Wednesday, September 17, 2008

Stat Arb Cleaning Up

Lots of independent corroboration that stat arb is making a lot of money in this tumult. Stat arb is basically trading based on the theme that stocks mean-revert over short horizons, and is based on statistical patterns. When applied to hundreds of stocks simultaneously, it has very low volatility. DE Shaw, Morgan Stanley, and Renaissance, are big players.

Anyway, for everyone selling indiscriminately in panic, their are lots of millionaires in hedge funds saying, 'thanks!' Note the lack of any large hedge fund blowing up in this mess.

7 comments:

Chad Patel said...

Hi Eric,
Just today I spoke to a few quant prop trader friends of mine that trade Stat Arb and they mentioned the same thing. Are you seeing this the same way or are there numbers you saw someplace?
Thanks.
ps, good blog.
Chad

Anonymous said...

oh please. you know better than this. if paulson hadnt dropped the bomb the same stat arb forced selling woudav smashed. this post oughta be removed.

Eric Falkenstein said...

uh, stat arb is delta neutral. If you have an example of a stat arb strategy that is/was long beta, I'm all ears. I've seen lots of strat arb strategies, and looked at their daily, intraday data, backtests--no delta. There is good and bad stat arb, but having a shadow long position has never been an issue.

Anonymous said...

delta neutral is a myth concocted by ppl who trade OPM and keep theirs in treasuries, just like 'intrinsic value' is a myth concocted by sellers for the buy-and-hold team. whatever it is though, the post sounded to me like stat arb is a smart thing, therefore my suggestion.

Anonymous said...

Eric, is stat arb the same thing that Andrew Lo has been lecturing on about "What happened to the quants in August 2007?" when they all had 7-std dev loss months?
'Cause Sept 08 feels a lot like August 07....

Anonymous said...

Doesn't stat arb just boil down to a liquidity play? Shouldn't stat arb strategies get paid for providing it in a time of crisis?

Also, I feel like stat arb is a strategy where you make or lose together, a la Aug 2007. A lot of hidden beta-to-strategy there, no?

Anonymous said...

Firstly, not all systems are mean-reversion. There are plenty of momentum-based systems, too, and their results (based on the managers I talk to) have been mixed. Clearly, momentum systems with shorter time horizons will have done better (!).