25-Standard Deviation Moves, Several Days in a Row

From Peter Thal Larsen of the FT:

FT.com / Capital markets - Goldman pays the price of being big: For years, Goldman Sachs’s asset management business has been the envy of Wall Street. While other investment banks have struggled to establish a hedge fund arm, Goldman has raced ahead. At the end of May, it had $151bn of alternative assets, making it one of the world’s largest hedge fund managers.

In the past week, however, Goldman learned that being a big hedge fund manager can be a mixed blessing.... Its Global Equity Opportunities fund suffered such heavy losses that Goldman on Monday took the unprecedented step of putting $2bn of its own capital in the fund. Goldman is the most high-profile victim of the recent upset in quantitative strategies....

For reasons that are still unclear, shares began to move in ways that were the opposite of those predicted by computer models. These moves triggered selling by the funds... [to] meet margin calls from banks. This in turn exacerbated the share price movements.... [T]he GEO fund... [b]y Friday... had lost more than 30 per cent of its value. Goldman’s flagship Global Alpha fund, which uses quantitative strategies across a range of asset classes, has lost 27 per cent of its value this year.

“We were seeing things that were 25-standard deviation moves, several days in a row,” said David Viniar, Goldman’s chief financial officer. “There have been issues in some of the other quantitative spaces. But nothing like what we saw last week.” Faced with this situation, Goldman had to make a decision. If the bank did nothing, the GEO fund would probably be hit with large-scale redemptions in the next few weeks as investors pulled out.... But Goldman came up with an alternative: it would commit $2bn of its own capital in an effort to shore up confidence and to allow the GEO fund to ride through the downturn. It also asked some clients to join in. A handful of investors with strong links to Goldman, including C.V. Starr, the investment vehicle run by former AIG chairman and chief executive Hank Greenberg; Perry Capital, the hedge fund managed by a former Goldman Sachs trader; and Eli Broad, the real estate and insurance entrepreneur, chipped in another $1bn. Goldman yesterday denied that the move was a bail-out, arguing that it decided the GEO fund’s portfolio represented a good investment.

“We do not want to leave the impression that we have some moral or other obligation to shore up a fund. People who invest in a fund know what they’re investing in,” Mr Viniar said....

The bank also attempted to play down concerns that it might be facing similar losses on its own balance sheet as a result of proprietary trading bets. Mr Viniar said that, while Goldman does use its own balance sheet to pursue quantitative strategies, its exposure was “quite small”.

A broader question, however, is what damage the events of the past week mean for quantitative strategies. Mr Viniar said.... “What we have to look at more closely is the phenomenon of the crowded trade overwhelming market fundamentals,” he said. “It makes you reassess how big the extreme moves can be.”

The line that Goldman Sachs is pushing for public consumption is that its quants identify trades with good fundamentals, and thus that a temporary market move that loses it money is actually a good thing for the strategy--an opportunity to load up, take bigger positions, and profit much more as prices come back to normal, if prices do come back to "normal." Goldman has put $2 billion of its money where its mouth is, and it looks like about ten of its biggest clients have each put up about $100 million. This is either a sign that they have confidence that GS is right--ten clients put in a billion!--or not--even by twisting arms they could only get ten clients? and only $100 million each?--depending on your neurochemistry.

Renaissance is also pushing the "we are well capitalized, and this is an opportunity" line.

Comments

[David]

I recall reading that price movements tend to be normally distributed near the mean and power-law distributed at the tails. Do the "hedge funds" (that aren't actually hedged) ignore this solely out of mathematical convenience? 25 standard deviations? Do they really believe that? Is this an arbitrage opportunity or does the correct math show that leveraged strategies are bound to fail?

[anne]

Brad DeLong:

"25-Standard Deviation Moves, Several Days in a Row"

Think carefully of what Brad is telling us about statistics and about justifications for inherently flawed investment models. Brad is making a most important point.

[the idler]

Extreme value theorem? Fat tails? Long tails? Give me a break! My models are right, and reality is wrong.

[anne]

Brad DeLong:

"25-Standard Deviation Moves, Several Days in a Row"

Think carefully of what Brad is telling us about statistics and about justifications for inherently flawed investment models. Brad is making a most important point.

[the idler]

Extreme value theorem? Fat tails? Long tails? Give me a break! My models are right, and reality is wrong.

[D-Slam]

Wow, that quote from Goldman will be the signature quote for this mess.

Our models are good. It's not our fault. We're just unlucky. REALLY unlucky. Even if ever atom in the universe were a hedge fund, making one trade per second for the last four billion years, we'd still be the unluckiest one of them all. We're sorry, but hey (shrug), s*** happens!

[James Killus]

Ah crap. Here I go an do some snarky commentary on The Black Swan by Nassim Nicholas Taleb:

http://unintentional-irony.blogspot.com/2007/06/central-limit.html

and about how decent statisticians and scientists don't make the mistake of thinking that a gaussian distribution is about the tails of the distribution (they start arguing about whether the distribution is "really" lognormal, or Weibull, or whatever), when this guy Viniar comes along and demonstrates that there are plenty of people who are even stupider than Taleb says, and they control vast amounts of money.

Judas Maude, Viniar should be horsewhipped for a remark like that. Anyone with half a brain would take their money as far away from Viniar as they could get it.

[Barry]

Think mixture distribution. You have one model for the overwhelming majority of activity. The trick is that the extreme highs and lows are out there, and you need huge quantities of data in an 'ergodic' market market to accurately measure the tails. In fact, you might need a huge quantity of data to spot the extremes.