Join 3,375 readers in helping fund MetaFilter (Hide)


The Minds Behind the Meltdown
February 23, 2012 2:51 AM   Subscribe

How a swashbuckling breed of mathematicians and computer scientists nearly destroyed Wall Street.
posted by veedubya (48 comments total) 18 users marked this as a favorite

 
How is this just now a FPP? I thought for sure this one would have been beaten to death. Kinda Previously
posted by AndrewKemendo at 3:09 AM on February 23, 2012


I think I first heard the word "quant" during the Enron scandal.

At no point have I ever heard the word in a positive context; it's always been "here's what the quants helped the companies everybody has good reasons to hate do"
posted by Pope Guilty at 3:19 AM on February 23, 2012


At no point have I ever heard the word in a positive context;

"Quant" is regularly used in the sciences to refer to anything or anybody mathematical, and does not have a negative connotation. Well, except for in conjunction with something like "That quant guy is going to give a talk. I'll be trying to stay awake," or "The quant guy tells me I can't use this statistical test in my paper because it doesn't make sense. They spoil everything."

/bitter quant guy
posted by Philosopher Dirtbike at 3:34 AM on February 23, 2012 [3 favorites]


"Quant" is short form in my eyes for auto-regressive techniques that don't reveal any new information. Somehow I doubt Wall Street is that lazy.
posted by Yowser at 3:37 AM on February 23, 2012


How is this just now a FPP?

Because it's not by Matt Taibbi?
posted by chavenet at 3:46 AM on February 23, 2012 [2 favorites]


Is there more after And there was nothing he could do to stop it. Or so it seemed? The article ends for me there.
posted by vanar sena at 3:55 AM on February 23, 2012


> Is there more after And there was nothing he could do to stop it. Or so it seemed? The article ends for me there.

It's a book excerpt, which is why it ends there. I suppose you'll have to buy the book to get the rest of the story.
posted by needled at 4:02 AM on February 23, 2012


"Quant" is short form in my eyes for auto-regressive techniques that don't reveal any new information. Somehow I doubt Wall Street is that lazy.

Really? Why? As this article by Felix Salmon points out, a lot of the problems in the housing markets were down to bond traders relying on one single formula, a measure of correlation called the Gaussian copula function. The whole Street thought it had found a new way to price the risk of mortgage backed securities, everybody started using it, and nobody remembered that the fact that the pricing model wasn't perfect, and that there's more things in the mortgage market, Horatio, that dreamt of by the quants...

Speaking as a English major who barely passed calculus, a lot of what the quants do is over my head. But one thing is clear to me --- a lot of what they do is based on analogies, at the end of the day. The idea that formulas and techniques which have successfully been used to model phenomena in the natural world can usefully be used to model phenomena in the markets. But nature's gonna do what it's gonna do, and when what actually happens differs from your prediction you know your model's wrong and you start over. The way the quants trade changes the way the markets move. In fact, there's so many of them in there now that a lot of the time they're just trading with each other, doing things like HFT that humans can't, and there's a legit question as to whether they are tearing at the fundamental price discovery mechanism of the market....
posted by Diablevert at 4:06 AM on February 23, 2012 [9 favorites]


Everyone's got a system, but the house always wins.
posted by gjc at 4:07 AM on February 23, 2012 [5 favorites]


The problem is that wall street applied a genetic algorithm approach and the selection criteria was "Which algorithm will let me steal the most money?"
posted by srboisvert at 4:17 AM on February 23, 2012 [13 favorites]


the problem is all these people got fired from NASA. or that there never was a NASA to hire them.

where's our jetpacks, our cheap solar power, our bullet trains? over in China because the United States started drinking its own IMF brand of kool-aid.

to wit: "fundamental price discovery mechanism"
posted by eustatic at 4:56 AM on February 23, 2012 [4 favorites]


The Wall Street Journal passes the buck from the institutions to the foot soldiers -- a swashbuckling breed of mathematicians and computer scientists and how those mathematicians and scientists then nearly destroyed Wall Street.

A group of intellectual individuals so smart that they used tools beyond those managing them:
brain-twisting math and superpowered computers to pluck billions in fleeting dollars out of the market.

(This is more poetic than most WSJ articles)

And then, not only are the tools beyond the control of managers, the market takes on a life of its own!
the selling had taken on a life of its own, leading to billions in losses.

