The Formula That Killed Wall Street

We recently discussed the perspective that the harrowing of Wall Street was caused by over-reliance on computer models that produced a single number to characterize risk. Wired has a piece profiling David X. Li, the quant behind the formula that enabled the creation of such simple risk models. "For five years, Li's formula, known as a Gaussian copula function, looked like an unambiguously positive breakthrough, a piece of financial technology that allowed hugely complex risks to be modeled with more ease and accuracy than ever before. With his brilliant spark of mathematical legerdemain, Li made it possible for traders to sell vast quantities of new securities, expanding financial markets to unimaginable levels. His method was adopted by everybody from bond investors and Wall Street banks to ratings agencies and regulators. ... [T]he real danger was created not because any given trader adopted it but because every trader did. In financial markets, everybody doing the same thing is the classic recipe for a bubble and inevitable bust."

Comments

the formula that killed wall street:

[Shakrai]

G+R+E+E+D

Re:the formula that killed wall street:

[Shakrai]

actually, greed is good. it's the great motivator. really, it's the only motivator.

It's a good motivator when it's tempered with wisdom. It's a bad motivator when you blinds you to the long term consequences of your actions. I've been saying for years that it seems like our entire economic system has been tailored to next quarters results at the expense of building/investing for the long term. Who cares if this quarter has record profits if you paid for those record profits with the future viability of your enterprise?

Nothing wrong with models.

[gravos]

There is nothing wrong with using a model. Models are good. They help us simplify the world so that we can understand it. For example, we have hundreds of competing climate change models that explain what is going on and predict what we should expect. We model the weather for forecasts. And so on.

But. And it is a big but. You must know the limitations of your model. By definition, a model is a simplification of a complex phenomenon. That does not make it flawed: that makes it a model. Overreliance on the model is your fault, not the fault of the model.

Re:Nothing wrong with models.

[morgan_greywolf]

However, there are some models that are just bad. If we take your climate change model example, simply going outside and measuring the temperature, and then comparing it to a temperature you took one the same day three years in a row and then plotting the statistical trend is a very poor model. Using that model, one might assume that we have drastic global cooling going on. It doesn't matter how much you rely on that model, if you rely on it all, you're going to be dead flat wrong.

Re:Nothing wrong with models.

[BrokenHalo]

You don't need an MBA to know there are bust-boom cycles.

You also don't need an MBA to know that there is a limit to the number of balls a juggler can keep in the air at any time before he drops one. And when one ball drops, the whole thing falls apart. As the truism goes, those who don't learn from history are doomed to repeat it...

Re:Nothing wrong with models.

[maraist]

I disagree. A model defines a static or pseduo-static system. It takes the non-linearities out of a system to make them as close to a linear, 1st order or 2nd order system as much as possible, such that you can produce matrices of inputs to outputs. All models also are accompanied by regions of legitimacy.. Namely the non-linear (or super non-linear) components press close to zero in these regions. Outside the regions, those non-linearities become too much 'error' for the model to be valid. Ideally, you can use separate models for different regions, and you have a nice continuum. But for that, you need to be able to first measure a region parameter.

The problem here is that you're talking about a model for an investment strategy that is inherently non-deterministic, non-linear and more importantly recursively adaptive. The region you're operating in, is part of the outcome variables.

Consider 3 investors each with equivalent information systems (including risk modeling, present-valuation, and product-viability forecast, whatever).

In a vaccum, a model, assuming a static system might be appropriate. Balance-sheets, due-dilligence, market trends, geo-politics, etc. are all valid. But consider that the other two investors have the power to effect the system. Consider that they can manipulate, propping up an industry, or willfully collapsing it (over-buying, or short-selling). By acting irrationally in the short term, they can sufficiently distort all the measureable parameters to your equation to force you to act inappropriately.

Thus by taking a short-term hit, one of your competitors can gain a much greater long term advantage.

Thus, KNOWING that you use certain models, allows your competitors to game the system.. Note they need to have significant resources in which to do this.. But the old addage that you need money to make money exactly applies here. Why would a wealthy person only accept 3% to 15% ROI when they can control certain markets and earn 500%.

Now explicit market manipulation is illegal. But there is nothing illegal about gambling (sadly). Thus betting against the 'known wisdom' is perfectly legal.

So now you have two camps.. Conservatives that trust their models (blind to the fact that people can manipulate them in the long-run). And advanced speculators who bet against the market. Over time, if one is considered unbalanced, then more and more itchy investors will switch from one side to another.. Until an equilibrium is reached where any and all metrics become meaningless - An equal proportion of investors will honor measureable data as there are people betting against the data. The raw data therefore has no material impact as to the future valuation of an asset. Note, as such a system evolves, the 'measureable' data will change over time. Namely instead of measuring the viability of a company, you measure the prospects for news and bet based on historical trends of the news outlets, not whether the news is good or not.

