"The Formula that Killed Wall Street"

[snot]

Don't think this changes anything for most of us, but interesting: http://www.wired.com/techbiz/it/magazine/17-03/wp_quant?currentPage=3 .

[girl gone mad]

1. Don't buy into this meme!

I had friends who worked on WS ('quants').

This is bullshit. The guys at the top wanted these complex mathematical formulas to hide behind. They knew many of the assumptions they were feeding to the underlings were flawed but they were creating plausible deniability for themselves.

[Doctor Cynic]

2. On the plus side, he's an almuni of my school! Woot!

[JohnWxy]

3. a very good article. These mathematicians should have talked to one engineer. He would have told

them that the real world has a way of making beautiful static equations produce ridiculous results. In the real world there is always a degree of uncertainty which NO SYSTEM can ever completely eliminate. This is like believing you have a betting system that will "beat the odds". You cannot eliminate the randomness of the real world - you have to prepare for it. OR another way to put this is when there are too many varables to keep track of you're probably going to miss something.

Another thing about Li's approach. I think he is taking the perspective of one entity extending credit and ignoring the interaction between entities. They should have simulated a whole system of such entities (i.e. a number of banks and insurance companies assuming default risks - AND interconnected by CDSs.) because they interact. ONe perturbration of the system (a bank failure) will propogate further stress throughout the system and could (and would have without Government intervention) cause other failures which produces more stress on the system - perhaps eventually taking the whole system down. The risk is going somewhere and when you don't have an infinite number of banks and insurers to diffuse the risk among the risk is going to reverberate back on you eventually (actually, rather quickly). Just because you bought a Credit Default Swap from somebody else and that takes the risk away from YOU that doesn't mean the risk goes out into space to nowhere. It still resides within the system.

A proper simulation would have shown what could happen. Simulations are used in just those cases where there are too many variables producing a vast number of permutations of interactions such that no number of discrete equations can cope with all the possibilities.

They should have simulated the entire system first.


Recommended.