The normally steely-eyed Captains of Wall Street were overcome with bible-like emotions; they were incapable of making decisions...
It was utter chaos driven by pure fear. Nothing like it had ever been seen before. This wasn't supposed to happen!

And whilst it was apparent what was going on, somehow the professionals whose job it is to know these kinds of things completely missed (ignored?) the signs:
Everyday investors had no insight into the carnage taking place beneath the surface, the billions in hedge fund money evaporating. Of course, there was plenty of evidence that something was seriously amiss.

So basically, the WSJ is telling the story of a few financial kings that hired alchemists, and told them to make lead out of gold. When the alchemists had done that -- with the side effect of destabilising the rest of the world -- the kings threw up their hands and wrenched. "How could we have known? How could we have known?"

Nice try. Someone still gave all those quants jobs, and celebrated quite a few outsized returns. Try again.
posted by nickrussell at 5:02 AM on February 23, 2012 [53 favorites]


"Blame the nerds"? Never saw that one coming. Of course the greedy alpha male types running Wall Street are blameless -- they're just so dumb and naive, you see, that they got hoodwinked by those goddamn nerds into making billions of dollars.
posted by indubitable at 5:15 AM on February 23, 2012 [21 favorites]


How a swashbuckling breed of mathematicians [Part 1] and computer scientists nearly destroyed Wall Street. [Part 2]
posted by Fizz at 5:17 AM on February 23, 2012 [3 favorites]


Am I the only one who thought of this scene after reading this post title?
posted by Kevtaro at 5:20 AM on February 23, 2012


Guess not.
posted by Kevtaro at 5:21 AM on February 23, 2012 [2 favorites]


Of course the greedy alpha male types running Wall Street are blameless -- they're just so dumb and naive, you see, that they got hoodwinked by those goddamn nerds into making billions of dollars.

I do love the "I am not evil; I just can't do my job" excuse. Let's be honest; is that really less of a reason to, say, drown you?
posted by GenjiandProust at 5:30 AM on February 23, 2012 [6 favorites]


At no point have I ever heard the word in a positive context; it's always been "here's what the quants helped the companies everybody has good reasons to hate do"

I'd say that's American society's bog-standard anti-intellectualism. None of this, not even Enron, happened because some people were too smart or knew too much math.
posted by Copronymus at 5:48 AM on February 23, 2012 [7 favorites]


The rule that needs to be implemented and is not even on the table is a minimum time to live for all exchange orders. Until that is put in place, the markets will be at risk of blowing up due to HFT.
posted by Crash at 6:00 AM on February 23, 2012 [1 favorite]


Eh. This is a fundamental misundertanding a bit here I think. In most cases the quant finance guys who were good at their job were saying "Hey this works, but its predicated on x,y, and, z being as true in the future as they were in the past. There is no reason to think that won't be the case, but things do change, and our business itself might be what causes it to change" and the non-quant guys they worked for said "Yeah but it works for now right? Let's sell the shit out of it". And the Quant guys who wanted to stay employed said "OK" and the ones who got fired pretty quickly kept trying to talk people out of using them.


Like look at the Gaussian Copula stuff in the Felix Salmon article (also discussed previously). It was an actuarial technique designed to examine couples habits of dying around the same time. Some quant gobbled it up and used it make CDOs work.

The rule that needs to be implemented and is not even on the table is a minimum time to live for all exchange orders. Until that is put in place, the markets will be at risk of blowing up due to HFT.

Ugh. The only people HFT harms are other speculating short-term traders. I have a very simple solution for the rest of you. Use limit orders.
posted by JPD at 6:03 AM on February 23, 2012 [3 favorites]


BTW - tho this book isn't about RMBS and CDO's its about a class of quant driven equity traders that blew up back in '07 in one of the first waves of the big collapses to come. It wasn't actually that damaging to the market as a whole other than creating some excess volatility for a little while.
posted by JPD at 6:05 AM on February 23, 2012


JPD - A TTL rule wouldn't impact anyone but the top echelon of HFT. What's your issue with the rule? I'm talking 50 - 100ms, not 5 minutes or something crazy.

Without the rule, you can place all the limit orders you want, but if the market is tied up due to an influx of orders which are immediately cancelled, you're not getting your order on the books. Another way to fix this is to ban quote stuffing, but the TTL basically does the same thing.
posted by Crash at 6:20 AM on February 23, 2012


I don't think you can blame mathematicians for the financial crisis any more than you can blame marie curie for hiroshima.
posted by empath at 6:29 AM on February 23, 2012 [2 favorites]


I have a very simple solution for the rest of you. Use limit orders.

