Aug 10, 2007

that was painful

Some serious market gyrations this week. While the indexes were moving up and down there was a much more interesting story behind the numbers. Stuff always goes up and down but this week something weird was going on. Quant funds were getting crushed. By quant I mean funds that use some type of analytic model to assist their stock picking (there are other types of quant funds). Our fund is part quant and we got fucking hammered Tuesday through Thursday. Hammered! Typically valuation metrics (stuff like P/E) and momentum metrics (price increases) go into these models. They are all proprietary models but a lot of Wall Street uses them and they all have a lot of similarities. In general these models pick cheap stocks that have good momentum.

This week though everything was reverse. Expensive stocks with terrible momentum were going up and the cheap stocks with good momentum were getting hammered. At least that's the top level story being spun around the street. But no one really had an explanation why this was the case.

The reality of the situation, which I have not seen written anywhere, is that there was some massive short covering going on. Or at least it seems that way. I suspect it was related to the sub-prime mess. Liquidity is drying up right now and highly leveraged funds are getting client redemptions (people want their money back). So these funds needed to either raise money by taking out debt (tough to do when liquidity is drying up) or sell their holdings. Well they can't sell their sub-prime holdings because no one wants them. So they sell their long positions and cover their shorts. Many of which could be ideas generated from their quant models. They could in fact be very good bets. But no matter. They need cash so they sell. Remember kids, leverage can be very bad.

I did a fair amount of number crunching this morning to get to the bottom of this. Typically a stock's short interest has no medium term relation to its performance. Short interest is the number of shares in a stock that have been sold short over the total 'float' (or number of stocks that are free for trading). In each of the first 7 months of this year the correlation between short interest and stock performance varied between 0.03 and -0.03. Correlation coefficients can vary between -1 and 1. 1 is highly correlated, 0 is uncorrelated, and -1 is inversely correlated. To put this in perspective, any correlation above, say 0.15 between a stock's performance and another metric means you can make some serious money. Granted the correlation lasts over the long term. Tuesday through Thursday the correlation was 0.4. Meaning heavily shorted stocks were going up much more than lightly shorted stocks.

Well the heavily shorted stuff is, in general, shitty overpriced stocks that quant models love to short. So everyone with any quant model was getting hammered over those two days.

But since the correlation over previous months is so little, it should tell you that this hammering we all took should reverse quickly. Within the month. And in fact it can be taken advantage of because while you are getting hit you should be stocking up on good stocks. You'd also want to short more but you effectively are after a bad day because compounding works the opposite way with shorts. The snapback came today. On a normal day we are up/down 0.0% to 0.5% relative to your benchmark. The last few days varied between 1.0% and 1.5%. Negatively of course. Today we were up 3% relative. A huge snapback. And I expect more to come next week.

No comments: