While AI has a chokehold on Wall Street, it can't beat human judgment...
At least, not yet.
A recent Harvard-led study put "AI versus human investors" to the test. It looked at whether an AI algorithm could predict exactly what a fund manager was trying to do when buying, selling, or holding a stock.
On the surface, the results were promising. Using data from 1990 through 2023, the model was able to correctly predict about 71% of mutual-fund trading decisions.
That data suggests most active management follows a pattern a machine can copy... and potentially automate.
But here's the good news for our anti-robot readers...
The 29% of trades the model couldn't anticipate also followed a pattern. They were the trades more closely associated with outperformance.
In other words, AI proved it could mimic the routine part of the job. As for the harder-to-see, non-routine decisions... it had a tougher time. And those were the ones that helped managers beat the market.
As we'll explain today, human judgment still brings plenty to the table. The best investing ideas come from what machines can't yet model.
Harvard's AI test subject learned the playbook that everyone already knows...
A lot of this goes back to what we talked about earlier this month, when we reviewed Howard Marks' recent remarks.
Marks looked at how AI can help us be better investors. New technology is pretty good at recognizing patterns. And it's true that good investing is partly about pattern recognition.
Great investing, however, is about knowing when the pattern no longer applies.
The Harvard model seems to have learned the first part very well. According to the study, it got especially good at understanding how managers react to repeated stimuli like changing economic conditions.
It could guess what might happen in response to interest-rate hikes or cuts or inflation trends.
We're not surprised the model was able to catch on... A huge portion of investment management is systematic.
And the study actually highlights an inherent weakness on Wall Street. Many managers simply follow the crowd.
The model could predict whether a fund would buy or sell a particular stock based on what its peers in its fund category did.
But when everyone is making the same moves, it's hard to beat the market.
That's where the 'unpredictable' part of investing comes in...
It's what sets skilled managers apart.
A great example is the role of "conviction." Most fund managers put their biggest weightings in the ideas they have the most faith in.
The study found that the top five companies in a given portfolio were the hardest for the model to predict.
In many cases, it was because the edge was based somewhat on speculation. The "edge" here depends on a judgment about what the market is missing before the evidence is obvious.
Managers were able to outperform the market by leaning into companies with three qualities... high sales growth, high research and development (R&D) relative to earnings, and positive earnings surprises.
Investing in such companies led to outperformance of roughly 4.25%.
These sorts of non-standard decisions can confuse a model trained on past trading data. If a company is pouring cash into R&D, it's hoping for a breakthrough that will lead to future earnings.
An algorithm can't predict whether those investments will bear fruit or not.
This is where human judgment earns its keep...
Like Marks said, there are non-quantitative factors – like trying to predict a company's future – that AI isn't as helpful with.
You can use AI for the basics. It can speed up the "rote" parts of the job. It can pull together data that tells you how interest-rate changes affect your portfolio... And it can summarize the habits of the crowd.
It's a great tool for the parts of investing that are repetitive and rely on data everyone has access to.
But don't expect AI to find the next great surprise before the market does. It doesn't know how to account for factors that are less obvious, like R&D spending. And it can't predict when a company will surprise Wall Street.
For that, we'll stick with good old-fashioned human intelligence.
Regards,
Joel Litman
March 20, 2026
