AI is no savior when markets get tough … but it can help, says Nickel Digital chief

When markets get tough, as crypto did at the end of January, investment companies need all the help they can get to make the right decisions, fast. No surprise, then, that many are turning to AI, the shiniest new weapon in the arsenal, to analyze and suggest ways of minimizing losses and even making a profit.

Almost all (96%) executives at a surveyed group of trading firms that collectively manage around $14 trillion in assets said AI is already playing a major role in core investment processes, according to research recently carried out by Nickel Digital Asset Management. But it’s not enough, a human hand is still needed, said Anatoly Crachilov, founding partner and CEO of the firm.

AI is transforming quantitative trading just as it is almost every other industry and human endeavor. Going beyond the large language models (LLMs) that seem to have permeated so much of day-to-day life, there are also machine learning and predictive AI approaches that analyze historical data to forecast what’s coming next. They’re weak, however, at identifying incorrect information that can lead to erroneous conclusions and poor decision-making.

“It’s a very tough market. AI will not save you; it’s not a savior,” Crachilov said in an interview.

Despite the slump in crypto prices that engulfed the market at the end of last month, London-based Nickel, which runs a multimanager platform allocating to more than 80 teams, remains positive for the year. “Perhaps an achievement in its own right,” Crachilov said.

The crossover between crypto trading and AI is becoming most advanced in areas like risk management. While AI might still struggle to outperform high-speed sniper bots targeting the latest low-liquidity crypto tokens, for example, a sweet spot is where sentiment and data-driven models can learn how to manage risk.

Each manager attached to Nickel operates within a well-defined risk framework that includes maximum drawdown limits at times of increased volatility. Sometimes human intervention is needed and an “old school” approach, Crachilov explained, as opposed to relying on data-driven, machine-learned automation.

“If the market goes into distress, like it went on a few occasions in recent memory, sometimes you have to exercise discipline and stop those managers who break [max drawdown] limits, whether it’s AI driving their strategy or not,” Crachilov said. “Ultimately, there is a hard stop on how much pain we would allow in the portfolio.”

Questions about how much human involvement there should be in AI-driven trading strategies, or the manner in which a human override is triggered, were too technical and nuanced for Nickel’s relatively high-level survey of managers, Crachilov said.

He said Nickel operates “a military-style operation,” where a rich data flow collects over 100 million data points from the underlying book every 24 hours. “While this part is very well informed, it still requires human involvement. And we’re still in conversation with managers, even in the middle of the night,” Crachilov said.

The natural evolution toward being fully automated still has to account for the possibility of erroneous or incomplete data feeds from places like crypto exchanges, according to Crachilov.

For example, a human would realize that data indicating a certain position is down 100% was probably the result of something being wrong with a data feed, he said. But an automated AI system might mechanically enforce a limit when it wasn’t required.

“You need a human overlay. The whole crypto ecosystem is still very fragile. And some of the exchanges may go into timeout for 15 minutes, or see wrong data, or produce patches of bad data, which may inadvertently force the system to shut some of the managers for no good reason,” Crachilov said.

It really comes down to the firm’s risk-management philosophy, which is to remove a single point of failure from any point in the process, said Nickel’s head of investor relations Charles Adams.

“If there was one autonomous agent which is monitoring the whole portfolio, let’s say something goes wrong with it, the risks could be potentially catastrophic,” he said. “The whole point is that we have this very well diversified fund split between over 80 managers today across hundreds, if not thousands of sub accounts on exchanges, and removing that single point of failure is very important to us.”