Skip to main content

5 Big Questions Regarding the Future of AI in Finance

5 Big Questions Regarding the Future of AI in Finance

Profit & Loss will be hosting a panel discussion, titled AI: Regular Quants with a Bigger Bazooka?” at our annual Forex Network New York conference on May 24th. For more details about the conference agenda and speakers, please click here.  To register for the event, which is free for buy side firms to attend, please click here.

By Gaurav Chakravorty, Co-Founder, qplum

There is a lot of conversation around Artificial Intelligence (AI) among different participants in the institutional investing pyramid.

Investors are wondering if AI can get higher returns by extracting unexplored alphas or if it can reduce costs, and investment professionals are wondering how machine learning and AI will impact their businesses.

Right now there is a lot of exuberance, optimism, skepticism and fear around AI and the impact that it will have on financial markets. Here I explore five key questions that are important to ask regarding this technology and its role in finance.


1. What is Artificial Intelligence?

AI can mean different things to different people. Let’s first talk about the experience of AI instead of the minutiae or the mathematical aspects.

For an institutional investor, AI is perhaps nothing more than another manager that they are allocating capital to. The experience of investing with an AI-driven asset manager will probably differ in these aspects:

  • Less costs. AI-driven asset managers can achieve a lot more cost efficiency with technology.
  • Less key-man risk.
  • Fewer biases hurting performance.
  • More focus on diversity of strategies. Most AI-based managers will be doing a multitude of trades.

For the portfolio manager, deep learning – the heart of modern AI – is a novel approach. The novelty lies in the fact that the technology tries to learn a hierarchical structure of concepts to understand what is going on in markets, before it starts predicting what will happen. In data science parlance, we spend more time learning the right representation of the data than in the supervised learning step. As Geoffrey Hinton, says in his paper in Nature on Deep Learning, truly novel breakthroughs in deep learning will come from unsupervised learning.

The novelty is replicating memory in addition to adaptability in the trading machine. Recurrent Neural Networks and Long Short Term Memory (LSTMs) networks have produced groundbreaking results in understanding sequences like speech and text. The hope is to apply the same methods to structured and unstructured forms of data, in investment management.

2. How is AI Currently Affecting Asset Management?

Every participant is going to get affected by AI. Machine learning methods have the potential to bring greater efficiency, higher returns and greater transparency to those who are using it. I think most people will agree with higher returns and greater efficiency of costs, but why transparency? With more data, it is harder for people to hide behind performance numbers.

The good: We are seeing a tremendous amount of interest and willingness to learn from allocators and their investors in AI/machine learning methods in investing.

The bad: There is a tendency among old-school allocators to lump the new quant with the old quant strategies and to just look at AI as “quant”.

The unknown: There is some skepticism about the new technology not being validated, and there are still many unanswered questions about the use of technology.

While there is a lot of excitement, there are more questions than answers right now. People are wondering if it’s real or if it’s hype. Portfolio managers and institutional allocators are devoting a lot of time to learn these new technologies.

There is fear, skepticism and excitement. Some old school consultants are afraid, investors are interested, portfolio managers feel fear and opportunity depending on their core strengths.

3. Does the Emergence of Artificial Intelligence Mark the End of the Human Discretionary Trader?

We would not call it an end but an evolution. The work of the portfolio manager will change drastically.

There is a limitation to machine learning. Machines can only analyse data that they have seen before. In real life there are often unexpected events that need to be handled. We think this transformation is perhaps similar to how we still need a pilot in a cockpit. The job of the pilot, however, has changed a lot. In many ways, they need to understand the strengths and limitations of the machine and how it exactly works. Technology can only accelerate and facilitate the decision-making process.

4. How Should Institutional Investors Adapt to AI?

Institutional capital allocators and due diligence professionals have to be aware of new technologies and what questions to ask to separate the best from the rest. Chasing returns is not the most optimal strategy.

Michael Weinberg, the CIO at MOV37, said it best at a recent roundtable he and I hosted:

"Where we are now is far more revolutionary for asset management than simple AI-driven quant. What we have now in investment management is the equivalent of AlphaGo Zero, but for investing. Just like AlphaGo Zero generates its own new ideas, so do Autonomous Learning Investment Strategies (ALIS). We think that this disruption to the asset management industry will come from outside discretionary and even quantitative managers. These so-called ‘third wave, ALIS managers’ exploit the confluence of data, data science, machine learning and cheap computing: their brains are wired differently. They are often hackers and computer gamers with a healthy disrespect for convention."

He also mentioned the fact that they spend a lot of time in trying to understand the strategy and the manager and competency in deep learning and AI. It is not just about chasing the last three years’ returns.

5. Will A.I. be “the” Strategy in Future or Just a Part of the Alternatives Pie?

This is perhaps the biggest question that the asset management industry is facing. In an old school fixed “asset allocation pie” method, CIOs, fund of funds and family offices need to bucket AI into some label.

Right now they are calling AI as quant and quant as a part of the market-neutral or alternative-investing section. However, we think CIOs who have a foresight of AI being the only strategy can achieve a lot more.

Right now quant-investing is about 1% of all asset management, under 7% of active management and under 13% of assets managed by hedge funds.

However, AI​ ​is​ ​not​ ​“just​ ​a​ ​strategy”​ to be delegated to a narrow corner in the dark confines of an alternatives black box, but rather a core part of a CIO’s investment philosophy.

In the technology world, Apple does not limit its use of AI to just Siri, but instead is extending its use across all aspects of its business. In a similar fashion, we can see CIOs extending the application of AI to all aspects of investing, and using AI to improve decision making around asset allocation, how to respond to macro events, and whether to allocate active or passive.

In my view, it is not unreasonable to expect in the near future that about a third of total assets would be managed or allocated using some form of AI.