This can only happen if you have a gaussian distribution of strategies. Namely a massive pool of investors operating independently with an equal liklihood of choosing one of an infinite number of strategies, such that an equal ration of buy/sell decisions could be produced.

You can think of it as the classic "Is the poison in your drink or mine" attempt at gaming the system. Any number of strategies can be employed to decide which action is best, but the more you employ, the greater resemblance to random-decisions is created.

The short is, no formula can adequately valuate a market that is based on such a recursively adaptive system. Determining the risk of a car accident, a plane accident, a flood, etc. These are deterministic to a large degree (short of global warming and legalizing pot). But the college that first advocated investment strategies based on such finite metrics should be unaccredited in my view. A car owner isn't trying to game the insurance market, but a stock holder or company seeking stock value is.

One word

[DigiShaman]

Diversity.

Or there's my financial formulae

[Rosco P. Coltrane]

- Don't spend the money you don't have
- Don't do credit unless you absolutely have to

I know I know, Wall Street are these big finance hotshots who do complicated things that have nothing to do with personal finances, but what is it they do apart from speculating and playing with money they don't have, or other people's money? They just hide that simple fact under abconce financial constructs, but that's all they do in the end.

Bring back some morals sanity in the credit business and there won't be anymore crisis of this magnitude. No need for math here...

Not Wall Street. Us.

[computersareevil]

It isn't killing Wall Street. Those jokers are getting $billions$ in free money.

It's killing us, the people who work for a living and have to provide all those $billions$ or suffer the inflationary consequences when the Feds just print it.

Re:Not Wall Street. Us.

[Notquitecajun]

A BIG part of the problem is Washington's tendency to reward economic losers at the expense of the people who know what they're doing, and I'm NOT just talking about the poor. There are plenty of the high-salary types who have some sort of governmental loophole or backing that saves them when they screw a big company up.

It's one reason we don't need to be bailing out bad companies, and instead rewarding or backing up the good ones with incentives and tax cuts so that they can really succeed and push forward.

Thou shall not calculate behaviour

[Anonymous Coward]

This brings me to the crucial issue. Unlike the position that exists in the physical sciences, in economics and other disciplines that deal with essentially complex phenomena, the aspects of the events to be accounted for about which we can get quantitative data are necessarily limited and may not include the important ones. While in the physical sciences it is generally assumed, probably with good reason, that any important factor which determines the observed events will itself be directly observable and measurable, in the study of such complex phenomena as the market, which depend on the actions of many individuals, all the circumstances which will determine the outcome of a process, for reasons which I shall explain later, will hardly ever be fully known or measurable. And while in the physical sciences the investigator will be able to measure what, on the basis of a prima facie theory, he thinks important, in the social sciences often that is treated as important which happens to be accessible to measurement. This is sometimes carried to the point where it is demanded that our theories must be formulated in such terms that they refer only to measurable magnitudes.
It can hardly be denied that such a demand quite arbitrarily limits the facts which are to be admitted as possible causes of the events which occur in the real world. This view, which is often quite naively accepted as required by scientific procedure, has some rather paradoxical consequences. We know: of course, with regard to the market and similar social structures, a great many facts which we cannot measure and on which indeed we have only some very imprecise and general information. And because the effects of these facts in any particular instance cannot be confirmed by quantitative evidence, they are simply disregarded by those sworn to admit only what they regard as scientific evidence: they thereupon happily proceed on the fiction that the factors which they can measure are the only ones that are relevant.

Hayek. Nobel Prize Lecture, 1974.

Preposterous!

[Comboman]

nature of bubbles is that they burst when they reach physical limits of the stuff of which they are made. In our case it was human gullibility.

Preposterous! Human gullibility is one of the few things that has no limits.

Re:Picking up pennies in front of bulldozers

[Chris Burke]

If they are random then why do they predict economic change with 100% accuracy?

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA!

*gasp gasp*

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA!

*pant pant*

HAHHAHAHA, oh God that's rich. Seriously, you meant that? HAHAHAHAHAHAAHAHAHAHAHAHAHA!

*wipes tears from eyes*

No other reaction is possible for such a statement. Is this a delayed posting from 2007? Not that it wouldn't be equally laughable, but at least it was conceivable to maintain the self-delusion that it isn't. Today? You're saying the economy is 100% predictable, and you're saying this today.

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHA!