You will hear people talking about “latency,” which means the delay between a trading signal being given and the trade being made. Low latency — high speed — is what banks and funds are looking for. Yes, we really are talking about shaving off the milliseconds that it takes light to travel along an optical cable.

So, is trading faster than any human can react truly worrisome?

The rest is here (if you want to read more about the practical ethics of quantitative finance). What interests me is the more general notion of mechanistic processes making our decisions for us and inhibiting our ability to recognize and react to possible hazards.

The problem isn’t simply speed (or delay), it comes down to mechanistic processes and structures that are inaccessible to human decisions, neutralizing the power of human judgement and intuition to deal with emerging patterns.

Read another recent New York Times piece about how valuable hunches are in battle:

“Not long ago people thought of emotions as old stuff, as just feelings — feelings that had little to do with rational decision making, or that got in the way of it,” said Dr. Antonio Damasio, director of the Brain and Creativity Institute at the University of Southern California. “Now that position has reversed. We understand emotions as practical action programs that work to solve a problem, often before we’re conscious of it. These processes are at work continually, in pilots, leaders of expeditions, parents, all of us.”

The mind accomplished a lot over a couple hundred thousand years and we still don’t understand how the most fundamental processes work. Let’s not sell it out just yet. The mind — including intuitions, emotions, the whole ball of crud — is still our best asset.

What the machine-processes give us is objective safeguarding and reference — like instruments in an airplane.

But don’t planes have autopilot too? Isn’t that the same as algorithm-generated trading?

No. I think the difference is that the plane still goes the same speed and the pilot can still notice emerging dangers and use discretion to take over the controls. A fighter pilot wouldn’t let the computer fly in a dogfight (do they still have those?) — though computers come in damn handy to analyze and prioritize potential threats and targets (something I read about in Popular Mechanics when I was a kid, I think for since-cancelled Comanche helicopter program: I don’t know how sophisticated the real flight and weapons systems are).

In quantitative finance, the computers don’t just take over the controls for a while to make things cheaper and easier; the computers have actually changed the nature of task, introducing a different set of directions we can’t handle. There’s no “going manual” anymore, and we lose the ability to identify and manage emerging patterns.

Automated functions must exist to help us identify and correct mistakes (either by freeing our attention to notice them, or by providing objective benchmarks and measures), but they should not drive the process.

No related posts.