Benzinga Interview with Alex Lu — 6 Experts Discuss How AI Will Change The Future of Wall Street (Part 1)
This is part 1 of a roundtable Benzinga conducted on Wall Street’s adoption of artificial intelligence. Click here for part 2.
Artificial Intelligence is slowly creeping more and more into our everyday lives and jobs. Wall Street is no different.
What makes the intersection of AI and the markets particularly interesting however is the inherent contradiction. Wall Street has long been a place that’s thrived on exploiting inefficiencies. AI, on the other hand, exists to make things more efficient.
With that in mind, Benzinga individually asked a panel of people specifically working at this intersection to give us their thoughts on how it will play out over email. The panelists are:
- Alex Lu, co-founder & CEO of Kavout, an AI firm that provides quantamental analytics
- Chris Natividad, CIO of Equbot, provider of the AI Powered Equity ETF AIEQ
- David Aferiat, co-founder & managing director of Trade Ideas, an AI-powered discovery engine
- Dr. Hossein Kazemi, Ph.D. and Senior Advisor to the Chartered Alternative Investment Analyst (CAIA) Association, an internationally recognized authority on alternative investment education
- Lane Mendelsohn, vice president of VantagePoint, an AI market forecasting software
- Nikhil Dhingra, co-founder & CEO of Tradagon, a fintech company that uses AI in its signals
Some answers have been edited for clarity.
1) On a scale of 1-10, how would rate Wall Street’s adoption of AI (with 10 being “We’ve reached peak AI”) and why do you say that?
Alex Lu, Kavout: Between 3-4. In the last decade, we have seen enormous success of AI in internet products such as Google’s search, Amazon’s personalized recommendation, and most recently AlphaGo. When we started Kavout in 2015, there were only a few big hedge funds investing in AI and machine learning to find an edge, and most Wall Street firms were either suspicious of AI’s value or didn’t know how to leverage and integrate AI in their existing business.
Chris Natividad, Equbot: 2. Banks and asset managers are only scratching the surface with AI. There are commercial instances of AI protecting consumers in the form of fraud detection, and emerging investment products that directly utilize the evolving technology. We see AI as the way forward as businesses want more information when making strategic decisions. If you can agree that the amount of digital information will only continue to grow, then AI will become a critical tool across industries.
David Aferiat, Trade Ideas: Wall Street Institutions: 4. Only the highest capitalized are building competencies. They are the biggest names of course, but only represent a minority (in number).
Retail Wall Street: 2. No one is implementing AI technology in the service of decision making.
See this stat for the overall economy.
Low-cost computing power became available about a decade ago but only during the last few years has the large amount of data needed to train AI has become available at a reasonable cost. The emergence of AI in asset management, along with several other trends involving digitization, is exactly why my colleagues and I at CAIA recently launched the Chartered Alternative Data Analyst (CADA) Institute.
Lane Mendelsohn, VantagePoint: I think we’re probably at about a 3. I think a lot of people at this point are hearing more and more about artificial intelligence, becoming a little bit more familiar with what it is, but being familiar with something and actually being able to implement something are two very different things.
We saw back in the early 1990’s, neural networks and artificial intelligence were becoming a hot topic. A lot of individuals were starting to experiment with them, a lot of companies were trying to utilize them, and it’s a tool and it really depends on how the tool is being implemented. I think probably in the next five years, we’ll get to a five. I think it’s going to be a while before we get to a 10 on the scale of adoption.
Nikhil Dhingra, Tradagon: If we are talking about true adoption of true AI, and not just adopting the buzzword (a la the term “Blockchain”), then I would say maybe a 4. There is still a lot of exploring, iterating to be done—as well as seeing the longer-term implications—before we can say that AI has been adopted and put into practice. I think lots of folks are just touching the tip of the arrow and exploring it carefully, but still in the due diligence phase.
2) What is currently the biggest challenge for institutional investors in how they use, or try to use, AI?
Alex Lu, Kavout: As I learned from my past career at Google GOOG GOOGL, Microsoft Inc MSFT, and Baidu (Baidu Inc (ADR) BIDU) of building large-scale AI and machine learning systems, the application of AI depends on three key elements: comprehensive data, superb computation infrastructure, and algorithms that suits the problems to solve.
Institutional investors need to be strong in all three to achieve great results and find the team to implement these elements. So the biggest challenge is to find the right group of talent who not only have extensive experiences in building AI systems, but also understand Wall Street.
Chris Natividad, Equbot: Unfamiliarity. There are structural issues within the majority of institutional investor processes that limit the adoption of AI. Institutions typically require long track records or backtesting results to gain comfort with new investment strategies/technology. We see inherent problems with this requirement.
Bloomberg recently noted, “An estimated 90 percent of all data in existence today were created in the past two years.” Getting comfortable with unfamiliar datasets and machine learning investment theories will most certainly be a major challenge for institutions as they move toward AI investment solutions.
David Aferiat, Trade Ideas: Institutional investors are faced with challenges to their existing systems and the need to apply the new tech of AI, quant modeling, and Big Data to their decisions.
Even outside larger IT projects, there’s less of a reason other than tradition to want a Bloomberg terminal at a cost of $24k/year when AI-generated analytics—with a performance record of capturing alpha—can be 1/4 the cost. This makes AI the new tables stakes. You won’t necessarily win using it, but you will definitely lose without a plan for accessing it.
Dr. Hossein Kazemi, CAIA: The biggest issue is the aversion of asset owners to “black box” strategies. Many consider AI as another version of algorithmic trading (to some extent this is true), and algorithmic strategies have not performed well in the past. While investors are comfortable with having AI playing an important role in many parts of their lives, they seem to prefer human judgment to AI when it comes to the investment process.
Another potential obstacle is that an AI approach to trading requires a whole new organization structure for trading operations. While it is desirable to put discretionary traders in silos to reduce group thinking and correlations among traders, this approach will backfire when applied to AI trading, which requires a team effort to test thousands of strategies in order to pick the best. Increased diversification is achieved through the training and test of AI models rather than putting traders in silos.
Lane Mendelsohn, VantagePoint: I think for a lot of institutional investors, it’s going to be very difficult for them to figure out exactly 1) what do they want the AI to do for them? And then I think the bigger challenge is going to be how do they implement it? Because you can’t just wake up one day and decide you want to implement artificial intelligence.
For our company, we’ve been developing artificial intelligence technology since the 1980’s, and we still haven’t perfected it. Now, we’ve gotten to a point where we can successfully and consistently generate forecasts that are in the 86 percent range, but it’s an ongoing process and the markets are constantly changing. It’s something that has to be an ongoing effort, which requires a lot of talent but also a lot of financial resources in order to refine and continue to perfect that technology.
Nikhil Dhingra, Tradagon: I think it is figuring out the right tools and human capital to put into place to leverage existing data and new data and make sense of it, and then instrument the appropriate infrastructure to act on it and actually get value out of it.
3) What is currently the biggest challenge for retail investors in how they use, or try to use, AI?
Alex Lu, Kavout: Compared to institutional investors, most retail investors have quite limited knowledge about AI, and this lack of understanding is hindering the adoption of AI. Most retail investors either take a sci-fi-like view to expect AI to make all the right calls for them, or don’t trust AI at all and resist its use in any scenario. I think education and having the right expectation would be the biggest challenge in retail space.
Chris Natividad, Equbot: Education and trust. A recently conducted EquBot survey revealed over 40 percent of people are still skeptical of AI. Further, only a very small percentage of the population is aware of the processing power, innovative potential, and strategic advantage of using AI.
The level of skepticism and overall lack of awareness is indicative of a tremendous opportunity. Retail investors who are further up the knowledge curve will also need to trust in their understanding of long-term machine learning investment concepts. AI investment algorithms continually evolve, so experiencing different market conditions will only make these investment platforms stronger over time. We believe early adopter retail investors willing to invest in AI investment solutions will benefit the most, as additional capital will serve as tailwinds when the technology successes become more widely known.
David Aferiat, Trade Ideas: Biggest challenge is throttling their expectations of what AI is and how to use it in the bigger picture—not just looking at it trade by trade and trying to mimic.
It’s the same challenge new traders have learning from a pro trader. You need to learn, and not just follow. Another challenge is closely related: adoption is harder when you can’t understand the details because they’re too technical or opaque. So retail investors end up having to convince themselves without really being able to understand the details. Working on simple, but elegant ways to give AI-based assessments of the markets from 30k feet with the ability to drill down is important.
Lane Mendelsohn, VantagePoint: I think the challenge for retail investors is everything that relates to institutional investors, but multiplied by 1,000x. For retail investors, they don’t have the financial resources that institutional investors have, and they don’t have the time to spend on it. Everything that I mentioned as it relates to institutional investors and the challenges that they have, it’s all magnified when it’s related to retail investors and traders.
Nikhil Dhingra, Tradagon: I’m not sure AI tools have been sufficiently democratized for the retail investor to really have control over how they apply it to their own trading. Right now, it seems like they can be part of a bigger AI play, where say, an AI platform is crowdsourcing trades and then running AI over that data to achieve a desired goal. But in that use case, the retail investor did not really play a direct role in it other than making the trades.
Send us a Message
Contact us
Contact us today to learn more about Kavout's products or services.
