Andy Lo wrote a neat article on how stat-arb has done in the recent downturn. What I found most interesting was that the strategy he examined as the archetype of 'stat-arb' was mean reversion using daily horizons: past day's returns to determine winners and losers, hold for a day. I knew pairs was a common strategy, and generalized into a long/loser and short/winner strategy, but I had no idea people ran this at the daily horizon I always assumed 5-15 day. That's a lot of transactions, clearly a strategy that would benefit from great order execution systems.
But what is really neat is it highlights that trades that work are often quite simple. It's always funny to see that a trade of going-long-losers-and-short-winners, is often managed by a PhD in math. I guess it's no different than a 3rd string violinist, playing easy scores, but trained to do much more difficult work.
No comments:
Post a Comment