http://online.wsj.com/article/SB10001424052748704509704575019032416477138.html

Out of control overpaid autistic computer geeks with unproven theories based on bad reasoning are to blame for the global economic collapse.

Covered in a new book:

http://en.wikipedia.org/wiki/The_Quants

We really need to get a lot of these people into prison, or group homes.

Yup. It definitely wasn't the management/C-level execs pulling down the really big salaries and giving the orders who were to blame.  Damn those pesky programmers / mathematicians!

"Margin Call" is an excellent film that tackles this very topic.

http://en.wikipedia.org/wiki/Margin_call_(film)

You get a clear idea of just how little upper management understood of the day to day trading, and what they did when the penny finally dropped, i.e. "we're so FUCKED, SELL SELL SELL!!"

"It definitely wasn't the management/C-level execs pulling down the really big salaries and giving the orders who were to blame."

I want to believe. Really. But isn't it the quants that decided on the math and implemented the algorithms the management relied upon?

Who is to blame for the Tacoma Narrows Bridge collapse? The engineers who designed it, or the politicians who ordered a bridge to be built? How much role in design did the politicians have? How many lines of code did the CEO of Goldman write in those algorithms?

"But isn't it the quants that decided on the math and implemented the algorithms the management relied upon?"

No, it isn't. The market decides on the maths and traders pass these decisions to quants. Maths is good. Greed is bad.

But quant-traders are the pure evil.

Well i agree it's a gamble, but more in an insurance company's way of thinking than a poker game.

Like, take a quote from http://en.wikipedia.org/wiki/The_Quants : "[quantitative strategies lead to] underestimation of the likelihood of chaotic, volatile moves in the markets".

When an insurance company lists a price for a car (or life) insurance, it's doing an actuarial calculus ( http://en.wikipedia.org/wiki/Actuarial_science ) based on statistic of car damages (or life expectancy). Similarly, a hedge fund company listing financial instruments it's computing the product price based on statistics of market moves.

In an ideal world, the insurance company or the hedge fund would add comfortable margins to these statistics (i.e. consider large volatile moves in the markets). Unfortunately this results in very expensive insurance policies or financial instruments which no one (or very few) would buy. And since the competition it's ferocious, there's a race to the bottom on how little these products can be priced. Which inevitably results in shrinking the domain of probable moves of the market from "very large" to "very small".

@Bluebeard

"and what they did when the penny finally dropped, i.e. "we're so FUCKED, SELL SELL SELL!!""

Which was the right thing to do (as long as you're the first to sell, which they were).

@Idiot

"Out of control overpaid autistic computer geeks with unproven theories"

They weren't theories. They were models. There's a difference between the two.

Anyway, the subprime mortgage debacle has mostly to do with fraud, naive expectations ("house prices can only rise") and pricing based on incomplete, stale data. Bad models (and bad modellers) were also there, but they were firther down the list of the culprits.

@Quant - yes they did what they could. What I found fascinating was how when the fire sale started at about 9 in the morning they were getting good money... by the afternoon they couldn't give the shit away. All in the space of one day....

This sounds a lot like what happened in the mid-80's, when Wall Street got a hold of copies of Lotus 1-2-3

The bosses didn't understand the spreadsheets and thought "Well, the computer says it's ok..."

Takeaway thought: You can misuse tools to get the result you desire.

Bottom line, I don't see the understanding up the food chain that these models are essentially computer programs, and that all programs have bugs. Nobody can think of every possible thing that can happen, and there's lots of assumptions built into these models that don't always work out in the real world (i.e., liquidity always > 0).

And when they fail, they fail catastrophically.

Look, if the odds of creating an economic catastrophe are 10%, but running an algo with that risk pays off 100%+, you're gonna make the bet. 

Can't blame the quants for the rules that allow this kind of bet from happening.

Well one thing we know for sure, computer models of global warming are correct even though, like financial models, they have never been used to successfully predict the future accurately more than a day out. And since computer models of global warming are correct, we must set up carbon trading exchanges and let Goldman Sachs and Wall Street take a fee off each transaction.

"Well one thing we know for sure, computer models of global warming are correct even though, like financial models, they have never been used to successfully predict the future accurately more than a day out. "

This wrong on so many levels that I don't know where to start debunking.

