Will The Next Warren Buffett Be A Bot?

Google+ Pinterest LinkedIn Tumblr +

This past year alone, Artificial Intelligence programs have bested world masters in the ancient Chinese strategy game, Go; have learned how to read mammograms and predict cardiovascular risk through eye scans with equal or higher accuracy than humans; and now AI has its sights set on the world of finance.

A NewtonX expert currently leading the Digital Innovation Practice at a large investment bank highlights that AI is already an integral aspect of many financial institutions. It’s used, not only in high frequency trading, but also in Exchange Traded Funds (ETFs), in various capacities in large fund management institutions (including Vanguard and Fidelity), and in sentiment analysis. In some companies, it has taken over the role of the research analyst by collecting and analyzing market signals in real time to make rapid decisions. It can rapidly digest reports including regulatory filings, quarterly releases, news articles, and even social media postings, and then make predictions based on how these signals have impacted the market previously.

Algorithms that interpret data have been used in finance for years, from simple regressions to code-based complex equations or even Monte-Carlo simulations. So what has changed with AI? We asked the former Chief Data Scientist of a large London hedge fund who leads the NewtonX panel on this topic — ”The difference between [traditional algorithms]and AI is the ability to process large sets of unstructured information in real time, without knowing a priori what it is looking at, and to update its problem solving methods as it processes the data”. In other words, it can adapt to unexpected data, can see patterns that humans have not trained it to recognize, and can find solutions that would be counter-intuitive to scientists. In a world where data is multiplying and financial markets are exponentially data-driven, it seems like a perfect storm is forming for AI. And at least two companies are exploiting this ability to offer ETFs and other funds that are completely powered by AI.

NewtonX conducted a micro-survey of ten senior executives in the financial services (fund managers and heads of trading at large banks) for in-depth qualitative assessments of AI uses in ETFs and other fund management applications — their insights inform the conclusions drawn in this article.

Democratizing Finance: How AI Includes the Common Man, but Excludes the Common Fund

IBM’s Watson powers the foremost AI ETF, called EquBot. We have interviewed as part of our research one of the data scientists at the origin of the project. EquBot claims that its use of AI both eliminates human bias and eliminates “wealth-based discrimination throughout financial markets”. It’s true that unlike traditional hedge funds, to buy into EquBot you need only to pay as little as the price of one share — and conceivably you are purchasing investment knowledge that surpasses that of a highly successful hedge fund manager.

That said, NewtonX experts stressed that this technology, while made available to the common man, is not available to the common fund.

“It’s a pretty thought, that AI could democratize access to professional investor knowledge,” said a fund manager at Fidelity. “But the reality is that larger funds will be the only ones that can afford the necessary data processing tools to effectively implement AI and interpret its output correctly, which will result in a small pool of investment giants.”

NewtonX experts also expressed skepticism toward the purported autonomy of EquBot. “AI is extremely useful for all of the tasks that IBM says EquBot performs, but I have yet to see an AI tool that can use the predictions and patterns it finds to make better investment decisions than a human,” said a Member of the NewtonX panel..

Indeed, between its conception in October and the end of the year, the EquBot E.T.F. rose 3.1 percent, while by comparison the Standard & Poor’s 500-stock index rose 5.1 percent. This could be attributed to normal variability in returns, but could also be a reflection of the bot’s impressive, but not superhuman capabilities.

How a Rational Machine Deals With an Irrational Market

There’s a reason that playing the market is high risk, high reward: the market is unpredictable. And the one thing that AI struggles with the most is unknown situations. As we recently explained in an article on AI in strategy games, the technology excels in environments with rules, parameters, and a wealth of data. Even situations with a lot of variation are well handled, as long as they are somehow reflective of similar variations that have happened in the past. But when the truly unexpected occurs, AI reacts much, much more poorly than humans do.

Experts agreed that AI is best used as a subsidiary to human decision making. “Most of us use some form of AI to report on patterns, trends, and consumer sentiment,” said a NewtonX expert leading the AI research department for a large financial services firm. “But it can misinterpret what it reads. The benefit of AI is that you can look at patterns and trends and then dive into individual examples of what led the computer to its conclusion. This expedites the process to a decision.”

Because the market is irrational, it follows that you would want to use a combination of predictive analytics and human intuition. When everything goes according to precedent, AI can accurately pinpoint opportunities and identify mispriced investments. When unexpected things happen, though, you don’t want an AI responding – you want a human.

Perhaps AI will become better at responding to the unknown, but as the former VP of an international European bank put it, “The market is always getting increasingly complex, so even if AI gets a handle on it at any given moment, it’s likely to lose that grip because of millions of tiny factors that can shift the market in any direction.”

The Future of ETFs is Reliant on AI, but not Comprised of It

Incorporating AI into finance is appealing to many people because it removes the mystique of finance. It’s no coincidence that EquBot’s mission is to democratize access to smart investing — and other similar tools have followed suit. TokenAI, for instance which is an AI-powered investment tool, claims to bring “Wall Street level tools to the crypto community and to the masses.”

In much the same way that automating the process of contesting a parking ticket benefited lower income populations, AI powered investment tools will too. That said, the higher end ETFs and other types of funds will include a combination of human and machine driven insights to make the best financial choices for the fund possible.

“Sure, AI conquered Go this year, but financial markets are much, much more complicated than a strategy game, and the goal isn’t always as clear,” said our NewtonX panel lead. “AI is definitely an increasingly important tool, but not an end-to-end solution.”

The data and insights in this article are sourced from NewtonX experts. For the purposes of this blog, we keep our experts anonymous and ensure that no confidential data or information has been disclosed. Experts are a mix of industry consultants and previous employees of the company(s) referenced.


About Author

Germain Chastel is the CEO and Founder of NewtonX.

Comments are closed.