The pandemic-driven disruption to the global economy has not slowed the pace of innovation across the fintech space, nor has there been any reduction of interest in leveraging the latest advances in data science to drive new growth opportunities, especially when it comes to asset management.
But terms such as “data science,” “artificial intelligence” and “machine learning” are frequently used but no so frequently defined in terms understandable to laymen.
That’s why Digital Wealth News connected recently with data sciences leader Homa Karimabadi about his role as Chief Scientist of AlphaTrAI, which seeks to transform asset management with a data science-driven approach that eliminates errors that can be caused by human bias and emotions.
In addition to holding a Ph.D. in Plasma Astrophysics from the University of Maryland, Dr. Karimabadi is in good company with the rest of the team at AlphaTrAI, where CEO Andreas Roell brings extensive experience across AI-driven start-ups, with former LPL Financial President Bill Dwyer serving as one of the company’s Board members as well as an investor in AlphaTrAI.
- How did you become interested in data science in the first place, and when did you become aware of the potential value of data science within asset management specifically?
As a physicist, my interest has been to understand and develop forecasting models for complex systems. Financial markets are the ultimate complex system and have many similarities to natural systems such as tending towards equilibrium but being inherently unstable and exhibiting periods of explosive growth. Further, the financial market is similar to quantum systems because participation in the market affects its behavior.
Early on, I recognized the importance of computations and intelligent algorithms to push beyond the limitations of the existing models. I introduced and led the development of intelligent algorithms in particle simulations and knowledge discovery in space weather modeling, which led to uncovering of phenomena and making predictions that could not be achieved with previous less accurate fluid approaches.
Along these same lines, noticing deficiencies of the existing models in financial markets, I saw an opportunity for significant advancement through the development and deployment of state-of-the-art intelligent algorithms to surpass the performance of lower fidelity models.
- Why do you believe the large mutual fund and ETF asset management firms have been slow to adopt the latest advances in data science into their businesses, and what makes AlphaTrAI better positioned to succeed in this emerging area of asset management?
Much like healthcare, the financial market is a heavily regulated industry with engrained legacy systems and practices. The inertia of established management firms is incongruent with the agility required to adapt and adjust to the fast pace of advancements in data science. A disruption in the industry is underway and is gaining momentum. As in other industries, new companies such as AlphaTrAI, unburdened with legacy practices, are forging a new path.
Data science is an interdisciplinary field and the Venn diagram of the disciplines that it encompasses, from mathematics and statistics to machine learning, is continually expanding. Deep learning is the latest addition to the field of data science. The guiding principle behind our algorithms at AlphaTrAI is that solving the market requires a holistic approach with deep learning as one of the disciplines rather than a be-all and end-all solution. This is reflected in AlphaTrAI’s multi-disciplinary team of physicists, AI experts, mathematicians, and statisticians.
- For all the talk of the power and potential of data science, there are those who say that data science, artificial intelligence and machine learning are largely still buzzwords insofar as asset management is concerned, and that the actual value of data science has yet to be proven. What do you say to these naysayers?
This sentiment and lack of widespread adoption of data science in financial markets are hallmarks of a legacy industry where poor performance has not only been accepted but rewarded, an industry that has been reactive rather than proactive to change. And it is precisely these characteristics that have made financial markets ripe for disruption.
The reality is that most human asset managers have underperformed the market and have struggled to provide consistent returns. In contrast, the most successful asset management fund in the industry has been the Medallion Fund where it is completely algorithm driven. We have modeled ourselves after the Medallion Fund, by hiring scientists instead of MBAs to bring a multi-disciplinary approach to solving the market.
We have first-hand experience with how innovative blending of techniques leads to superhuman trading strategies. We have seen how our models self-learn complex strategies from data, uncover complex patterns and relations in the data, consistently make successful predictions, and adaptively adjust their risk and reward balance.
- Name the top four things that data science applied to asset management can accomplish more effectively than experienced, human asset managers.
Financial market is akin to a multiplayer game where the rules of the game and even the characteristics of the players are constantly changing. Technological advances and rise of algorithmic trading are accelerating the pace of change and elevating the complexity of the game. As such, asset managers that have been slow to adopt technology and data science are becoming increasingly outmatched.
Data science driven asset management offers consistency, debiased decision making, reliable predictive capability and speed of turning information into action.
- Are there areas of asset management where data science cannot replace human professionals?
Data science is an enabling algorithm development framework and is meant to augment human cognitive power, not to replace it. After all, human data scientists are designing the algorithms and pushing the state-of-the-art in AI. Until we get to the point of strong AI and fully automated machine learning, human-in-the-loop is the optimal solution.
The adoption of data science in asset management changes the composition of the research team from MBA trained professionals to data scientists. Data science is also making its way into other aspects of asset management including compliance and marketing.
