WIRED: The Secret Formula That Destroyed Wall Street

[On the Road]

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. And it became so deeply entrenched—and was making people so much money—that warnings about its limitations were largely ignored.

Then the model fell apart. Cracks started appearing early on, when financial markets began behaving in ways that users of Li's formula hadn't expected. The cracks became full-fledged canyons in 2008—when ruptures in the financial system's foundation swallowed up trillions of dollars and put the survival of the global banking system in serious peril.

http://www.wired.com/techbiz/it/magazine/17-03/wp_quant

Four-page article, but very insightful into how so many people got such basic things wrong. Also linked is an article on how to prevent secret labyrinthine arrangements like these from getting out of control.

[Gin]

1. money was their drug of choice... and their habit did us all in!

[On the Road]

2. Money is a Drug for a Lot of People

but they don't bring down the whole economy.

The so-called financial wizards didn't understand correlation. That's what was wrong with the equation.

[gratuitous]

3. It's probably beyond my comprehension

But I'll bookmark it for later anyway. However, as the old saying goes, "When everybody thinks the same thing, nobody's thinking." When the real estate markets were beginning to heat up, fixing up older houses and then selling them was a good way to make some nice cash. Values were increasing anyway, and with some remodeling and some elbow grease, a marginal property could be made more desireable, and a tidy little profit could be turned in a relatively short time, like a year or so.

But as the market began to overheat, values were rising too quickly, and instead of taking some time and effort to do the fixing up properly, more folks began to get in, slap a little paint over some cracks, and call it good. Flippers turned properties around in months, if not weeks, and the inflating real estate bubble got hooked up to a jet engine of inflation. When the market finally stripped its gears and stalled, a bunch of people got caught, and a bunch of paper profits suddenly became bad debts.

I'm guessing that Li's little formula didn't have a provision for the eventuality of a market suddenly seizing up like that.

[On the Road]

5. That's a Pretty Good Account of It

for not having read the article yet.

The assumption was that individual foreclosures are not highly correlated. That's where the formula comes in. All you had to do was go back to the late 80s to see that that wasn't true. Some of the analysts bought it; other didn't. But by and large, all their bosses bought into it because it was the answer they wanted.

Check out that other article, too. It's about how using using financial markup language can lead to radical transparency, cut through the ocean of financial data, and prevent the wall of secrecy that allowed these things to get so big.

[NoodleyAppendage]

4. The product of Li's formula is disaster. I guess one of the entry variables is GREEDY BASTARDS. n/t

J

[WhiteTara]

6. he ran like a rabbit to get out of the country

he should be extradited back here for trial.

[Dreamer Tatum]

7. Wait - someone who wrote a formula should be on trial?

Why?

Should Fischer Black and Myron Scholes be in prison? How's about William Sharpe? Modigliani? Miller? Jarrow?

Bodie, Kane, and Marcus write a famous textbook about investments...should we toss them in prison, and stage a book
burning?

The people who discovered oxycodone and hydrocodone: wow, look how many people get addicted to those drugs. We should
totally put those guys on trial.

And my neighbor had a terrible accident in his Acura, so we should have the designers of the TL extradited, too.

(are you getting the idea?)

[HamdenRice]

8. I had no idea it was this bad. If you can understand, this is the most important article yet ...

about the financial meltdown. This is new. Even though back in the stone age (90s under Clinton's SEC), I used to design mortgage backed securities, and followed the news about them after I went into the non-profit sector, I had no idea that this formula had come to dominate pricing.

Frankly, I can scarcely believe or understand how such an obviously faulty formula was used to price trillions of dollars in debt securities.

I realize that the article may seem complex and throw lots of math around, but at the bottom, I can explain this in incredibly simple terms that any high school graduate can understand, and that any high school graduate would see was faulty. The fact that anyone thought this guy could win a Nobel Prize in economics, let alone would price even $10,000 in bonds based on this, is stupefying.

So if you want to understand Li's formula and why it was so disastrous, just read this sentence from the Wired article:

"Li wrote a model that used price rather than real-world default data as a shortcut (making an implicit assumption that financial markets in general, and CDS markets in particular, can price default risk correctly)."

