AI and active management are being positioned as the perfect partnership. Fund managers claim machine learning will finally give them the edge they’ve been seeking. But the academic evidence reveals a paradox: the better AI gets at analysis, the harder it becomes to beat the market.
After decades of disappointing performance, active fund managers believe they’ve finally found their salvation: artificial intelligence. The promise sounds compelling. AI can analyse thousands of companies in seconds, process datasets that would overwhelm human analysts, and eliminate the emotional biases that lead to poor decisions. A growing number of fund managers now tout their AI capabilities as justification for charging premium fees.
Should investors believe the hype?
Academic research shows AI can improve stock analysis. The problem is what happens next.
AI and active management — the laboratory evidence
The analytical edge is real. Researchers at the University of Chicago’s Booth School of Business built an AI model that analysed nothing but financial statements: balance sheets and income statements, stripped of company names and dates. The AI predicted the direction of future earnings with 60% accuracy, compared to 53% for human analysts examining the same information.
In another study, researchers constructed an “AI analyst” that outperformed 54% of human forecasts. Trading strategies based purely on AI predictions generated monthly alpha (returns above market movements) of 0.84% to 0.92%. These aren’t trivial improvements.
AI excels in specific circumstances. It processes high volumes of standardised information quickly, making it particularly effective at analysing smaller companies where data is transparent. It avoids cognitive biases that plague humans: anchoring to previous forecasts, overweighting recent news, herding toward consensus. And it identifies subtle patterns across thousands of securities simultaneously.
The most encouraging finding came from combining human judgement with machine intelligence. The hybrid approach (analysts working with AI rather than competing against it) beat pure AI models, achieving 57.3% accuracy versus 53.7%. Humans add value through institutional knowledge, understanding of intangible assets like brand strength, and contextual judgement during market stress. AI contributes computational power and pattern recognition.
Recent work by economists Paul Woolley and Dimitri Vayanos at the London School of Economics, now being explored by the Bank of England, suggests AI might disaggregate momentum trading from fundamental analysis in manager performance. This could identify which managers possess genuine stock-picking skill versus lucky timing.
Impressive results. But they share one crucial limitation: they’re pre-fee, pre-cost laboratory findings. The real world looks different.
The better AI gets, the harder it becomes to beat the market
Here’s the paradox active managers don’t advertise: if AI helps them analyse better, it makes markets more efficient. More accurate analysis leads to more accurate prices, which means smaller mispricings and less alpha available.
David Booth, co-founder of Dimensional Fund Advisors, identified this trap. AI, he noted in an article in the FT, is ultimately just artificial intelligence competing against the aggregate intelligence of millions of market participants. Markets incorporate all publicly available information into prices. “The market ensures that a price is the most accurate current representation of the value of a stock or bond,” Booth wrote. “It’s free and available to all.”
“The market ensures that a price is the most accurate current representation of the value of a stock or bond. It’s free and available to all.” — David Booth, co-founder, Dimensional Fund Advisors
AI-augmented funds will be victims of their own success.
The mathematics hasn’t changed. Nobel laureate William Sharpe established the iron law of active management: properly measured, the average actively-managed pound must underperform the average passively-managed pound, net of costs. It’s arithmetic. Before costs, active and passive investors collectively are the market, so they earn the market return on average. After costs, active investors lag.
AI doesn’t escape this. It simply raises the aggregate skill level of all managers competing against each other. A higher skill floor means a higher bar for outperformance.
Veteran investor Charley Ellis identified this pattern decades before AI arrived. In his book Rethinking Investing, Ellis explains: “While individual investment managers may be thrilled by the novelty and the power of their own firm’s technology, the larger reality is that almost all competitors already have, or will soon get, equal technology. Rather than providing a competitive advantage, technology is over and over again a great equaliser.”
When everyone has the same analytical tools, nobody has an advantage.
The evidence shows this equalisation in real time. Researchers tracking AI performance against humans found that in 2011, AI beat human analysts 64% of the time. By 2013, that had collapsed to 38%. Human analysts had gained access to similar tools. The edge disappeared in two years.
The real-world reality check
Academic studies measure gross returns before costs. Real-world active funds face annual management charges that typically range from 0.75% to 1.25%. But most active funds trade more frequently than index funds, so transaction costs add substantially to what investors pay. Then there are AI infrastructure expenses: data feeds, computing power, machine learning expertise. Even if AI-enhanced funds aren’t charging premium fees for the technology, they’re still active funds. Which means they’re expensive.
The studies show human analysts working with AI beat pure AI models. But that helps analysts stay employed, not investors earn better returns. The question isn’t whether AI improves analytical capability. It’s whether AI improves investor returns after costs.
“The question isn’t whether AI improves analytical capability. It’s whether AI improves investor returns after costs.”
No evidence yet suggests it does.
The edge erodes fast because AI tools aren’t proprietary for long. Commercial data vendors sell similar capabilities to multiple firms. Open-source machine learning libraries democratise core technology. As more capital deploys the same strategies, alpha decays. It’s a familiar pattern from quantitative factor investing: publish the factor, watch the returns disappear as everyone piles in.
Fund managers talk enthusiastically about their AI capabilities. But where’s the evidence of sustained after-fee outperformance? The burden of proof sits with active management. “We use AI” isn’t evidence of value. Net returns are.
As Ellis observes: “Professional managers are nearly equal in technology and in information and are doomed to become more and more equal in the days ahead.”
What investors should do
For most investors, AI strengthens the case for indexing. More efficient markets mean fewer opportunities for active managers to exploit. The cost advantage of passive investing remains insurmountable. If AI raises all managers’ skill levels equally, gross returns might improve slightly, but net returns after fees won’t.
“But some AI-enhanced funds must outperform?”
Perhaps, for a while. But which ones? And for how long? By the time retail investors identify yesterday’s winners, the edge typically evaporates. You’re making three simultaneous bets: your chosen fund has superior AI, this advantage persists despite competitive pressure, and fees don’t consume the gains.
“Shouldn’t I at least try allocating some money to AI-enhanced active funds?”
The burden of proof sits with active management. Demand evidence of persistent, after-fee outperformance over meaningful periods (five years minimum). Don’t accept “we use cutting-edge AI” as a substitute for performance data. Academic studies show what’s possible in controlled conditions. Actual funds operate in competitive reality where advantages disappear quickly.
“What if I use AI tools for my own stock picking?”
Individual investors face steeper barriers: data costs, infrastructure requirements, time commitment, technical expertise. You’re competing against professionals with better tools, more resources, and full-time focus. DIY active management with AI likely produces worse results than hiring professionals who themselves typically underperform index funds.
The simple test: if a fund manager emphasises their AI capabilities, ask one question: “What are your after-fee returns versus a comparable index fund over the past five years?” That tells you everything. Technology talk is marketing. Returns are reality.
“Technology talk is marketing. Returns are reality.”
The verdict — aggregate intelligence wins
AI represents genuine innovation in financial analysis. It will make markets more efficient, benefiting everyone through better price discovery. But here’s what AI in active management really means: artificial intelligence competing against aggregate intelligence.
Aggregate intelligence is millions of market participants making judgements, incorporating all publicly available information into prices, freely available to anyone through low-cost index funds, and getting smarter as more sophisticated tools become widespread.
Why pay premium fees to bet your AI is better than everyone else’s? Especially when advantages disappear rapidly while costs remain permanent.
For retail investors, the question isn’t whether AI improves stock analysis (it does). The question is whether AI-enhanced active management delivers better outcomes after costs. The evidence says no.
AI and active management might be the perfect partnership for fund managers seeking to justify their fees. But for investors seeking optimal returns, aggregate intelligence accessed through index funds remains the rational choice. Getting rich slowly through diversified, low-cost indexing remains the most reliable path.
The hype around AI and active management will continue. The arithmetic won’t change.
Resources
Cao, S., Jiang, W., Wang, J.L., & Yang, B. (2021). From Man vs. Machine to Man + Machine: The Art and AI of Stock Analyses. NBER Working Paper 28800.
Ellis, C.D. (2022). Rethinking Investing: Six Principles for an Uncertain World. McGraw Hill.
Kim, A.G., Muhn, M., & Nikolaev, V.V. (2024). Financial Statement Analysis with Large Language Models. University of Chicago Booth School of Business Working Paper.
Sharpe, W.F. (1991). The Arithmetic of Active Management. Financial Analysts Journal, 47(1), 7-9.
FIND OUT MORE
Want to find out more about how we can help you achieve your goals? Let’s arrange a meeting.
Financial Adviser and Financial Planner in Wharfedale
rockwealth Wharfedale is a fixed-fee Independent Financial Adviser situated in Wharfedale, North Yorkshire.About Us: Nestled in Wharfedale, rockwealth is your dedicated local financial planning firm. From tailored financial advice to pension and retirement advice, evidence-based investing, and inheritance tax planning, we are here to facilitate your financial journey in Wharfedale and the broader North Yorkshire area.
Interested to work with us?: Begin your financial journey with us through an Initial Discovery Consultation, completely free of charge and without any obligation. You can visit us at our office, schedule a video call, or call us on: 0113 541 9373.