Skip to content →

Tail risk, the police, and grad school

This past Saturday morning I walked past the Loews movie house on 68th and Broadway and saw, stationed outside, a NYC police car with sirens flashing, and standing in front of it, with his back to the car and with a holstered gun, a cop, his eye on the movie house … which was, I then realized, showing The Dark Knight.

My first instinct was: Isn’t this ridiculous? Someone crazy tragically shoots up one movie house and now they’re going to guard all movie houses showing that movie. This isn’t like political attacks on synagogues on Yom Kippur, which after September 11 led to police guarding synagogues.

A little later I saw the same situation at the Loews on 83rd Street.

But I’ve changed my mind. What the police were doing was practicing tail risk elimination.

At Goldman 12 years ago, the risk committee didn’t rely exclusively on VaR and poor probability analysis to estimate their risk. Though they estimated their VaR, they also considered disaster scenarios that were dangerous, and then took positions that ensured that their losses wouldn’t be unbearable if those events (a repeat of the Russian default crisis, a repeat of the ’87 crash, etc) occurred, no matter what their probabilities.

And that’s what the police were doing. Irrespective of the small odds, they were trying to cover themselves for a repeat. And that’s the best their imagination could do.


On a related note, I myself thought, and several people commented to me too, that it was significant that the man in Aurora had been struggling, apparently failing graduate student. The life of PhD students can be awful, and if you’re already severely disturbed, doing badly in grad school can be the last straw. I mentioned some examples of this in My Life as a Quant.

Published in blog Uncategorized