What will APPL be at tomorrow at 11AM EST? Please answer before 11AM EST today. Note I am not even asking for info about the whole market, just a single point to see if you can hit it.

"Bad models (and bad modellers) were also there, but they were firther down the list of the culprits."

Bad models are harmless. You spot fast when something is wrong, and it's usually when YOU are losing. And the remark about neglecting volatility in pricing models made me smile... People have as much idea about quant-fin as they have about autism.

>Out of control overpaid autistic computer geeks with unproven theories based on bad reasoning are to blame for the global economic collapse.

In our accounting class, we got to see the 60 Minutes interview with Kozlowski in prison. That unrepentant sociopath kept blaming the little people underneath him for the causes of his criminal misdeeds, just like the above quote.

Financial models are not intended to "predict the future accurately".

"What will APPL be at tomorrow at 11AM EST? Please answer before 11AM EST today. "

Like s7h_black said, quant models are not to predict the future. They are to put a price on your predictions. How you arrive at them, is a different story.

Did not know that. Thanks.

Volatility is messy.

The original geeks were Black & Scholes, who got the Nobel Prize for the CAPM (capital asset pricing models).
http://en.wikipedia.org/wiki/Capital_asset_pricing_model

They argued that you could tell how volatile an assets price had been historically, contrasted with the rest of the market.

They called that Beta, and got really famous by telling everyone just how clever they were.
http://en.wikipedia.org/wiki/Beta_(finance)

Then they started to believe their rubbish, and founded a firm to make tons of money. It was called LTCM (Long Term Capital Management). Things did not end well for them. Not pretty at all.
http://en.wikipedia.org/wiki/Long-Term_Capital_Management

Turns out that predicting future volatility, just by looking at historical volatility is not too helpful. A bit like driving by looking in the rear view mirror.

Now, a bunch of guys decided to get rid of the "This-is-not-a-casino" pretenses, and came up with something called Monte Carlo simulations to price risk, and by extension financial assets. But that's a story for another day.

Bottom line is that you can use computers to find inefficiencies and other arbitrage opportunities. Trying to predict the market though, using computers, is doomed to fail.

This is in no small part due to the fact that the market is 'human' after all, and very often displays irrational (and herd) behavior that few models, quantitative or otherwise, can predict.

"Trying to predict the market though, using computers, is doomed to fail."

I wonder if the same argument could be made about climate change, given the multiplicity of variables and complexity of the systems...

No, climate change is real. It's changing right now, today's temperature is different from yesterdays.

Climate change is real, as per Idiot

Climate prediction is bunk

Global warming is a religion.

"The original geeks were Black & Scholes, who got the Nobel Prize for the CAPM (capital asset pricing models)."

Black didn't get the Nobel Prize -- he wasn't an academic (worked in GS) so they waited with awarding Merton and Scholes until he died. And the model they got the prize for was the Black-Scholes model, not CAPM.

"They argued that you could tell how volatile an assets price had been historically, contrasted with the rest of the market."

OK, maybe you meant CAPM, because in the Black Scholes model you don't use historical data.

"They called that Beta, and got really famous by telling everyone just how clever they were."

Yes, you are talking about CAPM.

"Then they started to believe their rubbish, and founded a firm to make tons of money. It was called LTCM (Long Term Capital Management). Things did not end well for them. Not pretty at all."

OK, now you're talking about the Black-Scholes model. BTW, it wasn't volatility which killed LTCM -- it was correlation, which blew up due to the Russian crisis. Ends bad for everyone.


"Now, a bunch of guys decided to get rid of the "This-is-not-a-casino" pretenses, and came up with something called Monte Carlo simulations to price risk, and by extension financial assets. But that's a story for another day."

You need to use the Monte Carlo method of integrating stochastic equations when you introduce some extra degrees of complexity to the Black-Scholes model. The main idea remains unchanged.

"Trying to predict the market though, using computers, is doomed to fail. This is in no small part due to the fact that the market is 'human' after all, and very often displays irrational (and herd) behavior that few models, quantitative or otherwise, can predict."

This actually isn't the main problem here. Market's self-consistency is the problem -- you prod it, and it bites back. But the Black-Scholes model is not for predicting the market anyway.

