I just read the book “When Genius Failed” by Roger Lowenstein and can’t help writing about the story. I have tried to shorten it as much as possible but I fear that I may have overreached at several places. I recommend that the book be read in its entirety.
Our story begins in 1988. John Meriwether (henceforth called JM), an MBA from University of Chicago, was the head of bond trading and also held the post of vice chairman at Salomon Brothers, a Wall Street investment bank which would be acquired by Travelers in 10 years.
It was the time when the efficient market hypothesis (EMH) was percolating from the dour academic world to the raucous trading floors. Most of the Wall Streeters were mystified by the new theories emanating from the academic world, but JM, a math teacher and MBA from Chicago, felt kinship with his academic brethren. In a quick succession, he hired Eric Rosenfeld - a Ph.D. from MIT and assistant professor at Harvard, Victor Haghani - masters from London School of Economics, Gregory Hawkins - a Ph.D. from MIT, William Krasker - an economist and Ph.D. from MIT, and Lawrence Hilibrand - again a Ph.D. from MIT. The group quickly took over the arbitrage group at Solomon and started trading the inefficiencies of the swaps, derivatives and the bond markets.
In an efficient market, everything is priced correctly and no one has an incentive to trade. But the market in which these people traded was still evolving and hence was often times incorrect and inefficient. The opportunities were aplenty and they thought that they had a recipe for success.
The arbitrage group, owing to their mental superiority and introverted natures, became incredibly close to each other. They played golf together, went on trips, gambled on a card game called “liar’s poker” and felt a deep kinship. Not surprisingly, with their haughtiness, cryptic languages and phrases, and a sense of superiority, they started rubbing the Wall Street traders the wrong way. This led to a us-versus-them mentality in the group and was a precursor to the divide between Long Term Capital Management (LTCM) and the rest of Wall Street.
In 1991, Paul Mozer, a trader, made a confession to JM that he had submitted a false bid to the U.S. Treasury. JM went to the CEO of Solomon and even after agreeing that the matter was quite serious, they decided to do nothing about it. A few months later, it was found that Mozer had committed numerous other offenses against the Treasury. The matter was reported to the Fed and not surprisingly the Fed was quite angry at Solomon for hiding the previous offenses. Buffett flew in from Omaha, made the CEO quit and then tried to save the firm. He tried to save JM, the top moneymaker of the company who was also known to be highly ethical. All failed and the SEC charged JM with “failure to properly supervise Mozer” and JM settled for an undisclosed sum. This left an indelible mark on JM’s reputation.
Lose money for my firm and I will be understanding; lose a shred of reputation for the firm, and I will be ruthless. - Warren Buffett
The lost reputation dashed all hopes for JM to succeed as the CEO of Solomon. On top of that Buffett was never really comfortable with the mathematical modelling of the arbitrage group and did not feel comfortable trusting the entire firm to JM. JM then decided to open his own arbitrage fund, and Long Term Capital Management was born.
JM immediately started raising capital for LTCM, the target — a colossal sum of $2.5 billion. A typical fund at that time started with around $25 million. Such a large sum was a byproduct of the way in which the JM group traded.
Their trades followed from a simple idea — over time the spread between risky and less risky asset will tend to narrow. In an efficient market, this makes sense. The risk associated with an asset decreases with time; as range of possibilities narrow with time. It is harder to see what may happen in a year than in a month. The uncertainty decreases and the asset price becomes less volatile.
So, the group starts with such an inefficiency. Let us say that for some reason the five-year Treasury yields 1% and the ten-year Treasury yields 3%. There is a two percentage point difference between the yields. Let us do a thought experiment. We fast forward five years to the future. Now we may buy a new five-year Treasury which yields 1% or a ten-year Treasury with five-year maturity period yielding 3%. There is no material difference between the two Treasury bonds and hence we conclude that this state of affairs does not make sense. The spread between the ten-year Treasury with five-year maturity and a five-year Treasury should be zero. We conclude that the two percentage point spread in the yield will go to zero in five years. The group will look at such a spread and think that this yield spread is too large and there cannot be this much risk. They will bet on the spread to disappear.
Such a bet gives a very small return — two percent in this case. So, they will borrow money and do this trade in very big dollars. By increasing leverage 10-to-1, a two percent return can be turned into a 20 percent return — nothing to scoff at! If a trade went against them for a while, they did not back down, but did more of the same trade. They were confident that they could do no wrong and for a while, the market seemingly agreed.
It was difficult to raise the target capital of $2.5 billion for a group who had no track record. Furthermore, it was hard to like JM and his team who were very reticent about what they were planning to do. They were intensely private and some of them were quite crazy about privacy. Although it was known that Solomon’s profits were largely from the arbitrage group, the precise figures were not public knowledge.
The clients had another gripe. JM, from the very start, planned to leverage the capital 30-to-1 and possibly even more. Seth Klarman from Baupost Group, had grave doubts about LTCM’s debt and when given an offer, declined to participate.
Two developments helped LTCM raise the necessary capital. LTCM was able to interest Robert C. Merton, a star financial genius from Harvard; and Myron Scholes, co-author of Black-Scholes model for option pricing. In a stroke of luck, Solomon finally decided to split the earnings from the arbitrage group and the group got their much-needed track record.
With the two star geniuses behind them and a track record, the firm gathered steam. They were able to raise the money on their own terms, with a three year lock-up period — an unheard duration for a hedge fund.