I prefer the financial transaction tax.
posted by Talez at 6:39 AM on February 23, 2012 [8 favorites]


50-100ms isn't "the top echelon," its every HF firm. Its a rule that does nothing but protect very active human traders, and I personally don't see why they should be protected.

Its a waste of regulator time and political capital. Not to mention HFT is going away on its own. The numbers I've been quoted in terms of profitability over time is really striking. My guess is that the marginal player in the US markets is barely profitable today - and most players are marginal.

I'd argue Marie Curie is abrogating too much responsibility - its closer to the guys who worked at Los Alamos.
posted by JPD at 6:39 AM on February 23, 2012


I don't think you can blame mathematicians for the financial crisis any more than you can blame marie curie for hiroshima.

The quants aren't Curie. They are Los Alamos.

Look --- Based on the excerpt, the book looks a little breathless and inclined to over-attribute respto the quants. I too believe that the heads of these banks deserve more scrutiny and more blame for the ultimate consequences.

But there's a lot of blithe dismissal going on in this thread that I don't think is warranted.

Google's working on autopilot for cars, right? It's not too difficult to imagine a future in which cars basically drive themselves, and because computers can react a lot faster than humans, as that technology gets better and better it will enable traffic to go faster and for more cars to be on the road at once. And mostly, this will be awesome, because travel times will be quicker and there will e fewer accidents. But there will be a catch --- after a while, it could turn out that you can't have human controlled cars on the road, because they're too slow to keep up with the rest of the traffic. And therefore when something does go wrong, some unpredictable thing happens, there won't be a way for humans to fix it.

That, to a very rough approximation, is what the quants are doing to the Street. It's autopilot for trades. And the people who called them in and said "build me an autopilot for trading" deserve blame, but that doesn't mean that the possibility of the system being taken over by autopilots that humans can't quite control isn't a real problem.
posted by Diablevert at 6:51 AM on February 23, 2012 [6 favorites]


Jinx, jpd. I swear I previewed.
posted by Diablevert at 6:53 AM on February 23, 2012


Victor Frankenstein denies responsibility; blames monster for murders. News at 11.
posted by qxntpqbbbqxl at 7:06 AM on February 23, 2012


I just want to point out again that the cohort of quant traders this book focuses on actually had very little to do with the GFC. Only in a peripheral sense were they involved.
posted by JPD at 7:08 AM on February 23, 2012


1) quant strats are usually implemented as very risk-averse high sharpe strategies. Not at all how the rest of the street operates. Lets just brush aside the fact that most of the money lost on wall street was from carrying inventory on shitty mortgages.

2) Retail investors want to have ETF's and ETN's matching everything from the S&P500 to the out-month vol on the market, with leverage. These are some of the most popular investment vehicles out there now, and are only made possible with quant strats behind them. Are the consumers of these products responsible as well for 'destroying wall street'? We frequently see similar arguments about consumers being responsible for voting with their dollars.

3) Some of these strats are run as prop desks for banks, but by far the vast majority of the dollars in quant strategies are run by hedge funds. You know where hedge funds get their money? Pension funds are a huge source of funds, and pension funds in particular are attracted to the generally high-capacity low vol profiles of quant strategies. Are pension funds now evil as well? Thats right, your teachers-union pension fund is trying to meet its obligations by investing some of its money in eeeevil quant strategies.
posted by H. Roark at 7:10 AM on February 23, 2012 [1 favorite]


H.R - I think people just didn't RTFA - I'll admit I didn't when I made my first comment, and hell I've read the book the article is about.

(But I think the central permise of the book is that "high-sharpes" weren't actually so high)
posted by JPD at 7:14 AM on February 23, 2012


JPD, I still disagree. HFT can flood the market with quotes entered for the purpose of pinging for liquidity or tricking other algorithms to reveal intentions. Why should the market have to deal with any quotes entered for any other purpose besides buying and selling equities?

These actions have costs the entire market pays for, although miniscule individually. The only limit right now is set by SIAC, and each time the SIAC raises the capacity limitation for CQS, quote rates surge to fill it. In the case of market stress, a lack of spare capacity can lead to a drop in liquidity due to lack of clear pricing info.

