Back in October of last year I was on a kick about thin value and thick value. I used High Frequency as an example of Thin Value because it only cared about the spread between the bid and the ask. Comedy Central did a bit on it that I reposted in my creatively titled entry A Funny Bit about High Frequency Trading .
The stock market, as you’ve probably heard, had one of the oddest days in it’s history a week ago. At one point the Dow dropped almost 1000 points and recovered in mere minutes. I’ve been waiting to write about it to see what happened. I knew it wasn’t a fat finger as was suspected. But I thought it could be cyberterrorism. Someone decided it was time to make a lot of money (make the markets drop, buy $40 stocks at a penny and then sell when they rebound to $40). But nothing has surfaced about that either. What appears to have happened is a classic example isolated engineering. What I mean is that one part of a system expects the other parts to only vary so much and builds that into the business rules, or in this case, the algorithms.
It used to happen a lot at intersections that had car traffic and train traffic. The light controlling car traffic has to know when train traffic is coming and circulate the cars through the correct passage; basically red light those that are perpendicular to the train rails. It seems obvious now, but people have died because the rail and car traffic partners didn’t consider each other in their safety approaches. Each independently worked fine, but together they didn’t.
There are two follow up writings I’ve enjoyed today about the high frequency situation. The first is a blog entry by James Surowiecki called The Flaky Stock Market and the other is an article by Floyd Norris called Time for Regulators to Impose Order in the Markets .
I like the Surowiecki post because he agrees with my opinion that the markets, although flawed, worked as intended. The computers don’t consider context. You build rules for them to follow and those rules are based on certain probabilities. But no one can predict the future, so when things get out of whack there must be some sort of reasonableness associated with it. And that is the topic of my thin and thick value posts – high frequency trading doesn’t care about the equity in question, it only cares about price. Fundalments are ignored. And it’s intentionally ignoring the purpose of capital markets in the first place – to fund money to efficient uses, usually succeeding businesses. New information and interpretation of the information isn’t happening, so the herding of ideas occurs and that leads to pooled risk.
I liked the Norris article because it lays out how having competition for markets is good, but they need to be coordinated. Creating a vacuum in one can unintentional set off a panic in another. Here is where establishing business rules makes perfect sense. For instance, if the NYSE puts in a delay, the other markets should flip a switch and delay as well.
Ultimately high frequency trading is good because it lowers transaction costs, but it encourages lazy management; the speed is so fast that you hardly notice. It lulls you into a false sense that the system knows better than you do.
Working Thoughts 