Ever faster computer networks over the past twenty years spurred increased automation in modern financial markets, that is, the emergence of high-frequency trading algorithms. Since machines can react within one billionth of a second to a trading signal, message congestion arises: to put numbers to it, one fifth of all transactions on Nasdaq are clustered in sub-millisecond “microbursts” of activity (Menkveld, 2018).
From a trader's perspective, the response to congestion is even more speed: traders spend resources to reduce their latency (e.g., acquiring faster computer chips) and physical distance from the matching engine (e.g., co-locating their server with the exchange) to race ahead and beat the rush. But, therein lies the rub: investments in technology are lump-sum in nature, and co-location subscriptions are often at monthly intervals. This means that, during times of low congestion, most low-latency infrastructure acquired by traders remains inefficently idle. How often, then, is speed necessary? Menkveld (2018) documents that only 10% of the microseconds in a trading day will experience an activity surge. Within a microburst, exchange messaging activity can increase tenfold. Therefore, traders with hyper-speed access to matching engines do not use 90% of their capacity for 90% of the time.
Not only can idle capacity be costly to traders, but traders' speed races may also harm market quality: Shkilko and Sokolov (2019) find that when rain disrupts microwave networks, slowing down trading, both market liquidity and price stability improve. To counteract the negative impact of speed, some exchanges explicitly seek to temper speed races by slowing traders down, either via order delays (speed bumps as implemented on Toronto’s TSX Alpha), or frequent batch auctions instead of continuous trading. But what if stopping the high-frequency arms race altogether is not the only way?
In our new research paper (link), we propose a FinTech-oriented, market-based solution to the high-frequency arms’ race where speed resources are liquid. That is, we posit a framework where traders can spend resources on latency-reduction “on-demand”. The goal of our set-up is to suggest that we may need only to localize speed races to the microbursts in which they are desired to improve the quality markets, instead of stopping the arms race altogether.
Our proposal relies on a recent development in financial technology, decentralized exchanges (DEX). DEXs are currently being implemented in the crypto-asset space: see e.g., Binance DEX, EtherDelta, or IDEX. As opposed to exchange server rooms hosting the infrastructure, the limit order book data and trade matching software on a DEX are distributed as smart contracts in a peer-to-peer network. Each participant in the network (miner) has a copy of the exchange itself and may “rent out” computer power (CPU) to process incoming orders, the fee for which depends on the supply for, and demand of, trading infrastructure.
A DEX would allow for flexible dynamic pricing of trading infrastructure, generating a mechanism where trading speed is, in a sense, liquid. The mechanism is similar to Uber’s surge pricing: when a (trading) demand surge occurs, the excess demand pushes up the price of computing power, and miners allocate more computer power (CPU) resources to the exchange. In normal market times, miners can re-route idle CPU power towards other, more productive goals.
Our main finding is that a DEX can improve the allocation of latency infrastructure. In a centralized exchange, high frequency traders (HFTs) purchase co-location subscriptions on a monthly basis, poised to act on short-term opportunities when—or if—they emerge. By contrast, in a DEX, HFTs can rent technology on-demand: profitable trading opportunities simultaneously generate “micro-bursts” in trading activity and surges in the price of processing power.
A decentralized exchange might actually speed up price discovery. If trading at a centralized market resembles a marathon where runners need to pace themselves, trading at DEX mirrors a sprint where runners “leave it all on the track” over the short interval. Consequently, the expected time from receiving an informative trading signal to a price update is lower in decentralized markets.
In the end, intense HFT competition for speed during microbursts triggers a surge in the price of computer power, leading on-demand pricing of low-latency infrastructure to achieve two objectives simultaneously: HFT rents from low-latency trading decline, as does overall resource consumption.
Michael Brolley – Wilfrid Laurier University
Marius Zoican - University of Toronto Mississauga