If you think this is too high a threshold, make it 5 or even 1ms. The point is that without a floor which allows orders to potentially be filled, there's nothing to stop HFT from attempting to manipulate the market using these techniques.
posted by Crash at 8:15 AM on February 23, 2012


The Wall Street Journal: Breaking news of 2008, TODAY!
posted by KingEdRa at 8:31 AM on February 23, 2012


MD#1: Today's WSJ print edition is a great way to test your retention of what you should've read y'day on Bloomberg. A#1: [SILENCE]

- Twitter @gselevator, Jan 13, 2012
posted by Mister Fabulous at 8:40 AM on February 23, 2012


Yeah, sounds like a lot of blaming the scientists for the excesses and bombasticity of the traders and managers... Wall Street had a toxic culture that led directly to the crash; this reads like buck passing of the worst sort.
posted by kaibutsu at 8:50 AM on February 23, 2012


Approaching the singularity -- in global finance - "the model predicts a sharp increase in the size of market fluctuations when entering the machine dominated phase below one second... This kind of analysis really should have some bearing on the consideration of potential new regulations on HFT."
posted by kliuless at 8:55 AM on February 23, 2012


Call them enablers, then. Like the chemists who turn coca leaves into crack.
posted by IndigoJones at 8:55 AM on February 23, 2012 [1 favorite]


If only those damned chemists weren't providing us with such an incredible business model, we would be able to stop selling crack to all of those children...
posted by kaibutsu at 9:09 AM on February 23, 2012


Yeah, sounds like a lot of blaming the scientists for the excesses and bombasticity of the traders and managers...

You don't understand how quant strategies work. (of course, neither do I, but hey, I've got lots of opinions and this is the Internet). A quant strategy is a code. An "if...then" statement. They are often executed by computers without human intervention. You say "traders do x" but with a lot of these techniques, traders don't do jack. They decide on a general strategy and the computer executes it, much more quickly than a human could. There are still people making trades out there, sure. But a lot of the time when something decides "sell Coke stock" it's not because some human saw a headline that said, "coke profits drop." it's because somebody wrote some code that said "when the price of commodities x,y, and q exhibit x% price change over y period of time, execute trades h,g, and w." the trades are executed in milleseconds, literally quicker than you can click a mouse. This is how you end up with things like the flash crash. Autopilot crashing the cruise liner.
posted by Diablevert at 9:42 AM on February 23, 2012


Diablevert,

You're describing Algorithmic Trading, which is the basis behind automated trading. As several have pointed out here, this was not the only way in which "quant" analysis was used, and may not even have been central to the crisis. More important was the use of predictive risk models, particularly in application to complex derivatives. The main problem was managers using models they did not understand to evaluate securities they did not understand, with their main criterion for selecting a model being outcome=greatest rate of return.

I see this a lot in my own industry (advertising), where people with poor critical thinking skills take (much simpler) types of analysis and attempt to turn them into a one-size fits all solution, or a way to validate decisions that are actually based on who took them out to the best lunch, or who gave the best gifts at Christmas.

When all you have is a hammer, everything starts to look like snails.
posted by TheWhiteSkull at 10:33 AM on February 23, 2012 [3 favorites]


If only those damned chemists...


Oh please. You choose who you do business with. If some of the nasty bits rub off on you, well, you could have figured on that before you signed up.
posted by IndigoJones at 12:06 PM on February 23, 2012


I'm intrigued, I just wish there was a book available where I could find out more.
posted by mattoxic at 1:38 PM on February 23, 2012


As several have pointed out here, this was not the only way in which "quant" analysis was used, and may not even have been central to the crisis. More important was the use of predictive risk models, particularly in application to complex derivatives.

Yes, I know; that was me up the thread linking to Felix Salmon's explication of the Gaussian copula in Wired. Look, personally, the closest I've got to the quants is asking Andrew Lo a couple follow up questions after he gave a speech at a lunch I attended. But I am aware that the term "quant" encompasses a vast array of mathematicians and physicists and their various roles on the Street. And while I def think there are arguments to be made about the apportioning of Ultimate Responsibility for the crisis, for myself I've been a bit frustrated by this thread --- it seems to me to have an awful lot of knee-jerk Must Defend the Geeks and not an awful lot of understanding of how the mathematisation of trading, in all its forms, really has changed the financial markets. I wish more people were willing to engage with this because I don't think its something that can be safely left in the hands of those it serves. What I've been trying to do is break it down to something that might connect with people, and I think I'm obviously failing at that plus babysitting the thread, so I'll bow out.
posted by Diablevert at 1:40 PM on February 23, 2012


Ahahaha they're blaming the quants! Well then, if it's not an invisible omnipotent entity benign will of raping your finances or everyone's greed fault (hence really no one's fault) ...uhm... let's blame the most incomprehensible beings with an human semblance, the quants!

