Can Investors Outperform Artificial Intelligence?

September, 2023

Active investors such as hedge fund managers continually seek an informational edge for market trading. That includes using artificial intelligence (AI) to retrieve and process data attempting to make more profitable trades. Tools similar to AI that gauge sentiment from social media or scrape text from company financial reports long predate ChatGPT.

Any information gleaned from running an AI process that may be “material” most likely is a subset of a vast information set already known by “material” market participants (many likely with their own AI process). In “efficient” markets, information is almost instantly reflected in individual and aggregate market prices through the process of buying and selling securities. If by chance AI detects “new” information, an investor or his AI device acting on that information instantly incorporates that new tidbit as a price into that vast information processing array that we describe as “The Market.”

Markets gather all the dispersed information known or knowable about a company or sectors of companies. Market participants actively evaluate the impact of any new information. They assign their own value based on that information. Almost instantly that value gets expressed as a price as shares trade in real time. That internet-linked web of endless global interconnections, in effect, is artificial intelligence. And no government formally arranged it.

AI’s limitation in securities transactions is its inadequate predictive ability. AI’s forecasting ability works well when assessing patterns that are relatively stable or with set rules, such as in chess games. Activities in financial markets, by contrast, are fantastically complex and keep changing — so much so that no one can precisely guess how much or in what way another particular piece of information coming along may impact prices. Thousands of other inputs (if not tens of thousands) are all interacting simultaneously. AI processes trying to predict particular market prices are like self-piloting cars attempting to read stop signs with words, shapes, and colors that change every day.

Consider the AI-Powered Equity ETF, launched in 2017. It employs IBM Watson’s AI to analyze publicly available information to identify US stocks to buy or sell that hopefully would outperform US market returns (Exhibit 1). You may recall that an early version of Watson, “Deep Blue,” was a chess-playing expert system run on a unique purpose-built IBM supercomputer. It was the first computer to win a chess game, and the first to win a match against a reigning world champion under regular time controls.

Exhibit 1: AI Powered Equity ETF vs. Russell 3000 Index and S&P Technology Select Sector Index

Cumulative returns, November 1, 2017–May 26, 2023

graph:  Exhibit 1: AI Powered Equity ETF vs.
Russell 3000 Index and S&P Technology Select Sector Index

Source:Dimensional Fund Advisors and FactSet. Sample period begins with the first full month of returns for AI Powered Equity ETF. Past performance is not a guarantee of future results. Indices are not available for direct investment; therefore, their performance does not reflect the expenses associated with the management of an actual portfolio. Frank Russell Company is the source and owner of Russell Indexes. S&P data © 2023 S&P Down Jones Indices LLC.

While Watson can outwit even the best individual chess master, its intelligence pales in comparison to the aggregate wisdom created by a million individuals competing for profit making daily stock market trades. We are unsurprised then that the Watson-powered ETF lags broad US market returns and, by a much wider margin, US technology sector returns since its inception about five years ago.

Sure, AI can improve technical aspects of trading execution. But the aggregated market intelligence of many independent participants is powerful and ensures the price at the time a trade is made fairly represents the true value of each stock or bond based on all that is known or knowable at that moment. There’s no reason to think that AI fundamentally should revise our trading methods substantially anytime soon.

Better Planning to Pursue Better Outcomes

We expect positive premiums in our equity portfolios every day. But decades of market data show that unexpected returns dominate the performance we see daily, monthly, quarterly, and annually. Many numerical changes in reports are, essentially, statistical “noise” making account statements confusing. Ignoring daily, monthly, quarterly, and even annual return vagaries among funds that deliver equity returns is often necessary for a systematic investing process to be successful.

Premiums that drive better long-term outcomes can turn around quickly and unpredictably. They are easily missed without consistent representation in an informed portfolio allocation structure. If they are missed, returns that you may have waited months or even years for, but then suddenly quit due to lack of discipline — distracted perhaps by better-seeming recent returns elsewhere — are gone forever.

