Wall Street’s Math Wizards Forgot a Few Variables

http://www.nytimes.com/2009/09/13/business/13unboxed.html

by peter123 • 17 years ago

11 12 17 years ago

Accidentally bad math or intentionally bad math? I recall the great article posted here, written by the self identified godfather of the software to slice and repackage risk. And I always wondered, did he understand the difference between correlated and uncorrelated probabilities?

jhancock• 17 years ago

Good call. There is plenty of blame to go around. The models the quants used served their purpose of recording "profits" "right now". I think being short sighted was intentional or at least "unconsciously intentional".

joe_the_user4• 17 years ago

Yes, "They either didn't understand that, or intentionally ignored the possibility that, the random variables they were working with would eventually be correlated" is great summary of math BS for laymen.

However, I think that the "art" of the whole affair was looking at a domain just complex enough that even they themselves couldn't sure whether they were lying or just "being clever".

Dilpil• 17 years ago

"Quantitative analysis has been getting more common in the past 10 years. We had a major crash recently. Therefore Quants caused the crash."

This is a very popular sentiment, yet utterly false. The companies that failed hardest- Lehman, Sterns, Fannie/Freddie, rating agencies- are (were) notoriously non quantitative. They are frequently made fun of in Quant circles in fact. Were any of these companies run by Math Wizards? No. Most of them have (once again, had) zero Quants in their upper management. You cannot blame the failure of such institutions on Quants.

joe_the_user4• 17 years ago

Oh you've got me started...

A whole of array individuals and idea-systems bare blame for the bubble. From the Quants' model to Alan Greenspan's ideology to the to mark-to-market-accounting to the repeal of Glass-Steagal and onward, they justified what was effectively the largest bubble in history. And given its size, the blame can go many places...

Moreover, you're acting like the crisis was just the bankruptcy of a few companies - the ones conveniently without Quants. The crisis has, in fact, hit the entire economy. And moreover, all of Wall Street was effectively bankrupt at the point when Lehman failed. The Fed just chose to throw some trading houses to the wolves and bailed out the rest. If the bailed-out houses thought like you, I wish they'd let them go too.

And further, you are not in the least addressing the points in this and other articles, the problems with Black-scholes, the Gaussian Copula and so-forth.

nazgulnarsil2• 17 years ago

i have yet to meet a quant who understands what curve fitting is and why it's bad.

bdr• 17 years ago

Alright, alright, how many quants have you met, and did you quiz them?

chasingsparks• 17 years ago

Thought experiment:

You find an algorithm that is very reliable for forecasting the near-term prices of an asset. Your algorithm only works in contemporary, or at least recent, markets. You notice that by building a basket of many assets, you can increase your leverage as you have a reduced the volatility of returns. You DO UNDERSTAND that there is a possibility of major failure, but you assign it a low probability and decide to let it ride for the time being. After all, after a certain level of profit, you can scale back the risk. The profits are too good, so you never scale back. Thus, you are the one standing when the music stops.

The end.

joe_the_user4• 17 years ago

Ah, but you don't have to even be that clever...

If you just come up with an algorithm that mostly says 'buy', then you can look good for as long as a secular bull market lasts. And if your algorithm sounds amazing, then you can get lots of leverage without otherwise doing anything.

Of course, this story ends the same way, you're left standing when the music stops (along with most of Wall Street) - but with luck, the Fed will buy your toxic assets...

IgorCarron• 17 years ago

Left standing ? I am not sure.

http://www.nakedcapitalism.com/2009/09/ny-times-lehman-post-...

Wall Street’s Math Wizards Forgot a Few Variables

Related Stories