Lessons for Innovators in Quant Fund Sell-Off

In the media coverage of the losses sustained by quantitative hedge funds in recently, there's been a tendency to blame the computers. That's only partially right, I think. But there are, nevertheless, some lessons for IT leaders about innovation here.

Quant funds are investment funds that are managed electronically; computer models determine when to make trades. Quantitative trading has been popular with fund managers since the early 1990s because computers help them analyze more data about more stocks more efficiently than any human can. And, like any automated rules-based system, it streamlines the execution of transactions. Though fairly commonplace these days (not only hedge funds use them), they have been a major innovation.

In the post-mortem to the hedge fund sell-off, some fund companies have attributed the losses to the models. That might make you think that the problem was due to a lack of innovation in the computer models. Cue a scene with a bunch of data wonks in a basement near Wall Street crunching code, trying to come up with a bulletproof investment model that will make them billionaires.

Then I caught up with Ernest Chan, a quantitative trader and consultant, who told me the problem has not so much to do with the models (although some, he thinks, are actually over-engineered). Instead, he says, it's a manifestation of a basic economic problem. "There is too much money out there that is chasing high returns." In other words, technology applied to a flawed business strategy won't help you much.

In this case (and I'm simplifying), the hedge funds were racing away from an investment—mortgage-backed securities—that no wanted to buy even at a bargain-basement price. Fund managers had to raise cash to cover their losses, so they sold off whatever they could—including good investments that they should have been able to use as hedges against the ones that tanked. Chan says that when the funds need to raise cash, they'll sell off investments even when the models tell them not to. Add in a credit crunch – banks didn't want to lend money to carry the funds through their cash crisis—and it's no surprise the markets spiraled downward.

A financial services industry CIO who I talked to about this says the recent market behavior is about 25 standard deviations away from normal. "You can't even see the tail end of that bell curve," he says. So this particular scenario isn't likely to happen again. No trading model is going to be able to anticipate every eventuality, and everyone knows it (see this Washington Post story).

I see these lessons for innovators:

1. Sometimes it pays to make a big bet, but you're asking for trouble if your investment portfolio is stacked with high risk projects.
2. The more crowded the market, the smaller the returns. If you're going to throw a lot of money at a problem, you need to work fast to gain the greatest benefits from the solution.
3. You can't predict the future. However, you can revisit the fundamentals of your idea to make sure it remains sound. Then you can adjust your approach if conditions change.
4. Fail fast. When you hit a dead end, pull the plug and cut your losses quickly. If you have to wait to be told you've made a mistake, you've spent your credibility.

Thoughts?