Short Term Capital Mismanagement

LTCM sounded like a good idea.

Gather a bunch of geniuses together and use their models to detect mispricings in the market. Then bet on these mispricings correcting. What could go wrong?

This works when these corrections occur whilst you’re still solvent. It’s very effective, in fact. In their first 3 years, LTCM generated 21%, 43%, and 41% in their first 3 years. But when you employ leverage ratios of 30:1, when the market moves against you, you end up in a lot of trouble.

Such was the extent of this trouble in the case of LTCM that the financial avengers had to assemble at the offices of the Federal Reserve in New York to discuss if it would be more financially painful to let LTCM die, or to provide them with the gargantuan amount of capital needed to stay above water.

Here lies a valuable lesson. 4 of them, actually. On reading Roger Lowenstein’s When Genius Failed, I was struck by 4 major flaws that were responsible for this debacle.

Source: Goodreads

The Flaws of Leverage

No matter how good your strategy is, using significant leverage will eventually kill you with probability 1. The more leverage you apply, and the longer you extend the investment horizon, the more likely this is to happen.

Imagine a fund whose returns are drawn from a normal distribution with mean 0.3 and standard deviation 0.2. These returns are volatile, sure, but in expectation the firm makes 30% per year! Now let’s apply some LTCM leverage on top of those returns and leverage our fund 30:1. Remember, the fund makes on average 30% per year without leverage. This means they should be making an absolute killing when you leverage these returns.

But this leverage makes them vulnerable to downturns; over a 20-year time horizon, the firm has a roughly 60% chance of going bust. If we consider smaller but more volatile returns, the bust probability is lower but still very real. Drawing returns from a normal distribution with mean 0.01 and standard deviation 0.02, it’s a still-non-trivial 25%.

In reality the situation is far more precarious, because of annoying characteristics of financial markets like fat tails and volatility clustering, which I will mention later.

Source: Of Dollars And Data

The Flaws of Academia

The most prominent academic works in finance rely in some way on the normal distribution. This distribution is the easiest to work with because of its unique properties. Add other fundamental classical statistics concepts such as the Law of Large Numbers, the Central Limit Theorem, Pearson’s Correlation Coefficient, etc. and you have most of the tools used by the major contributors to academic finance over the last 50 years or so.

Source: Towards Data Science, Medium

These theories, although elegant, don’t quite fit reality. Returns aren’t quite normally distributed – they have fatter tails, particularly in the negative direction. The Law of Large numbers only applies in certain circumstances. Same with the Central Limit Theorem. Returns may be non-stationary. Correlations between assets appear to change depending on what time period you’re looking at. This means that employing correlations and covariances in the analysis of portfolio construction is a questionable practice at best.

Source: Two Sigma

At this point, most people actually know these things. Econometricians are discovering new techniques to “correct” non-stationary data. Other academics are accounting for the autocorrelation and kurtosis of returns in Black & Scholes’s famous options pricing formula by adding ‘jumps’ to the distribution that returns are drawn from. There are others.

But I am sceptical of these ad-hoc adjustments. They don’t get to the core of the problem: that financial markets are complex and cannot be summarised by a static distribution or process. These models and statistics can be helpful, sure.

But whenever your entire strategy is based upon them, you’re doing something wrong.

The Flaws of Human Nature


None of this would have happened if these guys (as far as I can tell, it was all guys) just de-risked their lives appropriately.

At the time of LTCM’s collapse, it seems as if most of the top dogs were worth £20M+. At this point, especially in year-2000 £££, there really isn’t that much point accumulating more. Think about it – what would you do with more cash beyond £20M? Buy another house that you would barely live in? Buy another supercar? Fly private everywhere instead of first class?

These things aren’t the cake, nor the icing, nor the cherries, nor even the sprinkles. They are the light dusting of icing sugar to make the cake look slightly more aesthetically pleasing. They don’t make a material difference to happiness. At all. Once you get to that level of wealth, you are simply playing for sport. Or to prove how smart you are. Or to get power or fame.

Which is fine – but take your current wealth off the table. It shouldn’t be at risk. It certainly shouldn’t be in a fund levered up the wazoo.

Jesus, even beyond £500K or £1M I would de-risk most of my capital. One can live a very satisfying life using the dividends of an appropriately invested £1M.


But the partners at LTCM couldn’t do this, because of their egos.

You can understand where these came from. If you deposit £10M into my bank account, my ego may grow a little. If you hand me a Nobel it also may inflate a little. If my fund was famous, commanded vast amounts of capital, and could dictate terms to the largest banks in the world, my head may get slightly larger.

But just because something is understandable, doesn’t mean it’s right. These inflated egos led to rejecting terms that the partners should have accepted. It led to internal fights that inhibited the level of internal communication. Putting ego aside could have allowed LTCM to properly assess how much risk they were taking, and pull back a little.


This would also require the partners to admit that maybe, just maybe, the models they were employing weren’t as accurate as they thought. Maybe we shouldn’t be leveraging 30:1 to bet on tiny modelled discrepancies in spreads converging to their theoretically correct levels?

To be fair, this strategy worked tremendously well for a while. And it’s easy to get carried away when you’re doing well – it’s in our nature.

Whether that was due to luck or skill is still up for debate. But even if it was due to skill, the thing about financial markets is that they adjust. Specifically, in this case, new players enter the game to exploit the arbitrage opportunity being enjoyed by others. This means that more debt has to be utilised to achieve the same returns. Another thing about financial markets is that they change. Relationships that could be historically relied upon can evaporate. Or reverse.

The academic models failed to account for these two facts. It was arrogance that prevented the partners from acknowledging this possibility before they embarked on this endeavour. It was arrogance that stopped them seeing this even whilst the ship was sinking.

The Flaws of The Financial System


Part of LTCM’s problem is that they didn’t fully comprehend the connections between their “uncorrelated” positions. They thought that, looking at historic correlations, there was a minuscule chance of losing a large chunk of their portfolio on any given day.

They didn’t appreciate the complexity of financial markets.

Correlations change. This is now well documented: in times of financial distress, correlation goes to 1. Imperceivable connections exist between seemingly unrelated assets. Changes in the price of pork futures can somehow affect the Mongolian stock market.

These ethereal links are pretty much impossible to take account of. Hence, even today there is an underappreciation of the real risks most investors are taking.

Too big to fail

When Greenspan first heard about the trouble LTCM were in, his reaction wasn’t to shrug his shoulders and continue sipping his iced tea by the pool (it was summer, and I assume he had a pool). He didn’t let LTCM fail, the result of a natural and necessary mechanism of free markets. Something that would have happened if 99% of other businesses if they had fucked up to the same degree. No – he gathered banks and begged them to bail out LTCM.

Why? Why should LTCM get special treatment?

Because of the contagious effects that their failure would wreak on the financial system.

Because financial institutions are intimately connected. So if LTCM was declared bankrupt and forced to liquidate all outstanding positions, the banks who dealt with and lent to and invested in LTCM would take a hit, too. Such was the size of these activities that such an act may just send the whole financial system into a disastrously negative spiral.

Source: Encyclopedia Britannica

The banks were forced to choose between saving LTCM, and eating the losses and potential other negative consequences that would result in their bankruptcy.

The government faces a similar decision when banks fail en masse such as in 2008/9. The problem is these institutions are so embedded in the system that 1 inee weenie failure could bring the whole shit-show down.

And the winner is…

And governments don’t want this to happen. Because allowing failure in the short-term results in short-term pain. Politicians operate on 4-5 year cycles; they don’t think long-term. They tend to select the option that maximises their probability of re-election. So failed banks tend to receive bailouts.

Who does this hurt?

The government may lose a bit of approval, ok. Bank employees may get fired, sure.

But who really loses is, drum roll please, you.

Because the money to bail out these banks has to come from somewhere. The government can’t just create more GBP without severe consequences. They need to use money they have accumulated via taxation and debt issuance.

This money comes from the general population. When banks fuck up, we pay.

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.