Longreads + Open Thread

Salt, Trading, Research, Wirecard Spies, App Spies, Growth, HFT



Trading at the Speed of Light: How Ultrafast Algorithms Are Transforming Financial Markets: A great history of high-frequency trading, starting with the gradual move from floor trading to electronic markets and culminating in the eternal race for lower latency. In most fields, "at the speed of light" means "fast enough to be instantaneous," but in HFT it's a nagging annoyance that certain fundamental constants of the universe limit just how quickly information can be transmitted from the futures markets in Chicago to equities markets in New York. This leads to immense investment in infrastructure so traders can get slightly faster, which feels like a wasteful and somewhat arbitrary competition—but it actually just replaced an earlier and equally-arbitrary competition, where one way to get preferential access to the order book was to be tall, loud, and willing to throw elbows.

The lower latency gets, the simpler algorithms have to be. You can't do some incredibly complicated math if it adds too many microseconds to the process of sending or canceling the next order, but in this domain, you don't really have to: at this pace, just having higher-than-average confidence that the next order will be a buy or a sell is enough to win. That also makes the business utterly brutal: in most fields, you're competing along multiple axes and can compensate for deficiencies in one place with advantages elsewhere. But if the goal is pure speed, you're either winning or losing. And as connectivity providers have learned to specialize in serving the needs of HFTs, they've raised prices—so firms pay a high fixed cost just to participate, and are either amortizing that cost over an enormous number of profitable trades or are bleeding to get to second place. (This dynamic works in the opposite direction, too; the longer-term the strategy is, the less concentrated investment management becomes, because there are so many more variables to think about over those scales, and the relevant variable changes.)

HFT is really a story of visible waste and invisible benefits: it's true that a great deal of mental horsepower has been expended in programming FPGAs to be slightly faster, building more microwave towers, carefully measuring the traits of individual fiber optic cables in order to make sure they have identical latency, etc. But all of that is automating away human labor that was zero-sum at the same level. So arguing about whether HFT is good or not means answering the intrinsically hard-to-answer question of whether one extremely clever electrical engineer forcing twenty quick-thinking floor traders to find another profession is a net win or net loss for humanity. (There are many fields where quick mental math comes in handy.) Those questions are fundamentally hard to answer without some sort of distributed system for collecting and synthesizing information that's widely distributed, i.e. exactly the sort of markets that HFT makes more efficient.

Open Thread

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.