Should stock trading get high-frequency spectrum?
Aug 23, 2023 | 5:26 AM
When I first saw the phrase “high-frequency trading” or HFT, I assumed it was an advanced engineering technique for trading off and managing spectrum use in order to increase channel capacity or improve signal-to-noise ratio (SNR). I was very wrong.
Instead, it’s related to a petition to the Federal Communications Commission by Wall Street-related trading firms to allow them to use some slices of spectrum to set up “private” high-power transmitters in the high-frequency (HF) bands under 30 MHz, also traditionally called the shortwave bands. So, they can transfer stock and related pricing between cities such as Chicago and New York a few milliseconds faster than achievable using optical links.
This tiny increase would allow them to “run ahead” of trades by others and take advantage of tiny price differentials to reap additional profit. This scheme has its own name, “latency arbitrage”.
The petition from “Shortwave Modernization Coalition” (SMC) is assigned Docket RM-11953 and titled “Shortwave Modernization Coalition Petition for Rulemaking to Amend the Commission’s Rules to Allow Fixed, Long-Distance, Non-Voice Communications Above 2 MHz and Below 25 MHz” (be prepared: it’s 105 double-spaced pages, all text). Their petition request would potentially put 50-kHz wide, 20-kilowatt signals immediately adjacent to seven amateur-radio (ham) bands.
In contrast to those power levels, amateurs are restricted by Part 90 high-frequency rules to 1,000 W peak envelope power (PEP). Even that peak-power number is somewhat misleading, as most radio amateurs operate at under 100 watts (it’s called “running barefoot”) and many are routinely under 10 watts while still achieving long-distance and even worldwide contacts. Further, SMC’s proposal would reduce the existing protection of -73 dB edge-of-band/out-of-band attenuation now in use for the 1,000-watt power limit to just -50 dB protection for their proposed 20-kW limit.
This proposal includes four likely transmission scenarios:
New York, NY transmitting west, such as to Los Angeles, CA
Chicago, IL, transmitting west, such as to Seattle, WA
New York, NY, transmitting south, such as to São Paulo, Brazil
Chicago, IL, transmitting east, such as to London, UK
What sort of decrease in propagation delay are we looking at? Basic physics and math show that for a typical RF-hop distance they would use, the saving between wireless atmospheric path and an optical-fiber path would be on the order of 10 milliseconds, and less in many cases.
The use of radio links for stock-trading HFTs is not a new development. RF links are already used with point-to-point microwave towers linking major trading centers. However, many of these systems use microwave links where there is more available bandwidth than the high-frequency bands. Also, since they are point-to-point systems, their emitted RF energy is more limited, and the beam spread is fairly narrow.
However, a long-distance microwave link takes many towers, since the tower range is generally limited to 30 to 50 miles maximum due to curvature of the Earth’s curvature and is a function of tower height (Figure 1).
Figure 1 Microwave relay towers, typically spaced up to 30 to 50 miles apart and using focused beams, have minimal RF “splatter” and have only modest potential for causing adjacent-spectrum RF interference. Source: Dr Jai W. Kang
Long-distance high-frequency latency arbitrage links are already in use. Unlike microwave towers, it only takes one antenna at each end of the link. However, that same link and its path are subject to all sorts of performance inconsistency issues and “skip zones.”
These are a function of atmospheric propagation conditions which vary with time of day, sunspot cycles, and many other factors which can only be predicted to a linted extent (Figure 2). Moreover, the effective data rate is low, but may be enough for the HFT application.
Figure 2 Long-distance RF links are subject to the vagaries of atmospheric propagation characteristics in addition to somewhat predictable changes. Source: Australian Space Weather Forecasting Centre
Software engineer and ham radio operator Bob Van Valzah has identified such high-frequency antennas in his Chicago area. He has also done the deep digging through the licensing firms and layers of trading corporations which obscure ownership and operation, as detailed in his blog “Shortwave Trading | Part I | The West Chicago Tower Mystery.”
This latest application is for higher power and frequency slices which may cause adjacent-channel problems. Obviously, the amateur-radio community led by its primary user association, the ARRL, has filed many objections to the proposal, pointing out the likely spillover interference and potential safety/emergency-related issues as well as day-to-day problems. There are about 760,000 amateur radio operators in the United States, according to ARRL, and while many are not active, the base number is growing.
But the issue is not which side has more members or money, it’s really about the best use of the limited resource of the electromagnetic spectrum and the harm that misuse of the spectrum can do to other users and services.
Keep in mind that the spectrum is an unusual resource. On the one hand, it’s limited, and you can’t make more of it. Even though the entire spectrum follows Maxwell’s equations, different parts of it have immutable attributes: compare low frequencies with terahertz ones, as one example.
On the other hand, it is infinitely recyclable and using spectrum does not consume it, unlike using a tangible resource such lithium, or even helium, which dissipates into the atmosphere and then outer space and cannot be recovered once gone. In that sense, we are very fortunate that spectrum is used but not consumed.
What’s your view on use of spectrum for limited-use, private links such a HFT versus broader applications? Does this spectrum allocation make sense? Is the interference risk manageable? Is it perhaps worth trying, since the decision can be reversed, and the used spectrum recovered? Or is reversal of allocation after such a large HFT investment not likely to happen, even if there are interference issues?
Comments