January 10, 2008
Inside the Black Box
The best thing about it only being January 10 is that I can say, without reservation, that this is the best thing I’ve read all year: n+1’s interview with a hedge fund manager. It includes a useful window into a little-known, but super-interesting, component of modern markets: quantitative trading driven by computer programs!
n+1: And so the computers themselves are making these trades?
HFM: You build the models and the computer does the trading. You actually do all the analysis. But itís too many stocks for a human brain to handle, so itís really just guys with a lot of physics and hardcore statistics backgrounds who come up with ideas about models that might lead to excess return and then they test them and then basically all these models get incorporated into a bigger system that trades stocks in an automated way.
n+1: So the computers are running the…
HFM: Yeah, the computer is sending out the orders and doing the trading.
n+1: Itís just a couple steps from that to the computers enslaving —
HFM: Yes, but I for one welcome our computer trading masters.
People actually call it “black box trading,” because sometimes you donít even know why the black box is doing what it’s doing, because the whole idea is that if you could, you should be doing it yourself. But it’s something that’s done on such a big scale, a universe of several thousand stocks, that a human brain canít do it in real time. The problem is that the DNA of a lot of these models is very, very similar, it’s like an ecosystem with no biodiversity because most of the people who do stat-arb can trace their lineage, their intellectual lineage, back to four or five guys who really started the whole black box trading discipline in the ’70s and ’80s.
If you read on from that point in the article you’ll learn about “ten-sigma events” — if that doesn’t sound like something from a dystopian anime series, I don’t know what does.
There’s also some really great discussion — and explication — of the whole sub-prime thing. It’s long, but the conversational style makes it pretty digestible.
(Thanks to PoN for the link.)