Predatory HFT Strategies: Is “Information Transmission Zoning” the Solution?

In a follow-up blog post to my recent two-part series on high-frequency trading (HFT), I examine a clever proposal by Khaldoun Khashanah, Ionut Florescu, and Steve Yang who have released a report sponsored by the Investor Responsibility Research Center (IRRC) Institute that may alleviate the issues of predatory HFT algorithms outlined in the previous two blog posts.

The problem of latency arbitrage is fundamentally one of information dissemination. Co-located HFTs are able to access, process, and transmit trading instructions more quickly than other market participants. One proposed solution is IEX’s “speed bump,” which slows down the information flow to co-located HFTs. However, this solution is piecemeal and does not address the issue at any other exchange. The IRRC report presents a novel solution that requires minimum changes to Regulation NMS and claims that it would, at a stroke, eliminate the problem of latency arbitrage without having a negative impact on liquidity. To determine whether it’s too good to be true, let’s look at their proposal.

They first introduce the concept of information transmission distance, which is the distance measured in time (not space) of a given market participant from a given exchange. In other words, if a microwave-tower-enabled HFT is 50 miles away from an exchange, it may still be closer in time to the exchange than a mutual fund 10 miles away that relies on cable connections. Information transmission distance allows a more accurate ranking of market participants in terms of their ability to access exchange trading information. In turn, this provides the framework for removing the advantages of co-location and the possibility of latency arbitrage.

Consider that the Securities Information Processor (SIP) is, let’s say, 1,000 microseconds away from the exchange. That is, when a trade occurs on the exchange it takes 1,000 microseconds before this information reaches the SIP and appears on the consolidated tape. Latency arbitrage would ordinarily occur because co-located HFTs could see the exchange trades 500 microseconds after they occurred and, therefore, have time to front-run this information. Given that the public information time horizon (i.e., the SIP) is 1,000 microseconds, the report suggests classifying HFTs according to whether they are closer than these 1,000 microseconds to the exchange.

The authors argue that HFTs that are in a zone closer to the exchange than the SIP should be classified as electronic specialists that effectively, if not technically, have privileged high-frequency insider information. The researchers note that some HFT firms only report losses one day per thousand days of trading — a return pattern that resembles a broker fee structure, not a trading return. As a result, the authors suggest they should have the same trade-facilitating obligations as specialists (to use the traditional NYSE parlance). Specifically, specialists are prohibited from trading ahead of investors who have placed orders to buy or sell a security at the same price. In other words, they would have negative obligations. This approach would continue to allow liquidity-generating HFT strategies inside the SIP information horizon that have been found to improve market quality. HFTs that are in a zone farther away from the exchange than the SIP would now, by definition, have the same level of access to trading information as the consolidated tape and so could pursue any legal (even predatory) strategy. However, latency arbitrage would now be impossible because the HFTs outside the information horizon would not see trade information before other market participants.

I think that this concept of information transmission distance zoning is a potentially very elegant solution that does not appear to require any changes in technology or regulations, and that may actually provide a systematic way of observing and regulating market participants in relation to their access to the fundamental unit of markets — information.

Critics will probably point out that HFTs are not trading on inside information, they are trading on public information but more quickly than the SIP. This is true, but I believe largely irrelevant; there are parallels to be drawn in legislation such as the EU’s Market Abuse Regulation (MAR). In the sections on information disclosure the MAR notes that when disclosing inside information, it is not good enough to simply bury a document on a website and argue that an informed investor could have found it; it is necessary to make concerted efforts to ensure the information is observed by market participants. Similarly, it is probably not useful to view a trading signal emanating from an exchange as representing the public disclosure of that trade — the majority of market participants won’t see it until it gets to the consolidated tape via the SIP.

Supporters of HFT often question the need to restrict a technically legal and ostensibly fair market activity. After all, any institutional investor could pay for co-location and greater computing power, so if they are not willing to make the investment, why punish those who spend the resources?

I believe the world of motor racing can give us a helpful analogy. Teams will often spend millions to develop superior technical solutions that give them an edge over their competitors in order to win. When these solutions are revealed, their rivals usually announce their intention to copy the technology in order to regain parity. At this stage the governing body of the sport typically bans the technical solution in question amid howls of protest by the innovating team. Clearly, the governing body is punishing the innovators and clearly this is unfair to them. However, the argument is that the product being produced — the motor racing event — is more important and having teams spend millions to develop mostly socially useless technology in order to end up with broadly the same parity as before threatens the financial viability of the sport and therefore the product.

Similarly, while it is unfair to limit the latency of HFT traders who have invested tens of millions into microwave towers and co-location, ultimately it is market quality and integrity that are important. Currently, market prices and depths can mean little in a world of phantom liquidity and price slippage, and this is not a good thing for market integrity. One solution is to point at institutional investors and tell them to up their game to the level of HFTs in terms of latency, but perhaps the information transmission zoning idea is a cheaper and quicker-to-implement solution.

If you liked this post, consider subscribing to Market Integrity Insights.

Image credit: ©iStockphoto.com/Peshkova