How a swashbuckling breed of computer scientists nearly destroyed Wall Street | Hacker News Reader

How a swashbuckling breed of computer scientists nearly destroyed Wall Street

online.wsj.com

23 pointsjakarta21 comments

Oh, sure, blame the guy that made you the billions in the first place for not knowing that you were fucking around with the underpinning factors. Real mature, Wall Street. Real mature.

There's enough blame in this situation to cover everyone involved.

On the level of modeling, the fundamental flaw was assuming that stock prices could be equated to independent, normalized, random variables. When these assumptions break down, you can still "make money most of the time" but you're downside becomes unlimited. Ponzi schemes make money most of the time too...

But on the level of the banks, investment houses and large investors, the problem was not realizing that an investment model is a leaky abstraction. You can't leave your investing to people who only see random variables. Prudence, the watch-word of traditional investing, encompasses more than probabilistic expectation (just as causation is more than correlation). Leaving fundamental decisions about investing to the quants is imprudent. Anyone not blinded by the shine of the new models realized this but that was a minority.

Of course, along with quant-investing in particular, one must include the whole Alan Greenspan school of "Market Rationality", including the move to "fair value" accounting. And that's a big elephant.

So lets see. A bunch of people bought stock. Stock rises and a bunch of people get rich. Somebody sells stock. A bunch of people get nervous and sell their stock, sending prices plummeting.

This story is as old as Wall Street itself. That quants were involved was incidental, thought it makes for some interesting twists.

Regular people overvalue houses, with the encouragement of the government. Banks cash in and some buy into the hype. Who to blame? I know, blame those geeks who crunched some numbers.

But whatever you do, don't blame the women who caused this problem in the first place:

http://www.youtube.com/watch?v=Ubsd-tWYmZw

Regular people are incentivized to pay as much in interest on their homes through tax deductions. With the tax deductions, paying an interest-only mortgage become MUCH less expensive than rent, so "buying" a house and planning to flip it after the interest-only period is up was an economically sound decision (and a lot of people still stand to gain.) This did drive up the price of housing as more people were entering home ownership.

No, tax deductions simply make home ownership competitive with rental companies, who can also deduct their interest expense.

My god, is that a real commercial or a parody? The look of sheer contempt on her face, as if she's about to throttle her soft, spineless husband...I'm speechless. Why would anyone want to be like either of those people?

Sure, let's blame the computer scientists, not the dangerous Wall Street culture of getting rich quick with essentially regulated gambling. That'll fix things ;)

That is a general tendency of blaming whatever cannot be explained or understood.

The primitive men blamed natural phenomena and diseases on the gods and witches. Today we blame it on the nerdy guys with glasses and pocket protectors.

When they make money, the financial giants quickly take credit for it: "oh look we are making money, _we_ are so awesome", when the nerdy guys screw up, then they are scapegoated, "oh look, it is those damn nerds and their voodoo, that screwed us."

Some of the "nerdy guys" actually did screw up. They deserve some of the blame. Hopefully in the future quants will be more careful about checking the assumption of their models, especially if it may have repercussions on their reputations and careers. That doesn't mean you don't also blame the others that were involved. There is more than enough blame to go around.

When the nerds with their voodoo were making lots of money, the firms didn't say "we don't know what these guys are doing, but they are making us lots of money, they deserve all the credit, they are the heroes", instead the CEOs gave each other big bonuses and pats on the back. But when everything went sour, they blamed the "nerds" (and I admit, nerds did screw up), instead of blaming themselves for not evaluating risk properly. If they couldn't evaluate risk properly (and I think they couldn't), they should have stopped what they were doing.

EDIT: syntax errors

Yes, some of them did screw up, the same way that early physicists screwed up. Quantitative finance, as we know it, is still a very young field. There are going to be hiccups along the way. Knowledge is not developed in a vacuum.

We had Archimedes looking at physics over 2000 years ago. It took until the late 17th century for Newton to develop the calculus, a tool that allowed us to really start advancing physics. Whether you want to start "quantitative finance" with Black and Scholes, Markowitz, or even Bachelier, we have had at most a little over 100 years of trying to pin this beast down. I think people fail to realize that.

The problem isn't the quants, it's the people that doesn't understand what the quants do and that a model is only a model.

This is much like saying that Smith & Wesson shoots people.

No, it's like saying the architect who built the house that collapsed in an earth quake must take some blame, not just the company that funded the building.

And I agree with that. The models were brittle and the math/CS people who created them will have to take responsibility for that. Obviously there are others who must share that responsibility.

The decision to massively dump positions was a human one. This is just another example of what happens when you get a bunch of independent agents with imperfect information scared -- they cut their losses. In a market, this triggers a spiral effect of lowering prices, encouraging more people to cut their losses. Wether this is automated or manual, this is not a new thing.

I dunno if it's just me or where they not clear as to what they meant by complex systems. Where they talking about just complicated things or more like: http://en.wikipedia.org/wiki/Complex_systems. I know it's not a really important point of the article but I can see how this type of stuff would make no sense to everyone but people who have knowledge of what complex systems are. Basically I can see how the management of these companies didn't want to interfere with what they didn't understand but seemed to work. But now it's easy to say that stuff was all crap and cause us billions. I hope that some of the science these guys were working on can be salvaged (the good parts at least).

Couldn't we model business cycles and also economic downturns / recessions also? Attempt to quantify bubbles if they possibly exist? Or would that be too computationally hard?

How would you model something like 9/11? There are too many random events that occur every trading day. Some are good. Some are bad. The best we can really hope for is that we notice things are going to hell before everyone else does. That allows us to get out early and save our own ass. That's the truth about the markets. The models we have help us to understand things during relatively normal times, but they are just models. You can't rely on them 100% of the time. At some point, you have to ask yourself, "Does this make sense here?"

I don't get it , don't most high frequency trading shops not hold overnight positions?

Not all quant work deals with high-frequency trading.

Reply on news.ycombinator.com