The main problem is that human beings have free will, and free will resists modeling with equations.

> I wonder if the same argument could be made about climate change, given the multiplicity of variables and complexity of the systems...

Although climate is very complex, I don't think that an evil mind is playing the system (religious people may disagree).

Markets however are played by the most evil species known in the universe, and that will actively destroy your predictions over time.

Actually, according to Taleb, Black and Scholes stole their theory from a blackjack guy named Thorn, I believe.

"Markets however are played by the most evil species known in the universe"

I wish it was true.

"and that will actively destroy your predictions over time."

They are pretty easy to predict -- they play to win.

"free will resists modeling with equations"

There are billions of times more atoms in the rest of the physical environment than in human brains.

You say we can't model the complex process of neurons firing in a human brain to predict what he will do, but you think one can predict a system that is trillions of times more complex.

How to delete this topic: http://crazyontap.com/topic.php?TopicId=151038?

"but you think one can predict a system that is trillions of times more complex."

Who says that?

"I wonder if the same argument could be made about climate change, given the multiplicity of variables and complexity of the systems..."

You're confusing modelling of the long-term trends in the global averages with short-term local weather forecasting.

We're engineers.  We can predict anything.  We work with simplified models of reality every day.

Doesn't mean we're always right.  But it's better than choas.

Sure, STH, and there were no Challenger, Fukushima, Deepwater Horizon, Hyatt Regency hotel, etc.

black, sure but that's because of this that and the other thing. If this that and the other thing, parts that weren't in the model, had never happened, then everything would be OK.

You see, the model is good, it's reality that has failed.

Global warming is true. The models don't match reality because reality is not behaving correctly.

The Shuttle launched over 100 times.  One launch failed because the weather was too cold -- and the Engineer SAID the weather was too cold, but were over-ruled by politicians.

One landing failed because the wing insulation was cracked on take-off.  The engineers said "take a picture of it", but were over-ruled.

I don't think you understand the purpose and reliability of "simplified models".  Apparently you think you can deal directly with reality, without realizing that you don't deal directly with reality.  The amount of reality that makes it in through the filter of your eyes and ears and brain is trivial compared to the REAL reality.

@Idiot

"Global warming is true. The models don't match reality"

They do.

@STH

"I don't think you understand the purpose and reliability of "simplified models""

Wow, aren't we hoity-toity today?

I was talking to S7.  Do you disagree with my point?

"The Shuttle launched over 100 times. One launch failed because the weather was too cold -- and the Engineer SAID the weather was too cold, but were over-ruled by politicians.

One landing failed because the wing insulation was cracked on
take-off. The engineers said "take a picture of it", but were
over-ruled."

Exactly the same thing can be said about the market. Millions of
transactions are being made every day, from global companies to
South-American farmers -- they successfully protect themselves from bankruptcy by buying financial instruments, new and more advanced companies grow thanks to the market investors' money in place of those obsolete or less efficient ones -- we can live better and safer now and in the future, have green energy, water treatment or cancer medicines, whole countries develop, etc. It happens a few times in our lifetime that this mechanism fails and that's what people notice. That's why your comparison is flawed. You want to compare something to your engineer's "simplified models", compare the Black-Scholes theory. If you want to set something against the market crashes, talk about
"the reality" icluding catastrophes caused by the above failures. Local "engineers" also can tell that it is too hot, but nobody listens (by definition).


"not chaos"? If you are an engineer, you should know that you have chaos in a mere LED and a semiconductor cable.


"I don't think you understand the purpose and reliability of
"simplified models"."

So what is that purpose and reliability of "simplified models" that you think I don't understand?
The "reliability" of "simplified models" lies in statistical tests and restrictive norms. Considering how little regulations
and legal responsibility there are in financial markets (thank your American neolibs), I'm surprised that the crashes are not much more often and severe.


"Apparently you think you can deal directly with reality, without realizing that you don't deal directly with reality.  The amount of reality that makes it in through the filter of your eyes and ears and brain is trivial compared to the REAL reality."

I'm not sure which part of my post ignited this philosophical tirade or made you think that you can read my thoughts. Im not a philosopher so I don't discuss my philosophy, but indeed, I do not follow the Kantian transcendental idealism.