Now, let me explain what this means. The problem was how to measure risk on various debts -- mortgages, mortgage backed securities, bank loans, and so forth. This was Wall Street's problem; they wanted to know how to measure and price risk of default.

Before this formula came out, most people would have said, you need to look at real world data. How often to people default on their mortgages? How often to corporations default on their bonds? That's referred to as historical data in the article.

Li said, there's no reason to look at historical data. Instead, if Wall Street wants to know the likelihood that a debt will default, just look at what Wall Street itself thinks is going to happen based on the prices of credit default swaps.

Here's an analogy. If you want to price a stock in a company, in the gold old days, you would look at that company's financial statements. How much revenue is it producing? What are its costs? How much profit is it making? How long is it going to make that profit?

Li's formula would by analogy say, there's no need to look at any of that data. The only thing you need to look at is what the stock price is according to Wall Street traders.

In the debt market, that's the equivalent of cds prices.

In other words, Li's formula is a perfect feed back mechanism, a circular loop. He's saying that to understand how Wall Street should price risk, look at how Wall Street prices risk.

In other words, it was a system of risk analysis and pricing without data about the real world. It was a math formula that said, what people think the risk is, is what the risk is, even if those people have no data, because they are using my data-less formula -- but dressed up in a seemingly incomprehensible formula.

[Odin2005]

10. Yikes! Somebody send that Li fellow to a basic logical reasoning class.

My god, a child could see through that BS thinking.

[HamdenRice]

11. I don't get it. How could anyone in a bank not have seen through this??? nt

[originalpckelly]

14. I seem to remember that people are fallible.

It's the primary problem of our fair capitalism, it's really not capitalism, it's all planned economics. It shouldn't be necessary to collect historical data in a free market on such a large scale. The problem is that they want to know the unknowable, or at least the unknowable without spending more money to figure it out.

I would guess that it takes more analysts (aka planners) to determine the risk level of an investment than it does to base it off of a bullshit feedback loop. It's really easy to look-up the price of something on an index, it takes a lot of work to go research the loans being made and the people receiving it.

What happened is that they wanted to get away with a real market, when in reality it's a planned economy. Price indexes work in actual markets with actual products, like goods and services, not esoteric financial products. There is a decoupling of the perceived value of a product and the actual value of the product, and it would seem to be an open invitation to faults like what's gone on. There is no direct feedback, aside from dividends, in stocks. A company can be running a loss, which is totally unknown to the public, until it releases it's earnings. The stock price doesn't reflect the diminished value of the underlying company until the information is released, when in reality the value of the company was always decreasing as it lost more and more money. It's this latency between an event occurring to increase/decrease the real value of something and the reaction to the event that is killing our system. Feedback builds up, and then is released in shock waves, instead of smoothly correcting a price down, it happens all at once.

It's an information problem, and these idiots tried to solve it, but failed miserably. The only way to solve it is to reduce the cost to the system to react, to decrease the latency between the feedback and the reaction on the part of a decision maker.

[originalpckelly]

16. You can't measure a system without disturbing it. The observer effect:

http://en.wikipedia.org/wiki/Observer_effect_(physics)

You can't plan (through measurement) an economy, without disturbing the economy (cost of measuring.)

[daleo]

27. Nice explanation. n/t

[DLnyc]

20. I think the question should be "why didn't they WANT to see through it?"

I personally pointed out to more than one person in the 90's that their system of assessing risk based on the market price of derivatives was a catastrophe waiting to happen. The response I got, each time, was essentially that "we're making tons of money doing it this way, everyone else does it this way, and if we use a more realistic estimate of the risk, we won't be competitive."

[KittyWampus]

33. some did. I remember Krugman writing about it. And apparently AIG had accountants rooting through

books at the Financial Division and then told to buzz off when he started to ask questions.

[BlooInBloo]

24. Here's Li's paper - what Li ACTUALLY said. Kindly point SPECIFICALLY to his logical error...

Else admit you don't know what the fuck you're talking about.

http://www.rybinski.eu/resources/non-modules.d/dispatcher/dispatch.php?id=2373

Thanks!

