Aisan stocks were up in early trade, with Tokyo jumping more than two percent – Copyright AFP Kazuhiro NOGI
Artificial intelligence has begun to shape investment decision making, with investors drawing upon the opportunities presented by algorithms to gain predictive data insights. Technology plays a vital role in the world of investing as companies look for ways to integrate artificial intelligence capabilities into the research and selection process. With the power of machine learning, the capacity to evolve, adapt and identify patterns, investment managers can harness this technology to potentially enhance performance and improve efficiency.
As an example, in January 2022, SoftBank announced it was investing $146 million into AI investment solutions provider, Qraft Technologies to accelerate the development of artificial intelligence in asset management. As the industry has been late adopters of technological innovation, Softbank’s investment could pave the way for solutions in a future market environment where achieving alpha may be challenging in the years ahead.
Digital Journal took the opportunity to speak with Robert Nestor, Formerly Head of BlackRock iShares Factor ETFs and President of Direxion ETFs. Nestor is currently the U.S. CEO at Qraft Technologies. Qraft is an artificial intelligence enabling invest-tech company that develops and operates deep learning-based algorithms for asset management and wealth advisory firms.
Digital Journal: How is artificial intelligence improving upon traditional fundamental analysis in security selection?
Robert Nestor: Some people believe that integrating AI into investment management is about programming the robots to make all the investment decisions. However, real practitioners in this space understand that human intuition is very powerful, so attempting to completely remove humans from the process makes little sense. While traditional fundamental analysis can build from some facets of AI techniques, it is still inherently a human dependent decision-making process. However, humans, and machines for that matter, have limitations.
What makes us innately different from machines is our emotion, which most people would say is a positive thing but generally not a value add to investing. By integrating AI technology, the emotional aspect can be largely eliminated. More importantly, the human brain, while powerful in its decision-making, can’t compete with the scale, scope and speed potential of machines. AI in investment management is inherently about combining the best of both worlds, meaning we use AI techniques to scale analysis and decision processes, but always with some manner of human oversight.
DJ: What are the types of data that AI algorithms process in the investment decision making process?
Nestor: Data is the energy source of any AI process, and AI investment processes are no different in that regard. The more the better, and, in theory, there is no limit to what could be relevant, that is, what could prove predictive of investment returns. In practice, typically only a very small subset of data is generally of value. More than 99 percent of data is just noise with no predictive power, but you can’t know that until it is analyzed. Data can span the gamut from traditional stock specific data (price, earnings, etc.) to less traditional data such as employee growth or patent levels. But it is not just raw data that could be relevant, derived data such as different (often unconsidered) data relationships, and their correlation and intuition to secure returns. Those potential relationships can run into the hundreds of millions, potentially even billions, very quickly. There is also macroeconomic data and so called unstructured data such as web traffic, image recognition, etc. Much of the latter is still in its infancy in AI driven investment processes but is likely to rapidly become more integrated over time.
DJ: Why has traditional asset management been slow to adopt artificial intelligence in decision making?
Nestor: I often say the most powerful force in nature is inertia. It is a very different approach than what’s standard in the profession today, and most people don’t want to change what they know and feels comfortable. I think there are four primary reasons the industry is slow to adopt AI:
- There is still a widespread, fundamental lack of understanding of AI. While that clearly hasn’t impeded widespread adoption in other industries, I’d venture that most people feel it is more important to understand how AI will help manage money than it is to understand how AI dictates driving directions in Google Maps. It’s hard to disagree with that view.
- The early AI practitioners went to other industries such as healthcare, marketing, etc.
- There is a significant cost to building AI driven investment platforms and capabilities from the ground up.
- There is a fear of decision-makers being replaced. The objective is not to remove humans from the process, but you do need less people in the process.
That said, I think there are clear tailwinds to AI adoption including:
- There is likely going to be a lower return investment environment going forward compared to the last decade.
- There will be continued fee/margin pressure in asset management driving further push for scale.
- The increasing influence of ESG in investment decision-making, which is a massive data cleansing challenge.
- There are increasingly complicated investment challenges as a wider population enters retirement and draw down from their portfolio.
DJ: How do you effectively build out artificial intelligence capabilities as an investment manager?
Nestor: It is not easy. It starts with the people, which may not sound intuitive for what is largely a technologically driven process. But you must have people who profoundly understand AI processes and techniques such as machine learning, deep learning, attention models, etc. along with deep investment acumen to apply it correctly. You can’t be effective without pronounced experience in both areas. You also need time and a highly disciplined approach to build, research, hone and evolve the AI models. It can take years in order to be confident in the efficacy of what has been built. Qraft started in 2016, and few platforms have our experience set. You need to feed the beast — data is the lifeblood of the AI process, and wide data sets can be expensive to acquire and cleanse.
DJ: How is artificial intelligence impacting investment fees?
Nestor: Candidly, I don’t think it has yet, as its adoption is not nearly widespread enough yet. But I think it inevitably will, and could possibly have a profound impact. A process that is largely technology driven will simply offer more scale and efficiency over time, reducing the cost to invest for alpha goals. It will take at least a few years to play out, but AI will inevitably play a meaningful role in investment decision-making processes, and that will drive down cost for all investors.
DJ: How is AI a positive force for individuals interested in ESG investing?
Nestor: Obviously, interest in ESG issues is fundamentally impacting investment decision making. More and more investors are seeking to better the world through their use of capital, on their terms.
However, there are major practical challenges. First, reliable longitudinal data to define companies’ ESG attributes is still in short supply, but is methodically being uncovered, and there is some really interesting work being done to accelerate it by firms such as EMAlpha.
Second, not everyone agrees exactly how and for which cause(s) they want to allocate capital. The desire for more customized solutions in this space is likely to grow enormously. Data and scaled delivery are the hallmarks of AI, and techniques being utilized here will only accelerate the access points for more customized solutions and confidence in the investment decisions driven by ESG considerations.