High-Frequency Trading Slowdown

Stock markets are not anymore large venues filled with people shouting and rushing to scalp Blue Star shares. Computers stole the show back in the 90’s, as algorithms did in the 2000’s. Algorithmic trading (AT) and high-frequency trading (HFT) have come to set the rules of the trading world, and have essentially isolated human intervention to programming and maintenance duties.

In AT, computers collect market large volumes of data and built-in algorithm based programs send instructions to trading platforms autonomously. HFT, an extension of AT, is a slightly different phenomenon. HFT strategies are based on the same mechanics, but typically involve short holding periods and smaller profits per trade. High speed and low latency are key: algorithms read real-time market feeds and execute orders within sub-seconds to seize on minimal price trends before other operators. The following graph shows price movements of “E-mini” futures (ES) and SPDR ETFs (SPY) on the same underlying index (S&P 500). By reducing the frequency scale to milliseconds, it is clear how high-speed can profitably insinuate into flashing volatility.

(Retrieved from “The High-Frequency Trading Arm Race: Frequent Batch Auctions as a Market Design Response”)

When HFT was first introduced in the U.S. markets, completing an order took a matter of seconds. By 2010, execution time had been reduced to milliseconds and today it only takes nanoseconds (thousandths of microseconds) to implement most HFT strategies. Nevertheless, firms are constantly engaged in a technology arms race, competing for faster trade processing in terms of connectivity, data access and real-time analytics. High performance computing (HPC) is set to be the next ultimate weapon, and overclocking its silver bullet. Overclocking is a matter of allowing computer processors to exceed their official speed specifications, which enables high-frequency traders to run more operations simultaneously, shave milliseconds off trade speeds, reduce system latency and ingest greater amounts of data in less time. Time that can be saved also by placing computers next to an exchange’s matching engine (co-location), for the shorter the cables, the shorter the delay. Few serious investors however can afford such application of the speed-time-distance formula.

HFT is a light-speed journey, but what’s the cost of the ticket?

As Mike Lewis decried in Flash Boys, “the market is rigged” and high-frequency traders are those to blame. The problem is that traditional exchanges are an unfair playing field, where classic investors lose out to high-speed dealers. Indeed, information is money and profit belongs only to the fastest at acquiring and transmitting data. In two words: information edge. In one word: front-running.

Besides enhancing – literally – the tools of the trade for faster performance, HF firms can get the best deals by deceiving other market participants. It took some time for regulators to develop solid rules, but any market manipulation is in principle illegal today, though new ways to work the system around are always being thought of. “Spoofing”, in particular, is a tactic which involves placing large volumes of fake orders in an asset that get canceled at high speed before they are filled. When such large-scale phantom orders show up in the order book, they create the illusion of a greater buying or selling interest than there is in reality. Other traders react to this illusion, creating a false spike in demand/supply leading to price anomalies. The spoofer can then use previously acquired options to make significant profit.

Before the 2010 Flash Crash regulators used anti-fraud and anti-manipulation statutes to punish spoofing. When the Dodd-Frank Act was signed into law on July 21, 2010, the Commodity Futures Trading Commission explicitly focused on spoofing as a violation of U.S. federal law. Over the pond, the Market Abuse Directive (“MAD”, now regulation) addressed market manipulation practices, preparing the ground for MiFID II.

Market manipulation is nevertheless hard to detect and analyze, and it can be confused with legitimate sophisticated HFT practices. There is a consensus that, on average, HFT has increased liquidity in the markets and eased their fragmentation. On the other hand, the huge amount of HF deals intensifies volatility, which in turn widens bid-ask spreads and easily kills investors’ confidence in markets integrity. The magnitude of such a risk is likely to turn systemic given the strong inter-linkages between financial markets, and through high frequency any detrimental effect can spread in a blink, as seen during the Flash Crash of 2010 or last October’s GBP drop.

HFT is hard to tame. Once unleashed, algorithms could end up rampaging to Dukas’ Sorcerer’s Apprentice theme before a powerless Mickey Mouse. Knight Capital, a market maker, played Mickey Mouse for a 45-minute period in August 2012, losing $440 million. A newly-created trading algorithm made millions of faulty trades, buying stocks at the higher “ask” price and instantly selling them at the lower “bid” price. Unfortunately, the glitch was no fantasia.

In response to questionable trading practices and HFT drawbacks, New York witnessed the birth of IEX in 2013. Also known as Investors Exchange, it is a stock exchange that aims at leveling the playing field for traders by slightly delaying market pricing data to all customers through a so-called “speed bump”. Despite having opposed the launch of IEX, Nasdaq and other exchanges are now looking to introduce their own speed bumps, both to meet SEC requirements and to remain competitive. The NYSE was the first to slow down high-speed traders by introducing speed limitations on trading of small- and mid-cap company stocks.

The HFT marketplace has gotten increasingly crowded. Competition is stiff and the “arms race” to develop the best algorithms and build telecom freeways has driven up operational costs. The industry is showing signs of weakness also in terms of profit potential. According to a Financial Times report focusing on Virtu Financial Inc., it was found that this publicly quoted HFT firm’s trading income was closely related to the VIX index. A high VIX level meant high volatility, and that meant more income for Virtu, while lower volatility meant less money. In recent years, the markets have experienced low volatility, which contributed to declining profits of the fastest traders. In the face of lower volatility, the prospect of stricter regulations, “speed bumps”, and the fact that losses can quickly run in the millions represent the major obstacles to HFT’s future growth.

High-frequency trading thus seems to have reached its peak and is now settling at low speed, cooling down the markets for the next break-though and giving time to regulation to catch up.