[originalpckelly]

26. Here: from page 10

"The suggested approach is contrary to the use of historical default experience information provided by rating
agencies such as Moody’s. We intend to use market information rather than historical information for the
following reasons:

• The calculation of profit and loss for a trading desk can only be based on current market information.
This current market information reflects the market agreed perception about the evolution of the market
in the future, on which the actual profit and loss depend. The default rate derived from current market
information may be much different than historical default rates."

Perception doesn't indicate anything about actual risk of default. Perception can be based upon incomplete information. Note he says "current market information" and that's the key here.

An example:
A stock is trading at $40 dollars based upon the perception that earnings the company reports will be in line with expectations. If the company misses, then the price of the stock will probably decrease. The actual value of the stock was always decreasing (as the company didn't make as much as expected), but that was not actualized in the price of the stock until more information was presented.

That's stock price, but I'd bet it's probably the same for the price of a CDS. The problem is that the perception is not one based upon direct observation, but indirect observation.

That's my first criticism, simply because most of the other stuff in the paper is couched in terms of art for the financial industry.

It would seem there is a latency of self-regulation for an observer, as distance from the object of observation increases, then the latency of self-regulation will increase. (The latency of self-regulation being the time it takes for an adjustment of the price.)

[BlooInBloo]

28. "most of the other stuff in the paper is couched in terms of art"...

I.e., you don't know any math.

Thanks for admitting you have no idea what he's talking about, at least.

[originalpckelly]

29. No, I'm terrible when it comes to numbers.

It's a totally foreign language to me.

kinetic energy = (mc^2/sqrt(1-(v/c)^2))-mc^2

or in better formatting:



Dividing v by c, shows the fraction of the speed of light that the object is traveling at. As v approaches the speed of light, c, the number given by dividing v by c becomes larger. This number is then squared. It is then subtracted from 1, to reverse the decimal and then the square root is taken to reverse the earlier squaring. Then it is divided into the rest mass, which is found by mc^2. As the number on the bottom becomes less and less, then it causes the number on the top to be denominated in smaller units, making the number on the top of the division sign more. It's like inflation. The number that results from this operation is the combined energy of both the rest mass, and it's kinetic energy. To find for only kinetic energy, you subtract the rest mass (again mc^2.)

As one gets closer and closer to the speed of light, it takes more energy to keep going, and this equation suggests that as one approaches light speed, the energy required to accelerate past it is undefined.

Math is just a language that demonstrates relationships in quantitative terms. I do not fully understand some of the stuff he was talking about, but I'm not totally ignorant either.

You also didn't respond to my statements, which I believe show the inherent flaw in the stuff he did. Observers play an important role in our universe.

[BlooInBloo]

30. I didn't respond because it was fucking stupid.

EDIT: And congratulations on your ability to quote something.

[originalpckelly]

31. What? How is it fucking stupid to think prices don't always reflect the accurate value of something?

It's not fucking stupid, it's common sense.

[HamdenRice]

34. He's exactly right, Bloo. The math may be right, but the basic assumption is wrong

Rather than figure out what the risk is, he simply assumes that the market knows what the risk is (through the price of cds).

Obviously the market did not know what the risks were. That's why we're having a market meltdown.

No matter how sophisticated the math is after that, the paper is basically wrong.

[originalpckelly]

12. He made a feedback loop, like someone programming a computer.

What a stupid fuck.

This is why planning is so bad, it's all planning too, and that's why they were in a position to make such a silly and easy to see mistake: people tend to see what they want to see when they can make a lot of money off of it.

[sixmile]

13. In other words -Free Market Capitalism

The value of an asset is whatever you can get someone else to pay.
If no one wants to buy, you lower the price until you hit zero - then you have a non performing asset. Worthless.

The real scam here is that property lasts forever, and these 'toxic' assets, though not highly sought after at current prices, definitely have A price.
The government and its media lapdogs have sold us a bill of goods. I say let 'em fail. If free market capitalism helped them get rich, let's see it that works on the way down, too.