Some quants are the contemporary, geeky and computerized incarnation of useful idiots, inasmuch they believe they're following the rules of "the market" and that "the market" is the best possible thing on the world, unwittingly Panglossing all over themselves, for they possibily didn't read Candide and tought history was a boring subject a school (it is, if it's not taught with a passion).

Others, like people at ACompanyThatSinceTheGreatDepressionKeptOnFucking You (hint: G.S.), know exactly their rather sophisticated "risk immunization strategies" are CRAP, but as their frail ego can't stand being judged by others, they seek comfort in the illusion of their supreme knowledge, and the bits of money that's thrown at them (and sometimes not even money! cars, luxury travels, whatever enables them to keep their ego afloat) is a pittance, when compared to people like Madoff got, without getting caught.

That' sad, for many quants, at the least some I got to know, are brilliant, energic, intelligent people; their talents turned to analyzing something useful would yeld great things.
posted by elpapacito at 2:10 PM on February 23, 2012


Ironically the guys at GS who ran a strategy like those described in the book in TFA got carried out in the '07 crisis the book focuses on. Google "Goldman Global Alpha"

Like the story here as nothing to do with HFT, the Gaussian Copula or Algorithmic Trading.
posted by JPD at 2:52 PM on February 23, 2012


JPD, I still disagree. HFT can flood the market with quotes entered for the purpose of pinging for liquidity or tricking other algorithms to reveal intentions. Why should the market have to deal with any quotes entered for any other purpose besides buying and selling equities?

I don't understand why this matters? Why do you care so much about price discovery? The market is never rational at any one second in time. This is just another irrationality out there. As long as you decide what price you want to buy and sell your shares at, then none of this matters.

The only compelling argument I've seen for the evils of HFT trading has to do with them putting the market makers out of the business, but then not being there to step in and provide liquidity in times of stress. Well guess what - the market makers where thieves too. The HFT are better thieves. You just need to figure out what the cost of a lack of liquidity at moments of extreme stress is. Me personally? I view liquidity or lack thereof as another irrationality to profit from.
posted by JPD at 2:57 PM on February 23, 2012


the WSJ is telling the story of a few financial kings that hired alchemists, and told them to make lead out of gold. When the alchemists had done that -- with the side effect of destabilising the rest of the world -- the kings threw up their hands and wrenched. "How could we have known? How could we have known?"

I love this metaphor so, so much.

The process of turning lead, or shit, into gold, if successful, will only serve to make gold as common, and therefore as valuable, as lead, or shit. The catastrophe is built right into the goals of the endeavor.
posted by gauche at 4:49 PM on February 23, 2012


The rule that needs to be implemented and is not even on the table is a minimum time to live for all exchange orders. Until that is put in place, the markets will be at risk of blowing up due to HFT.

Oh it's worse than that. The most appalling financial news I ever heard was about high speed trading. A project was started to build a new fiber optic undersea cable for data transmission, IIRC it was between Europe and New York. Of course there are tons of undersea fiber data cables. This one would be unique in that it was highly optimized for the shortest cable length. By cutting a little slack out of the line, by straightening out some runs around obstacles, they figured they could shorten the cable length enough that any signal sent through their cable would consistently arrive a few nanoseconds before a signal from any other fiber route. This would give high frequency traders an advantage over their competitors using the other "slower" fiber data routes. The company figured this speed advantage could easily recoup the billion dollar budget to build the new fiber.

Now this is goddam sickening. When traders building trading methods that are trying to beat the laws of physics, trying to gain a trading advantage by going faster than the speed of light, the entire market model is fucked.

But regulations on trading isn't sufficient. You have to regulate the traders too. Kick the crooks out of the market.
posted by charlie don't surf at 6:06 PM on February 23, 2012


« Older Grounded....  |  "The legislative goal of HB116... Newer »


This thread has been archived and is closed to new comments