit margin. The larger the investment fund, the more this becomes a problem. They could trade in only the most liquid markets (e.g., stocks) or via customized contracts for swaps and options.

When LTCM crashed they had positions in securities with over a trillion dollars of face value. At the time many investment banks were losing hundreds of millions of dollars on similar market bets. There was concern at the US Federal Reserve that if LTCM defaulted on their contracts it would cause chaos and a market crash. This in turn could place the US economy in danger. So the Fed organized a bail-out. A consortium of banks and trading houses put four billion dollars into LTCM. In return the consortium took possession of LTCM's market positions. The investors that had money in LTCM got ten cents on their invested dollar. The partners were largely wiped out. Once LTCM's market positions were unwound the rescue consortium made money on their investment.

LTCM's mistakes were made fatal by massive leverage and lack of liquidity. If not for their huge leverage, LTCM could have survived these mistakes, or at least survived without such breathtaking losses. So while the mistakes themselves did not have to be catastrophic, it is interesting to consider how LTCM's "world view" contributed to their losses. At the heart of LTCM's trading strategies were two core beliefs:

1. The academic view is that the fluctuations (volatility) of a given stock and, in fact, the entire stock market follows a random course. The most articulate expression of this arguments can be found in Professor Burton Malkiel's book A Random Walk Down Wall Street. According to this view, volatility is distributed in a bell curve (a so called "log normal" distribution), just as people's height and weight are distributed in a bell curve. The larger the movement away from the mean (the center of the bell curve), the larger the movement in the stock price and the greater the potential risk. If volatility falls in a bell curve, risk can be estimated. The calculation of volatility assumes that the way a given security or set of securities has acted in the past will reflect the way they will act in the future. Since the past is known, the future can be modeled.

2. Markets are perfectly efficient. The actions of market traders will price securities correctly. A "mispriced" security will be returned to its proper price by the market. This is sometimes referred to as "perfect market theory". Markets may be out of balance at some point in time, but they will always move back toward balance.

When markets are stable and no events like the "Asian melt-down" or the Russian bond default perturb them, the assumptions above tend to be true. In fact, during the first three years of LTCM's history, markets were very placid.

Efficient markets show linear behavior that can be described with calculus and statistics. Academics like this because it leads to elegant results which make good journal articles, with nice equations. There is, in fact, some reason to believe that markets are efficient. But there should always be a caveat: markets are efficient, on average, over a long period of time. Unfortunately for those who actually trade in the market, short term effects can be anything but "efficient" and perfect. Unlike physics, where theory is eventually tested against experimental results, much of economics seems almost willfully ignorant of the way the market behaves.

Those who are active in the market, like George Soros, tend to discount academic theory. Anyone who hears news reports about the stock market will realize that rather than being perfectly efficient, markets are not always placid, but sometimes act like a manic depressive, gripped by wild euphoria one moment and crashes the next. A whole language has grown up to describe market "mood", complete with mascots: the bull and the bear.

Markets exhibit avalanche events: a seemingly small trigger can produce a massive wave of change in the market. In the case of LTCM, the avalanche was triggered by the Russian bond default. LTCM was not the only trading firm to lose large amounts of money. Virtually every investment bank lost money when interest rate spreads widened unexpectedly. This was largely because they were using the same kind of trading models. The black-scholes option model and its more sophisticated offspring (binomial trees) are popular with trading firms because they usually provide a good model for pricing stock and bond options. These models work well when the market is "efficient" and orderly. But markets regularly make a transition between order and chaos.

The movement from an efficient market to one that is sometimes wildly inefficient, where, in the case of LTCM, there are persistent mispricings of bonds, is not described by statistics or calculus. These events are best described as dynamical (or chaotic) systems. Dynamical systems are systems where action in the system feeds back into the system, producing a complex result. In markets these actions are produced by traders in the grips of panic selling (or in the case of the Internet bubble, euphoric buying).

By believing in perfect markets, LTCM left themselves exposed to huge risk when the market changed and became chaotic. One question buried in the saga of LTCM is: would it be possible to recognize the switch in regime, from a market that is relatively efficient, where mispricings are corrected in the short term, to a market which is chaotic, where mispricings can be unexpected and long lived.

A number of computer based market models have been proposed that act like real markets. Some of these rely on market models based on a set of dynamical equations. Others rely on a market composed of rule based market actors, which simulate trading in the market. These model produce volatility curves that have "fat tails". That is, bell curves that show extreme events on either side of the center. Although these models act like a real market, they do not necessarily mirror the market.

In theory a model could be built that would predict market regime shift. If we had unlimited amounts of computing power and the ability to simulate humans, with their judgment, fear and greed, we could build a simulation that would perfectly model market behavior. When such a synthetic market got certain inputs, like the Russian bond default, the market will become chaotic. The market could then be run forward in time, telling us what the likely result would be.

There are obvious problems with this thought experiment. Humans and the interactions between them are vastly too complex to simulate. Market information, like the trading volume or close price of a stock is internal market information. Humans act on information (news) that is external to the market. The interpretation of news depends on past news (the Russian bond default might not have produced such an extreme result if it had not been preceded by the Asian melt-down). Finally, even if such a predictive market simulation could be built, it might not be able to recognize a regime shift in time for the information to do any good. The 1987 stock market crash happened in a very short period of time, for example.

Building a market simulation that could predict a regime shift is a daunting task. But it might not be an impossible one. It may be that traders generally act on a limited set of rules, which could be modeled. Instead of trying to interpret news, it might be possible to use the internal characteristic of the market (increased volatility and internal trends) as triggers for the market actors. Black-Scholes and related option pricing models made a great contribution in the past. Dynamical market models may be the next regime.

Web based references:

• LTCM Speaks, by Joe Kolman, DerivativesStrategy.com

• Salon's review of When Genius Failed, by Alan Deutschman

Disclaimer

I work on software for quantitative finance. This does not make me an authority in this area. Although I would like to think that I am a master software engineer, what I don't know about quantitative finance literally fills bookshelves. Nothing that is written here should be interpreted as reflecting the views of my employer. Nor should it be read as an indication of the kind of work that may be going on at my employer. The speculation about actor based modeling has been influenced by work at the Santa Fe Institute more than any other source. Finally, www.bearcave.com is my domain and these opinions mine alone.

Ian Kaplan - October, 2000
Revised:

Book review table of contents

back to home page