[originalpckelly]

15. This is an information problem.

It's going to raise some very interesting questions about the foundations of our system, and I suspect just as in physics as the great problems have come, they have been the build up to some kind of breakthrough understanding of the universe. It is probably true about economics as well.

[DLnyc]

19. Very well said. You picked up on exactly the idea I had picked up on:

From the wired article:
"Li came up with an ingenious way to model default correlation without even looking at historical default data. Instead, he used market data about the prices of instruments known as credit default swaps"

Sadly, though, this screwy idea has been around for quite a while. In the early 90's I was working on coding some models in the swaps and swap-options area. It was absolutely STANDARD PRACTICE then to derive various key (volatility) parameters from prices in the market. So, basically, the answer to the key question, "Gee, how risky are these things anyway?" was being answered by saying "They must be exactly as risky as all the other players think they are." Unfortunately, all the other players were using exactly the same logic, so no one had any idea, and no one worried about, how risky the things actually were.

[Dreamer Tatum]

22. Said yet differently

All market information and adjustments are perfect, immediate, and common knowledge.

Idjits.

[TahitiNut]

35. Shorter explanation: "Wall Street 'consults' its own navel."

The (so-called) 'technical' approach to market analysis is nothing more nor less than lemmings worshipping their own herd mentality and presuming that "getting out front" is the same as 'success.' (Splat!)

[originalpckelly]

9. Capitalism is all about math and physics in the final analysis.

:P

That's what's so wonderful about it, it just sucks when people get the math wrong.

[indypaul]

17. Or when one accepts it as their master and not their servant. n/t

[JudyInTheHeartland]

18. It's time to restore science to it's rightful place!

Wait... What?

[BlooInBloo]

21. Despite what idiots say, it was the banks' POLICIES that failed, not the analysis...

EDIT: "warnings about its limitations were largely ignored."

"Li can't be blamed," says Gilkes of CreditSights. After all, he just invented the model. Instead, we should blame the bankers who misinterpreted it. And even then, the 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."


EDITEDIT: It is HILARIOUS watching idiots read these things, and conclude that their idiocy is really the best policy, however. :rofl:

[DLnyc]

23. Could we say that the banks' policies failed when the regulatory system

failed to prevent those policies? True, banks were greedy and deliberately ignored, or even hid, risk in order to leverage more and more steeply. While I don't 'justify' this behavior (in fact, there is very little that banks do, or want to do, that I would justify), the reality is that banks WILL take higher and higher risks in an environment where government chooses not to regulate them. So, while I agree the banks essentially gambled away a large portion of the entire world's assets while enriching their (relatively few) executives, strictly speaking that is what one should always expect to happen when one lets the banks design their own regulatory system.

[spin]

25. Murphy's Law applies to economics....

or maybe Finagle's corollary to Murphy's Law:

Anything that can go wrong, will—at the worst possible moment.

Murphy's Law - ''If anything can go wrong, it will'' - was originally formulated by and for engineers. Alan S. Blinder, professor of economics at Princeton University, offers a new version in his forthcoming book, ''Hard Heads, Soft Hearts: Tough-Minded Economics for a Just Society'' (Addison-Wesley). To wit: ''Economists have the least influence on policy where they know the most and are most agreed; they have the most influence on policy where they know the least and disagree most.'' What the new Murphy's Law reflects, Blinder says, is the systematic tendency for good economics to make bad politics, and vice versa.
http://www.nytimes.com/1987/09/20/magazine/to-start-with-never-never-land.html

[NashVegas]

32. The Problem With All Formulaic, Spreadsheet Strategies

Is that when you take them outside of the controlled laboratory and pit them against a human population, results can be unpredictable.

Just about any online gamer whose worked with these types of strategies can tell you what eventually happens: the winning strategists cheat. In some cases, they cheat the way Bernie Madoff did, by deceiving the population. In others, they buy off congressmen who can enact bailouts.

[indypaul]

36. It's the same story as in 1929

Too many people borrowed money from Peter to pay Paul

That made Peter sore and you cannot do business with a sore peter.