Skip to content

Should You Use AI in Investing?

“Models beat human forecasters because they reliably and consistently apply the same criteria time after time. In almost every instance, it is the total reliability of application of the model that accounts for its superior performance. Models never vary. They are always consistent. They are never moody, never fight with their spouse, are never hung over from a night on the town, and never get bored. They don’t favor vivid, interesting stories over reams of statistical data. They never take anything personally. They don’t have egos. They’re not out to prove anything.”  James O’Shaughnessy –  What Works on Wall Street

 Artificial intelligence (AI) suddenly looks like it may revolutionize the financial industry (in addition to many other industries!). From robo-advisors to algorithmic trading, AI is creating new opportunities and challenges for investors and money managers. But can AI help you achieve your financial goals? In this blog post, I will explore how AI is being used by Wall Street today and whether you should use it in your investment process.

Wall Street is Already Using Ai and Machine Learning

One of the most successful hedge funds in history is the Medallion Fund, run by Renaissance Technologies. Founded in 1982 by Jim Simons, a former math professor and code breaker, the fund has generated an astonishing 62% annualized gross return and 37% annualized net return from 1988 to 2021.

The fund started as a quantitative trading fund utilizing statistical analysis and mathematical models. Over time the fund has hired an army of PhDs in fields like computer science, physics, mathematics, and statistics. The fund is very secretive but it is widely believed that they are on the cutting edge of machine learning, data science, and artificial intelligence. According to Greg Zuckerman’s book The Man Who Solved the Market, the Fund has utilized factors that influence the market that most investors are not even aware of: 

“Renaissance staffers deduced that there is even more that influences investments, including forces not readily apparent or sometimes even logical. By analyzing and estimating hundreds of financial metrics, social media feeds, barometers of online traffic, and pretty much anything that can be quantified and tested, they uncovered new factors, some borderline impossible for most to appreciate.” 

Renaissance Technologies is not the only hedge fund that uses AI and machine learning to inform its investment decisions. According to a survey by BarclayHedge in 2018, 56% of hedge fund respondents said they used AI in some capacity in their investment process. Given the rapid progress of AI technology since then, it is likely that this number has increased significantly.

Can the Average Investor Use Ai to Invest?

I think the average investor can use AI as a tool to help them make better decisions, much like they would use a financial advisor. However, I don’t think the average investor can use AI to beat the market consistently.

To the extent that AI can find inefficiencies or opportunities for outperformance in the market, those edges will be exploited by the most sophisticated players on Wall Street. Firms like Renaissance Technologies, Bridgewater, AQR, and Two Sigma have access to resources, data, and talent that the average investor cannot match.  As more and more firms adopt and implement AI, any advantage that early adopters had will diminish over time. The average investor will still be left with the average return of the stock market index, minus any fees or costs.

Thus asking AI to give you the best stocks to buy or the best sectors to invest in is unlikely to result in outperformance. Even if it could generate a list of stocks that had the potential to outperform, I think executing such a strategy would be difficult for the average retail investor. What would you do if those stocks underperformed during the first year you held them? Would you stick with them? Ask AI to generate a new portfolio? Almost every investing style goes through periods of underperformance. If you beat the market 70% of the time (which would make you one of the best investors in the world), that means you’re still underperforming 3 years out of 10. The problem with AI in this scenario is that you wouldn’t have enough confidence or conviction to stick with its recommendations. You wouldn’t know if you’re underperforming because AI was wrong or because you’re just going through a temporary rough patch.

Where AI Can Help You In Investing

I think AI’s best use case in the investing world is in financial planning and as an advisor/coach. I’ve seen it generate useful advice on general saving and investing tips. Investors’ biggest problem is their own behavioral biases. Human behavior works against us in investing. We tend to invest in things that have gone up and have good stories (which is often when we should sell!). We tend to sell when things have done poorly and headlines are negative (which is usually when we should buy!). According to the following chart from JP Morgan, the average investor has significantly underperformed benchmarks:

That underperformance has come from the average investor’s poor behavior. Chasing performance after it has happened and selling at a loss when things get tough.

You can use AI (or an advisor or Google or whatever works for you) to come up with a systematic approach to your investments. Ideally, this is something you can easily stick to regardless of market conditions. I asked AI to generate an investing strategy for a typical moderate risk tolerance investor and it came up with an answer that would likely serve many investors well:

A moderate portfolio allocation model could consist of 50% stocks, 40% bonds, and 10% cash. Within the stock allocation, 30% could be invested in U.S. stocks and 20% in international stocks. Within the bond allocation, 30% could be invested in U.S. bonds and 10% in international bonds.

This portfolio allocation model has an average annual return of 9.3% and a standard deviation of 10.8% from 1926 to 2021.

Of course, this is just an example and not a recommendation. Your optimal portfolio allocation may vary depending on your goals, risk tolerance, preferences, and behavior. You should also review and adjust your portfolio allocation periodically to reflect changes in your circumstances or the market conditions.

As long as an investor followed such an approach, they would earn reasonable returns over time. Of course, having AI help generate such an investing approach doesn’t guarantee that you’ll actually stick to it. Sometimes that’s one of the most valuable aspects of financial advisors – they act as gatekeepers between you and bad decisions. Want to panic and sell out during a market crash? You’ll have to pick up the phone and call your advisor, who should advise against it. Want to put all your money into crypto and NFTs? You’ll have to call your advisor and discuss.

But as AI continues to improve, I think it can definitely help with financial planning questions. What’s the best way to save for college? How much do I need to retire? What’s a safe portfolio withdrawal rate? I think these are the types of questions that AI will eventually be very good at answering.

Scott Caufield, CFA, CPA