Examining annualized return premiums derived from financial science going back decades, shows that small cap stocks have beaten large cap stocks, value has outperformed growth, and high profitability stocks have outgained low profitability stocks over long periods. How should you think about the performance of these premium dimensions1 over shorter time periods for more informed planning of your most essential goals?

Exhibit 2: Frequency of Small Cap, Value, and High Profitability Outperformance

bar graph: Frequency of Small Cap, Value, and High Profitability Outperformance

Past performance is no guarantee of future results. Investing risks include loss of principal and fluctuating value. There is no guarantee an investment strategy will be successful. Indices are not available for direct investment. Their performance does not reflect the expenses associated with the management of an actual portfolio. Source: Dimensional Fund Advisors. Number and percentage of quarters where market, size, value and/or profitability premiums were negative are calculated using monthly return data from July 1963 to March 2023. Market: Fama/French Total US Market Research Index minus the one-month US Treasury bill. Size: Dimensional US Small Cap Index minus the S&P 500 Index. Value: Fama/French US Value Research Index minus the Fama/French US Growth Research Index. Profitability: Fama/French US High Profitability Index minus the Fama/French US Low Profitability Index. Profitability is measured as operating income before depreciation and amortization minus interest expense scaled by book. The Dimensional and Fama/French Indices represent academic concepts that may be used in portfolio construction and are not available for direct investment or for use as a benchmark. Index returns are not representative of actual portfolios and do not reflect costs and fees associated with an actual investment.

Exhibit 2 supports the economic philosophy we believe will be most successful. The data show a multi-factor approach based on premium dimensions for investment management not only likely to outperform a simple indexed arrangements (popular with Vanguard) but also actively managed funds, EFTs and other schemes that pick stocks.

Performance during overlapping periods formed each month (e.g., January to December, February to January, etc.) over rolling periods of one, five, and 10 years, show these premium dimensions were positive over most one- and five‑year periods. They continue to increase over longer stretches. The value dimension, for instance, beat the growth dimension in 80% of 1,027 measurable 10-year periods.2 We reasonably infer for planning beyond rolling 10-year periods that calculations would improve closer to 100%. Of course, no investing period, however long, can provide certainty of realizing all the premiums.


The so-called Fama-French three-factor model (now with five factors) has proved itself in academia over the past three decades as the best economic model for estimating expected stock returns. It is an informed framework for organizing expected equity returns in an intuitive way. It has a risk story that makes sense for planning purposes: small and value stocks as a group have higher expected returns than big and growth stocks. They do so because we believe they have more risk. Bearing such risk consistently should reward an investor with higher returns over the long-term when they stick with a dimensionally diversified strategy.

No model is literally true, of course. But the dimensional multi-factor model Professional Financial has applied for almost thirty years has systematically guided better planning for our clients and helped us avoid many serious investment mistakes others have made chasing returns. Systematically targeting size, value and profitability premiums for our portfolio equity allocations gives your planning focus in uncertain times. Along with maintaining the right lifestyle and insurance, your odds of realizing a lifetime of financial security for you and those you care deeply about are dramatically improved. And most of all, planning with CFP® professionals can offer greater peace of mind.


1. Based on monthly rolling returns, computed as follows: Dimensional US Small Cap Index minus S&P 500 Index, June 1927–December 2021; Fama/French US Value Research Index minus Fama/French Us Growth Research Index, July 1926–December 2021; and Fama/French US High Profitability Index minus Fama/French US Low Profitability Index, July 1963–December 2021. Size premium: The return difference between small market capitalization stocks and large market capitalization stocks. Value premium: The return difference between stocks with low relative prices (value) and stocks with high relative prices (growth). Profitability premium: The return difference between stocks of companies with high profitability over those with low profitability.

2. Small vs. Large: 1,124 periods of 1 year; 1,076 periods of 5 years; 1,016 periods of 10 years. Value vs. Growth: 1,135 periods of 1 year; 1,087 periodsof 5 years; 1,027 periods of 10 years; High Profitability vs. Low Profitability: 691 periods of 1 year; 643 periods of 5 years; 583 periods of 10 years.