In the time it took you to read those two words, high-frequency algorithms operating across New York, London, and Frankfurt allow financial traders to make millions of dollars.
Although many countries have attempted to regulate the meteoric rise of high-frequency trading, no plan has been more ambitious than Germany’s High-Frequency Trading Act (HFT Act). Rather than regulate trading speed alone, the HFT Act targets the complex core of high-frequency trading: financial algorithms.
And there may be some evidence that the HFT Act is working—at least in part. In a recent paper, Nathan Coombs, a Research Fellow at the University of Edinburgh, grappled with the complexities of trying to define, identify, and monitor well-guarded financial algorithms, and concluded that the HFT Act—although far from perfect—has had a notable degree of success.
Coombs first describes how algorithms have revolutionized trading markets in Europe and the United States—replacing the iconic scene of traders shouting orders at each other on a crowded exchange floor with automated computer systems that execute orders using complex mathematical codes. He explains that these high-speed algorithms allow for trading speeds measured in microseconds, beyond human perception. Over the past decade, they have made headlines for their ability to cause billions of dollars to dissipate and reappear within a matter of hours.
Coombs defines trading algorithms as “iterative decision-making procedures”—but he adds that the precise meaning of that definition has vexed regulators worldwide.
The problem with determining what constitutes an algorithm, Coombs argues, is that a financial algorithm is not just the code itself—it is the entire process of a trading strategy. In other words, he observes, algorithms “live a double life: identifiable with their strategies as well as with the computer code in which strategies are operationalized.” These dual components of an algorithm—code and strategy—make it difficult for regulators to pinpoint their efforts.
In response to growing public concerns about high-frequency trading, BaFin—the German Federal Financial Supervisory Authority—began implementing the HFT Act in 2013. The Act’s stated objective is to “mitigate the potential risks arising from the speed and complexity of algorithmic high-frequency trading methods.” Public unease about speed-trading initially fueled the Act’s passage. According to Coombs, however, the main problem that German officials identified when drafting the legislation was determining whether intentional algorithmic strategies were causing manipulative market fluctuations.
The algorithm-tagging rule is an integral provision of the HFT Act that addresses this causal uncertainty. Each time an algorithm generates a trading decision, the rule requires that the firm identify that algorithm by assigning it a number. This novel approach of using identifiable numbers to tag algorithms, Coombs argues, balances two conflicting concerns: high-frequency firms’ right to keep their algorithms secret against the need of regulators to be able to identify market manipulation.
Critics of the HFT Act focus on the sheer impossibility of tracking thousands of constantly evolving processes. For example, Coombs notes that one critic has argued that, even with increased resources for regulators, the prospect of accurately monitoring all of the high-frequency algorithms is grim—even without taking into account potential compliance issues.
Another critic has claimed that attempts to create market transparency erect a “wall of incommensurate, uninterpretable, overwhelming information.” And as recently reported on RegBlog, this critique may be amplified by findings of “fragmentation, overlap and duplication” throughout the U.S. financial regulatory sphere.
Coombs acknowledges that the HFT Act does not do a perfect job of tagging and monitoring high-frequency algorithms—especially given the up-front challenge of defining what an algorithm is. He also recognizes that the HFT Act’s requirement that firms report “material changes” to their algorithms contains some ambiguity. Because changes can occur as frequently as every few weeks, whether a change is actually reported can vary depending on how a firm interprets “material change.”
But Coombs argues that such criticism misses the point. “In a perfect world,” Coombs recognizes, the tags that regulators use would “mirror” algorithms. However, some tolerance for imperfections, he argues, is necessary since regulatory resources are insufficient to achieve perfect oversight.
Coombs maintains that tolerance for the imperfections of the HFT Act has had positive returns for Germany’s financial market. Coombs identifies two major benefits of the Act: improvement of algorithmic transparency and of the industry’s culture.
The first benefit addresses high-frequency algorithms’ reputation on Wall Street for being well-guarded—allegedly even more restrictive than some areas of the Pentagon. Named to reflect this heightened security, the algorithms are stored inside a technological vault known as a “black box.”
Algorithm tagging gives German regulators data that sheds some light on what goes on inside of a black box. Since the data regulators receive is imperfect, monitoring the relationships of algorithms by itself does not reveal market manipulations. The data does, however, equip regulators with a better vantage point to investigate foul play. For Coombs, regulating in the shadows is better than the complete darkness of the black box.
The second benefit of the HFT Act that Coombs identifies is an unexpected cultural shift in the high-frequency trading industry. After interviewing a compliance officer, Coombs explains that in the “early days,” traders freely tested algorithms on the market—brushing aside compliance officers who tried to monitor complex strategies.
Coombs argues that the HFT Act provides a strong mandate for compliance officers to not only enforce the law, but also to increase their technical knowledge. In doing so, he believes that compliance officers can better ensure traders are able to understand and justify the strategies they are implementing.
Financial regulation should respond to an increasingly complex market, Coombs concludes. Rather than demand an “all-or-nothing” approach, regulators around the world should be receptive to imperfect data-gathering—and should reflect on Germany’s ambitious response to a practice that continues worldwide.
How regulators react to Germany’s HFT Act remains to be seen. One thing is certain, however: traders will not be slowing down any time soon.