The Beauty of High-Speed Trading

As an academic, you often study things that your friends and relatives are vaguely aware of, but never ask you about. Sometimes, however, you get lucky. Sometimes you wake up and find that a famous author has written a book covering one of your research topics, instantly making it relevant and a topic for dinnertime conversation. On Monday, this happened to me.

For those who may not have heard, Michael Lewis, of Moneyball fame, has written a new book on high-speed trading entitled, Flash Boys: A Wall Street Revolt. It’s currently the number one seller on Amazon.

What exactly is this high-speed trading that Lewis is talking about? In short, it is completely automated (computerized) trading that seeks quick profits using extremely fast connections to markets. It’s big, making up about half of all trading in the market. It’s also tiny, being dominated by a handful of small (relatively speaking) proprietary trading firms.

What does Lewis have to say about it? Well, Lewis thinks high-frequency trading (as high-speed trading is often called) is bad — I mean really, really bad — and that high-frequency trading firms, in collusion with market exchanges, have rigged the market against you. Strong claims, no doubt. But is he right? I’ll get to that in a second.

First, let me share with you a little about my research (for more, see here). Most academics find beauty in what they study, no matter how ugly it is, and I’m no different. I want to describe for you the beauty I’ve found within the beast that is high-frequency trading. Put simply, it is the following: High-frequency trading synchronizes prices across global financial markets. It knits exchanges together into one gigantic marketplace where the prices of financial securities move in unison, much like a school of fish (see the movie below where the prices of 40 securities are tracked over one day). Why does that matter, you say? Let me try to explain.

When you trade in a financial market, something quite amazing happens. Information about your trade travels at nearly the speed of light outwards from the exchange; it traverses through mountains, under seas, across microwave towers, and along other telecommunication routes on its way to nearly every other major exchange on the planet. As the signal ripples outward, these other exchanges become active — they “light up.” Why? Because, high-frequency trading firms are updating the prices of all other related securities in response to your trade.

The ripple that I’m describing reaches everywhere across the globe, and it is spectacular to watch. Brad Katsuyama, the hero of Lewis’ book, was awestruck when he first saw it. As the signal from one of his trades hopped almost instantaneously from one exchange to another, he told his colleagues, “You see. I’m the event. I am the news.” It was his eureka moment, the moment when he first came face to face with the soul of high-frequency trading.

Interestingly enough, Lewis and Katsuyama interpret the rippling effect of high-frequency trading as evil. Based on the 60 Minutes segment on the Flash Boys book and on the NY Times excerpt, it seems to be the focal point of their “rigged” claim. I’ll explain why they think rippling is bad later. Right now, I want to show you how it can be good, extremely good, perhaps even elegant and beautiful. I also want to show you how it can be very scary, bad even, but not at all in the way that Lewis and Katsuyama state. To do so, I want to revisit the fish.

Why exactly do groups of animals, such as fish, synchronize their motions? It is because synchronization allows them to link together, forming one giant organism that can observe its environment with “many eyes.” If one fish spies a potential food source and darts after it, the rest will follow. If another sees a predator approaching and moves out of the way, the rest do so as well. In ecology, the benefits of synchronization are well-known: Groups that synchronize make more accurate decisions, and they spend fewer resources on information gathering.

I think financial markets are similar, and I have good company in that opinion. In markets, the state of the economy is monitored by a large number of investors who broadcast changes to each other and the rest of society via price movements. Just as in animal groups, synchronization aides in this process. It allows the group to make more accurate decisions and to spend fewer resources on information gathering.

In practice what does this mean? It means that synchronization should increase the accuracy of prices and decrease the costs of trading for average investors. Lo and behold, this is exactly what has happened in markets as they’ve become more synchronized. For a single transaction in a liquid US stock, pricing inaccuracy reduced by 30% from 2000 to 2005 and a further 40% from 2005 to 2010. The cost of a single transaction reduced by a whopping 70% from 2000 to 2005 (partly due to decimalization and smaller order sizes) and a further 50% from 2005 to 2010. Large orders by passive mutual fund companies are also paying less. Vanguard has reported a drop in their transaction costs of 35% to 60% over the last 15 years. Pretty nice results for your average investor.

So what is Lewis’ beef? Well, this is where it gets interesting. Lewis and Katsuyama claim that high-frequency trading front-runs large traders. What does that mean? It means that when a large trader comes to the market and places an order, high-frequency traders try as hard as they can to propagate information about that order quickly throughout the market. It’s the ripples thing, which I’ve claimed is generally good.

So why do Lewis and Katsuyama claim it is bad? Well, because it means that the large trader’s information has leaked to the market quicker than they wanted it to and their transaction costs are higher. But here’s my take: You know that thing in animal groups about fewer resources being spent on information gathering? You’re seeing that in practice here.

This is why Lewis’s “rigged against the little guy” story depends on some rather convoluted reasoning. Basically, it goes like this: Actually mom and pop are okay when they trade in the market, it’s the large hedge funds and institutional traders who are being front-run. But, hold on, that’s bad for the average guy because their pensions and retirement funds are actively managed by institutions.

I’m not quite buying it. There is little to no evidence that actively managed funds benefit average investors. In fact, there is a well-known academic paper that suggests any excess returns earned by these funds are optimally extracted by the fund managers. To be honest, I’m not overly concerned if fund managers make less money. (By the way, mom and pop, you have put your retirement money in an index fund, right?)

Okay, so I don’t quite buy Lewis story, but I also believe ripples can be bad. How so? Well, let’s think about fish again. If a small group of fish get a wrong signal, that signal can propagate throughout the entire group and everyone converges on the wrong answer. Furthermore, who’s to say that high-frequency trading gets these ripples correct? Maybe they’re jiggling things the wrong way? Finally, how comfortable are you knowing that the entire financial system is connected by super-fast machines with little to no human involvement? The flash crash propagated throughout financial markets precisely because the markets were being run by machines.

One last note: If there’s one thing that everyone tends to agree on, it is the following: Current financial markets are really complex, perhaps overly complex. When you have myriad order types that not everyone understands, two different ways of obtaining market data and pricing orders, and outdated regulation, there certainly can be room for questionable activity, such as quote stuffing or regulatory arbitrage. My story above is a bit of an oversimplification. But so is the idea that high-speed trading can only work to the detriment of the little guy — clearly the little guy, so far, has seen some significant benefits.

Austin Gerig is a Senior Research Fellow in the CABDyN Complexity Centre, Said Business School, University of Oxford.