`

Please Stop Talking About High-Frequency Trading and Blockchains

I really only have deep familiarity with two concepts in finance today: the capital markets case for blockchains and high-frequency trading. Imagine my frustration since the two are the most ill-understood topics in the popular press* and the oil-and-water of technological applications. Yet, despite this, I often hear people pattern match and try to think of ways to join the two. A lesser woman would have pulled her hair out by now.
 
High-frequency trading (HFT) is a type of algorithmic trading which places emphasis on the latency at which a system can react to changing order books and place orders in the market. It has become something of a boogeyman in recent years. It exists at the confluence of market structure, the competitive landscape of exchanges, and regulations. HFT saw its heyday after the passage of Regulation NMS, which ended upstairs trading and incented institutional players to execute their orders on the exchanges. These institutions need to comply with this constraint without tipping their hand. As a result, their trades are now internalized by large execution brokers or sent to dark pools. This converges into an ecosystem of aggressive price takers and defensive market makers trading at high speed on exchanges. To be sure, HFT has also grown in other countries and other areas of finance which aren’t subject to the same regulatory constraints, but the general lesson is that HFT is not a necessary condition of finance but rather a specific response to market and regulatory forces.
 
In Capital Markets
Perhaps the worst pitch I have ever heard was from a blockchain vendor who wanted to “solve” HFT using blockchain technology.** This made no sense on several levels, but perhaps the least coherent aspect of the pitch was the suggestion that blockchains should be some sort of trading platform. When offering liquidity, the market-maker needs to be able to change quotes fast.
 
The structure of a blockchain does not facilitate low-latency applications. Blockchains are good at recording things, including complex things. They are not an appropriate platform for trading. It makes more sense to conclude trades outside of a ledger and use the ledger for settlement purposes. Some people have proposed blockchain HFT schemes where they pass transactions among a group and use signature schemes as a method of off-chain transaction netting. This makes sense, but it’s more of a solution looking for a problem as better mechanisms for trade recording exist.
 
In Cryptocurrencies
Recently, I was at an Ethereum meet-up where someone asked about the ability to co-locate servers to an Ethereum exchange for some sort of HFT application. In an environment where transactions run at 20 to 30 basis points, and where market movements are determined by insiders (Bitcoin market fell markedly before Bitfinex was hacked), shaving a few milliseconds of latency should be the least of a market maker’s concern. Though some exchanges do offer better rates for adding liquidity, the high fees of taking liquidity mean that those market makers are unlikely to be picked off by faster traders.
 
We can imagine many circumstances where market makers and HFT shops would benefit from blockchains or distributed ledger technology in post-trade applications. However, in capital markets, trading platforms do not suffer from any problem which can be solved with blockchain technology. In cryptocurrency land, there are bigger fish to fry.
 
* - Except for anything written by Matt Levine. Matt Levine is basically infallible.
** - In the same pitch, I was told that derivatives should be fully collateralized. 0 for 2.