"I can calculate the movement of the stars, but not the madness of men" — Isaac Newton
LTCM had one of the best financial people at the helm. Luckily when they started in 1994 the stock market was seeing a very volatile year (in May, the 30- year Treasury dropped 16% from its recent high — a large move for bonds). The first year LTCM earned near 20% after fee — in a year when an average bond investor had lost money.
It is better to be roughly right than precisely wrong. - John Maynard Keynes
In the letter to their investors, the partners did a curious thing. They calculated the exact odds of them losing a certain amount of capital. The calculations were based on historical bond prices and volatility trends.
“Just as a handbook of poker might tell you that the odds of drawing an inside straight were 8.51 percent, the professors calculated that Long-Term would lose at least 5 percent of its money 12 percent of the time. The letter went on to state the precise odds of the fund's losing at least 10 percent, as well as 15 percent and 20 percent.” - Lowenstein
Their thinking was similar to what the Black-Scholes model says about option pricing. They defined risk as volatility. If the portfolio was less volatile, a little too quiet, then they borrowed more to increase the volatility. The idea was to pick positions which were independent from each other — like coin tosses. If two coins are tossed, getting a head on the first one has nothing to do with getting a head on the second one. If you think of your positions as having random movements, then using the historical data you can find out a number beta that essentially defines its correlation with a chosen index. If one can do this with many independent positions then the “noise” dies down and with very high probability you can predict what the outcome of the portfolio will be.
For example, if we toss N coins and count the number of heads — then it will be approximately N/2. If we keep increasing n then we expect that with very high probability the number of heads will be within a small range around N/2. This is the law of large numbers and is denoted by the bell shape curve.
The curve plots the chance of getting a particular number of heads. We see that most of the chance is concentrated around N/2, and the probability of getting anything which is not in a small range drops quite fast. Suffice to say that the exact odds can be calculated.
The prices of stocks and bonds do not behave in this way. The assumption that they are uncorrelated and independent is wrong. When the assumption itself is flawed, there is not much hope that the conclusions will be correct. If you are asked about a position in your portfolio, you will say that you expect it to be worth 50 percent more when the market realizes the value. But if you were asked about the chances of being 20 percent up in three months -- you will probably be bewildered by the question.
“A manager that has become overconfident by using a bad process is like somebody who plays Russian roulette three times in a row without the gun going off, and thinks they’re great at Russian roulette. The fourth time, they blow their brains out.” — Daniel Loeb
But using these definitions of risk and leveraging their portfolio accordingly, LTCM earned 59% before fees and 43% after in 1995 — the second year of existence. The group had an equity of $3.6 billion and assets of $102 billion. At the end of 1995, the group was leveraged 28-to-1, an astounding ratio. The fund continued to put up amazing numbers and earned 57% in 1996 (41% after fees).
Their only complaint was that their portfolio was not volatile enough — implying that they needed to use more leverage.
A major flaw at LTCM was lack of independent risk controllers. Generally in investment banks there is a person or group responsible for managing risk and making sure that the trades and leverage are not going to bankrupt the firm. At LTCM the partners were responsible for monitoring their own trades.
"When there’s too much money chasing too few deals, asset prices are driven up, prospective returns are driven down and risk rises." — Howard Marks
As the size of the fund grew, it became more and more difficult to find good investment candidates. LTCM started facing a lot of competition from the investment banks like Goldman Sachs, Lehman and Bear Stearns. The partners will find an opportunity and by the time they were prepared enough to make a large investment, the spread would close. Bit by bit the situation worsened and it became quite difficult to make money on bond or mortgage spreads.
The traders then came up with a few more spreads to bet on which were based on equities. In an ideal world, they would behave in a similar way as bond spreads. Another attraction was that equity positions will be uncorrelated with bond positions — or so they thought.
One idea was the spread on dual listings. A major position in their portfolio, Volkswagen, had two listings — ordinary and preference shares. The preference shares offered superior voting rights and hence traded at a premium. They looked at the historical premium that the market had paid and then came up with a statement about the current premium spread. If it was too high, they would buy one and short another - in the hope that the spread will narrow and they will make money. The size of these trades were stunning.
Long-Term bet $2.3 billion-half of it long on Shell, the other half short on Royal Dutch-without, of course, any assurance that the spread would contract. - Lowenstein
They also started betting on merger-arbitrages. When a company announces merger at a particular price, the stock price jumps but remains a bit below the announced merger price. This spread is in a way the risk of the deal falling through. As the fund had no idea if individual mergers were going to work out, they took a basket of such deals and bet on all of them.
"In a crisis, correlation across asset classes goes to one." — James Montier
In 1997, the crisis started with the collapse of Thai currency baht. The government had no foreign currency to support a fixed exchange rate and its effort to peg the baht to dollars failed spectacularly. To make the matter worse, it had acquired a lot of foreign debt and the country was effectively bankrupt. There was a crisis of confidence in Asia and the equity as well as bond markets plunged.
In August 1998, the unthinkable happened. A major economy, Russia, devalued the ruble and defaulted on its debt. The investors piled into the safety of Treasuries and the spreads widened so high that according to LTCM’s model they were an historical impossibility.
Long-Term, which had calculated with such mathematical certainty that it was unlikely to lose more than $35 million on any single day, had just dropped $553 million — 15 percent of its capital — on that one Friday in August. — Lowenstein
Every market LTCM was in lost money.
It lost $1.6 billion on swaps, $1.3 billion on equity volatility, $430 million on Russia and emerging markets, $371 million on directional trades in developing countries, $286 million on equity pairings, $215 million on yield curve arbitrage, $203 million on S&P 500 stocks and $100 million on junk bonds.
The partners had at the peak $1.9 billion of their own money invested in the fund. All was wiped out.