Wednesday, 15 August 2007

The Trouble With the Quants

First, we had the subprime mess. Now, the quants are getting chopped up by the market implosion. Quantitative funds, with their automated trading strategies, are responsible for a huge chunk of the $1.5-trillion hedge-fund industry. But many the the biggest and most prestigious saw their computer models fail in recent weeks.

Most notoriously, Goldman Sachs' $9-billion Global Alpha quant fund is down 27 percent so far this year, with more than half of those losses coming last week. Another $3.6-billion Goldman quant fund, the Global Equity Opportunities fund, lost more than 30 percent last week.

What went wrong? First, many of the funds rely on the same fundamentals-driven formulas based on stock valuation, piling into the same strategies and securities. Worked great on the way up, but the exit doors jammed on the way back down.

Second, the funds were leveraged by four times on average but in some cases reportedly up to 10 times. And third, as the assets in these funds exploded in recent years, they became market movers and couldn't exit without causing themselves - and other quant funds - greater damage.

I think a couple of the lessons are:

(1) Manage Your Risk. Some vet traders say risk management is the most important lesson they learned in their careers. We should listen to such advice. Regardless of whether you follow a mechanical system or trade off a chart, you've got to define your risk for each trade, put on a position that's appropriately sized for that risk and exit if the trade exceeds the loss you defined. No second-guessing on that last one. Just do it.

One common approach - the one I use in my trading - is the 2-percent rule: Don't risk more than 2 percent of your assets in any single trade. In other words, figure out where your exit would be, calculate the loss from your entry point, then put on a position that, in the event of such a loss, wouldn't cost you more than 2 percent of your total assets.

I have no idea what the geniuses at Goldman Sachs were up to, but some of the massive losses we've been reading about suggest excessive risk. Take a look at the past drawdowns in my setups (see the "Trading Setups" link above). Forget leverage. Based on those drawdowns, even my best setup has a maximum portfolio allocation of 25 percent.

I've no doubt that the brainiacs at some of these quant funds have come up with more profitable and less volatile setups than mine. But if we assume that's the case - and, thus, those setups had less historic risk that actually permitted such high use of leverage - then the trades should have been cut short just as soon as the recent losses exceeded those very small potential drawdowns.

Something clearly failed either in the conception or execution or, very possibly, both.

(2) Diversify and Learn. This means trying and testing new strategies and markets and always learning. I like using the Commitments of Traders data in my trading because it's completely independent of the price action and thus brings a fresh perspective that's not widely followed. At least not yet.

And I don't think that's likely to change, either. The COTs data, at least my reading of it, generates signals that work on a longer horizon than many investment funds are able to use. On the other hand, that's what makes it attractive for a long-term portfolio like that of many ordinary folks.

The COTs really are the great leveler - the people's friend in the markets!

At the same time, I think it's important for me to keep trying new strategies for this data and refining my existing ones.

For further reading on the quant mess, have a look at Rick Bookstaber's interesting piece at his blog: here

Other interesting articles: CASTrader Blog, Zero Beta and Abnormal Returns

No comments: