Many high-frequency trading (HFT) firms have disappeared into larger firms as merger activity has increased recently. Those acquired include Chopper, Infinium, Teza, RGM Advisors, and Sun Trading. It also includes Getco, the Chicago-based firm founded in 1999 by two former floor traders that almost defined the field for some time.
The consolidation of those HFT firms that focus on the US equities space is especially marked, both in itself and as a share of US equity trading. This January, the Financial Times said that the two largest firms in this space, Virtu and Citadel, between them “account for around 40 percent of daily US trading flow.”
The Fragility of Competition
A new paper by Donald MacKenzie, a Professor of Sociology at the University of Edinburgh, looks into this consolidation, and into what MacKenzie calls the “fragile” character of competition among traders in the HFT context.
MacKenzie expresses some doubt about the Financial Times’ number, which “could be an exaggeration.” But it is noteworthy that at least a near approach to that 40% figure does not strike him as implausible. Logically, as the volume in more and more markets comes to be dominated by HFT trading, it is mathematically inevitable that consolidation in the latter will mean consolidation in the former.
One obvious reason for the fragility of competition in HFT is the cost of entry. Recruiting and retaining the necessary highly qualified staff will always be a considerable cost. Beyond that, the exchanges stand as gatekeepers. Citing a paper by Charles Jones of Columbia, MacKenzie says “a medium-sized proprietary firm trading US equities would also have to pay around $1.8 million a year for data feeds from the equities exchanges owned by the three main operators, NYCE/ICE, Nasdaq, and CBOE.”
Something More Profound
We’ll return to the issue of new entrants. But MacKenzie is especially concerned with what he calls “something more profound” in the changing nature of the technology now used to trade US equities. The trouble with a race to be the fastest in making these transactions is “not simply that it’s expensive, but that you can have a winner: that, at least in a particular market … one HFT firm—or a small number of firms—may achieve an advantage in speed that’s very hard and very costly for their rivals to overcome.”
Part of the problem is the decline of “jitter,” the random variation in processing time for transactions in exchanges. Jitter occurs for several reasons, some relating to system architecture, others to the randomness built into the physical world, such as the heat-induced agitation of electrons. Whatever its sources, jitter is an equalizer. Given jitter, the highest speed trader doesn’t always get the trade. Heck, given enough jitter even the tenth fastest trader trying to get a particular trade will win sometimes: perhaps often enough to stay in business, while working on ways to move up in that league table.
But the equalizer is failing. As system architecture becomes more sophisticated, the level of jitter drops, and this makes even small increments in speed, differences that would once have been swamped by the jitter, critical advantages.
MacKenzie says that he first interviewed high-frequency traders at the Chicago Mercantile Exchange in the period 2012-2013. They were not very complimentary about the CME’s systems. One trader discussed a scenario in which his own firm’s round-trip time for a transaction is 0.9 of a millisecond. Another firm’s (in this scenario), a full millisecond, which is to say a microsecond of difference. The latter might do as well as the former on the CME—and, it should be added, on other exchanges where signals from the CME serve as guides to algorithmic trading.
The trader appears to have recounted this scenario to MacKenzie with an air of grievance, as if the fastest speed should by the laws of nature “win” the trade.
If that counts as a grievance, it has been addressed. It has been addressed all too well, MacKenzie says. That one microsecond difference would now, in a time of reduced jitter, loom quite large.
Despite the costs, are there new entrants? Yes: they do exist. But MacKenzie says that as far as he can tell only two of them, XTX Markets and Headlands Tech, are of the same scale as the many firms that have disappeared of late. Competition still exists, but “its health should certainly not be taken for granted.”