[Artificial intelligence (AI) was built to absorb, analyze and fashion recommendations out of huge piles of datasets at a fraction of the time that human efforts can achieve. Many were further designed with the ability to learn—called machine learning—to hopefully, perpetually, keep getting better and keep adjusting to changing dynamics in the environment they are working in. So, how did AI with machine learning function during the coronavirus pandemic?
To better explore and understand artificial intelligence technologies in a crisis scenario, we talked with Institute member David Aferiat, co-founder and managing principal of Trade-Ideas—an award winning A.I. and machine learning Fintech firm with a SaaS-based revenue model. We wanted to explore how AI helps advisors and traders make sense of investing when nothing is making sense.]
Bill Hortz: From your perspective, can you explain more about the nature of this current market disruption and trying to respond to this environment?
David Aferiat: The Covid-19 virus is an enigma. More is unknown than known. And initially medical professionals were treating it against known medical models—flu and pneumonia—only to discover that the virus was completely different from either. Markets reacted in much the same way—against the “known.” The simple fact is that we have never been here before.
That said, there are navigation tools—most importantly artificial intelligence—that are designed precisely to operate in unknown oceans. It is important to point out that these are navigation tools designed to help people make better decisions, rather than crystal balls that make decisions on their own or for others. These tools though provide three key underlying guide markers: information, predictability and confidence.
Hortz: How does artificial intelligence such as yours make that happen, to provide that navigation?
Aferiat: Artificial intelligence sources massive data sets such as market data, social media, news, and runs these sets overnight through dozens of multiple algorithm scenarios across tens of millions of simulated trades. The result: daily active trading strategies that enable portfolio managers and traders to make better decisions, based on likely outcomes.
As trading proceeds, risk on/risk off guardrails enable portfolio managers and traders to know where and how their strategy is executing intraday. When the guardrails are hit, then traders can stay/go/modify their approaches in response.
Hortz: Did your AI system you call “Holly” surprise you in anyway in its functioning, analysis or stock picks during the recent coronavirus meltdown?
Aferiat: Holly is a highly robust system and able to immediately recognize and respond to events. So, actually, there were no surprises. Holly does not have emotions around events—it merely responds to what the data is telling it. In terms of functioning, we were “already ready” for any sharp market actions.
Hortz: What was Holly’s greatest value-add through the chaos in the markets?
Aferiat: The greatest value-add is making sense of nonsense. What Holly was able to deliver is information that led to rational strategies that could provide some confidence and predictability, all tracked in real time. Helping investors make sensible decisions in the middle of chaos, fraught by immeasurable unknowns, is a tremendous benefit.
Hortz: Were there times you had to adjust or tweak the AI system and/or the algorithms used by Holly during this timeframe?
Aferiat: Not really, Holly is always market ready, and the modifications and building of new approaches is continuous—it is not predicated on events—no matter how seemingly catastrophic. For our teams, it was business as usual, technically and culturally. Adjustments and tweaking are constant, not anomalous. Holly’s algorithms also teach themselves at all times.