Oil crash shows investors still trust flawed model

May 9, 2011

By Peter Thal Larsen
The author is a Reuters Breakingviews columnist. The opinions expressed are his own.

LONDON — Are oil investors clinging to flawed financial models? That’s the strong impression given by last week’s sharp oil price drop. If market movements followed a normal distribution, such a plunge would be extremely surprising. But what’s really astonishing is that investors still appear to be relying on risk management techniques that were comprehensively rubbished by the credit crisis.

The central flaw lies in believing that price movements in financial products follow a normal distribution around a central mean. Small fluctuations are common. Larger jumps or falls — the narrow “tails” of a distribution charted on a graph — are more rare. So anyone using this analysis to look at the oil market would have been very surprised by last week’s selloff. Clive Capital, the commodities hedge fund, has written to investors to point out that the drop in the Brent crude oil contract was a five standard deviation event, according to the Financial Times. In a normal distribution, such an event should happen on just one trading day in every 1.7 million — or about once every 7,000 years.

But this type of analysis has been comprehensively undermined by the credit crisis. To begin with, extreme movements happen much more regularly than they should if normal distribution data held true. Moreover, these fluctuations do not happen randomly, but tend to be clustered together. In other words, the “tails” are longer and fatter than the theory suggests.

Indeed, a glance at historical price movements shows the flaws of assuming oil price movements are normally distributed. As my Reuters colleague John Kemp has pointed out, the Brent crude price has moved, as it did last week, by 9.5 percent or more on 33 days in the past two decades.

Ever since the credit crisis struck, mathematicians and risk managers have been busily debating how risk models can be improved to better reflect the real world. It’s questionable whether any model can accurately capture complex, fast-moving and ever-evolving financial markets. But that does not excuse fund managers from clinging to an approach that was shown to be so disastrously inadequate four years ago.

One comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/

Would just be easier to quote NNT and “Black Swan.” But interesting to see that people are still not understanding that gaussian model’s cant be applied to stock market

Posted by agj1221 | Report as abusive