Injecting artificial intelligence into the trading process could be just the thing to sooth equity investors' volatile nerves.
Concern over the impact of quantitative algorithms and high-frequency trading on equity market volatility — highlighted last week by comments from Treasury Secretary Steven Mnuchin — has led some sources to view AI methods like machine learning and robotic process automation as a possible counterweight to volatility. They cite the promise of AI as a way to automatically incorporate past behavior into current activity, potentially averting some major market swings.
"I expect the AI systems will likely 'smooth' the market, rather than aggravate it," said Roman Ginis, CEO of Imperative Execution Inc., New York, which launched the AI-assisted dark pool IntelligentCross earlier this year. Mr. Ginis said he expects AI to "reduce volatility in the same way as automatic flight control systems can keep the planes more stable during turbulence than when human pilots were involved."
Mr. Mnuchin in a Dec. 18 interview with Bloomberg News said he felt "market structure has led to a lot more volatility," citing high-frequency trading as a partial cause. HFT, using quantitative algorithms to enhance trading strategies, has been blamed for extreme market stress in instances including the Flash Crash of 2010, although volatility has been below its historical average in most of the years since the 2008-2009 financial crisis, according to the CBOE Volatility index.
And while sources contacted for this story disagreed with Mr. Mnuchin's assessment, they did say there is some promise that AI could reduce volatility.
Richard Johnson, vice president, market structure and technology at Greenwich Associates, Darien, Conn., said he doesn't see how the use of AI "would necessarily cause a systemic increase in volatility. It is still early days, and of course there are bound to be events where a rogue AI algo causes increased volatility. But the rate of these occurrences would not be greater than they are today with non-AI algos."
Also, he added, "The VIX is also known as the fear index … which speaks to the fact that volatility is often driven by emotions. AI algos do not have emotions (and) do not fear."
Mr. Ginis agreed with Mr. Johnson's assessment, adding that AI will reduce volatility because it can "make the adjustments (in trading) in small steps, as the deviations get registered and acted upon immediately, and the machines don't have emotional overreaction or underreaction to new information."
Quantitative trading that already relies heavily on AI "has improved liquidity conditions by reducing bid-ask spreads and microstructural volatility" — the volatility associated with price discovery, said Marcos Lopez de Prado, principal and head of machine learning at AQR Capital Management LLC, Greenwich, Conn.
"There are multiple benefits to algorithmic trading — greater liquidity, lower volatility, lower bid-ask spreads, lower latency and lower costs," Mr. Lopez de Prado said. "Additionally, algorithmic trading has empowered regulatory agencies and compliance officers by allowing them to go back years and analyze computer logs, looking for market manipulators, and machines follow the rules and they do not have conflicts of interest."
Said Aaron Hantman, CEO at broker-dealer Tourmaline Partners LLC, Stamford, Conn.: "AI or institutional investors, even retail investors, are always looking to learn from the past. There may be credibility in looking at the market in a more systemic way; that's what I mean by AI. If you can backtest past market movements, you can learn something from that."
Volatility could be diminished "if one assumes that AI will implement better arbitrage strategies," added Brian Conroy, retired president of Fidelity International, London. "The key word would be 'could.' "
What AI can do in equity trading, said Mr. Conroy, is "analyze patterns of trades, what effect they have on order placement, ranking trading venues, moves between venues in the market. It's an incredibly sophisticated world that is made easier to navigate through this technology."
One way AI could lessen volatility is by using it to trade passive equity strategies, said Mr. Ginis.
"For me, an AI system is a cycle of three components: measurement, learning and actuation," Mr. Ginis said. "A good candidate for it in trading is an index-following strategy that tries to stay close to the Standard & Poor's 500 via constantly measuring its portfolio composition vs. SPY (the SPDR S&P 500 exchange-traded fund), adding this data point into a learning function (e.g., a neural network), which can then trigger an actuation if the deviation from the index is too far." That action, he said, "would be a buy or sell order the strategy would place in the market to adjust its portfolio."
"Of course, human traders have been doing this cycle for years," Mr. Ginis added, "with the learning function being the wetware between their ears. We can now build machines that do this better."
Mr. Ginis said the IntelligentCross AI-powered dark pool further reduces volatility because it was "built to minimize market impact — i.e. the price move after each trade. With our trades moving the market less, there's less overreaction to price moves and less volatility."
Tourmaline Partners' Mr. Hantman thinks it's early days to say whether AI could have any impact, positive or negative, on volatility. "I think it's too early to tell," he said. "Volatility is usually triggered by things like political impacts and market forces. I'm not sure that volatility could be impacted by greater use of AI."
Greenwich's Mr. Johnson agreed with Mr. Hantman on the triggers of volatility and said Mr. Mnuchin's statement that high-frequency trading is a cause of volatility isn't supported by empirical data.
"HFT has been prevalent in equity markets for over a decade and much of this time has been characterized by low volatility," Mr. Johnson said. "Any astute market observer will tell you that the current market volatility is being driven by macroeconomic and geopolitical concerns, not the minutiae, or should I say 'Mnuchiae,' of market microstructure."
Money managers could be unwittingly getting the benefits of AI in their stock trades, Mr. Johnson added.
"If you're asking if the buy side is using algorithms powered by AI and provided by brokers, not just yet," Mr. Johnson said. "I think it could be that the buy side is using it and doesn't know it, where the broker is running it and doesn't tell them. At the end of the day, the buy side doesn't care how brokers do it, just that they get the execution and performance."