The Physical Nature of Technology: An Example of High Frequency Trading

Economists typically think of technology as an idea, but in some way it interacts with physical form. These physical forms influence how the technology is applied and its social effects. In his da Vinci Medal speech, Donald MacKenzie considers some implications of this idea in “Material Political Economy.” (technology and cultureJuly 2024).

One classic example he mentioned was a conflict that arose several centuries ago, during the feudal era, over the technology of grain milling. Feudal lords generally preferred a centralized system in which commoners brought grain to watermills or windmills owned by the lord and paid the lord for milling. However, many commoners would have preferred to avoid paying lords and instead mill their grain by hand. As a result, feudal lords sometimes sought to destroy handmills where they could do so. In this environment, technological choice is clearly not about efficiency in the abstract sense, but about the interaction of efficiency and existing social structures.

As a modern example, I was intrigued by MacKenzie’s discussion of ultra-fast high-frequency trading (HFT) for financial firms. He points out that when these companies were established in the United States in the 1990s, “HFT firms were sometimes excluded from deals or faced material barriers protecting the slow systems of incumbents.” But his focus is on more recent developments.

How fast is “superfast”? Every year, the European futures exchange Eurex publishes data from which the response times of the fastest HFT algorithms can be inferred. Eurex’s 2023 measurements suggest a state-of-the-art response time (for the market data packets that trigger the actions of the trading system) of 8 nanoseconds, or one billionth of a second. Light in a vacuum, the fastest physically possible signal, travels only about 30 centimeters, or roughly 1 foot, in a nanosecond. This is simply not a useful measure of HFT speed. Ensuring that messages travel as close as possible to the speed of light in a vacuum is an important practical concern for HFT. For example, fiber optic cables are not fast enough. This is because the refractive index of the glass in the cable core slows the laser light signal to about two-thirds the speed of light in a vacuum. Therefore, where possible, HFT firms send trading data and orders via microwave, millimeter wave or laser light signals transmitted through the atmosphere. It moves in the atmosphere at about the same speed as in a vacuum.

Therefore, high-frequency trading is not an abstract technological innovation, but an implementation in the world of matter, distance, light and microwaves. Mackenzie wrote:

HFT programmers cannot afford to regard computers as abstract machines, as presented during their college education. It should be viewed as a collection of metals, semiconductors and plastics through which signals pass, and ensuring that the signals pass through as quickly as possible is a prevalent concern. For example, the preferred programming language in HFT is C++. This allows for “close to the metal” programming without the need to operate through layers of abstraction like other languages. Moreover, since around 2010, existing computer systems, even if programmed in C++, are not fast enough to perform HFT in many markets. Trading algorithms are programmed directly into the hardware of a silicon chip known as a Field Programmable Gate Array (FPGA). There have been repeated rumors that companies are moving beyond FPGAs and into fully custom integrated circuits.

One result is what MacKenzie calls a “speed race.” In the HPT world, algorithms are programmed to react very quickly to new information. However, if the fastest algorithm places an order and reacts first and the price changes, the slightly slower algorithm will realize that it is reacting to “old” information. They desperately try to cancel orders, while faster algorithms try to take advantage of “old” large amounts.

Any bid or offer for the underlying stock put forward by the market-making algorithm immediately becomes “stale,” as market participants describe it. For example, if futures prices have fallen, purchasing stocks at the existing bid price is likely to result in a loss. Therefore, market-making algorithms rush to cancel old bids as quickly as possible, and liquidity-taking algorithms race to execute old bids before they are canceled. The difference between winning and losing in these speed races is now measured in billionths of a second, according to Eurex data.

This kind of “speed race” is happening every minute. Another physical manifestation of this technology involves towers and locations for transmitting signals from Chicago-based markets to New Jersey-based high-frequency traders.

The need for ultra-fast speeds makes very specific physical locations extremely valuable, and those who own or control them can therefore accurately calculate rents. Fiber-optic cables or wireless links that transmit data from one financial transaction computer data center to another must follow as closely as possible a geodesic, the shortest path on the Earth’s surface between the two data centers. In 2010, computer scientist Alex Pilosov led the build-out of the first microwave link for HFT between Chicago, where futures are traded, and northern New Jersey, where data centers where US stocks are traded. Pilosov kept a low profile in this operation to avoid alerting potential competitors. But about a year later, the owner of an attractively located microwave tower where he had leased space told him that other people were also looking to place antennas on that tower. He said: “I said, ‘I’ll tell you what’s going on, but you have to promise me that you’ll charge me three times what you’re charging me. And I promise they will pay.’ And it happened. That happened.” Likewise, Mike Persico, who has built both millimeter wave and atmospheric laser links at a New Jersey stock trading data center, reports that owners of high-rise buildings near relevant geodesics where this equipment can be deployed suddenly find themselves in possession of a very valuable asset. will. Persico says: “Sometimes these landlords end up with the equivalent of a Willy Wonka golden ticket.[d] These attributes were the furthest thing from their minds and suddenly… . . It becomes very profitable.”

Even economists might wonder whether the resources invested in high-speed transactions are improving the economy for ordinary workers or consumers. But I would also add that the development of new technologies often takes a circular path through a variety of applications. And I wouldn’t be surprised if it turns out that, as prices fall over time, high-speed communications will open up uses that we haven’t yet dreamed of. .