AI agents and “vibe-coded” trading systems are giving ordinary investors access to strategies once reserved for hedge funds — but automation may bring a new generation of risks to the market.

Tech_07092026
A futuristic home investor works alongside an AI trading assistant.

A quiet technological shift is changing the way individual Americans invest. Instead of simply asking artificial intelligence to explain earnings reports or suggest stocks, a growing number of retail traders are allowing AI systems to build strategies, monitor markets and, in some cases, execute trades on their behalf.

The result is something that would have seemed extraordinary only a few years ago: ordinary investors effectively operating miniature quantitative funds from laptops at home.

AI agents can now scan thousands of securities, monitor price movements, evaluate portfolio risk and follow trading instructions continuously. Some investors are even using natural-language prompts to create their own algorithms without having significant programming experience — a practice increasingly described as “vibe coding.”

The phenomenon marks the latest stage in the democratisation of financial technology.

A generation ago, sophisticated algorithmic trading systems were largely confined to hedge funds, investment banks and proprietary trading firms employing teams of mathematicians, programmers and quantitative researchers. Today, generative AI is beginning to compress many of those capabilities into tools accessible to anyone with a brokerage account and a computer.

One example reported by The Wall Street Journal is Colin Edsman, a stay-at-home father who has built a small team of AI agents to assist with his investments. His agents perform different functions: one searches the market for opportunities, another monitors existing positions, while another prepares periodic performance reports.

The unusual part is that none of them are human.

They are AI agents operating through Anthropic’s Claude system, running continuously from Edsman’s laptop.

Trading platforms are beginning to prepare for the same transition.

Moomoo, which introduced agentic trading capabilities earlier this year, believes automated agents could eventually represent a significant share of activity on its platform. Neil McDonald, the company’s US chief executive, told the Journal that individual investors using such systems were effectively becoming “mini hedge funds.”

The shift is significant because retail investors have already become far more influential in financial markets.

Commission-free trading, smartphone brokerage applications and social-media investing communities dramatically lowered the barriers to participation during the past decade. Artificial intelligence is now removing another barrier: technical expertise.

Until recently, developing an automated strategy might require knowledge of Python, financial databases, APIs and quantitative modelling.

Generative AI can now produce much of that code from ordinary instructions.

An investor might tell an AI system to identify stocks that have fallen below a particular technical threshold, compare them with earnings expectations, exclude companies with excessive debt and purchase only those meeting predetermined risk criteria.

The AI can then generate the code needed to carry out the strategy.

That is the essence of vibe coding: describing what a program should do rather than manually writing every line of software. IBM describes the broader practice as building software largely through natural-language interaction with AI models rather than conventional programming.

Applied to investing, the implications are considerable.

A retail trader can potentially create dozens of experimental strategies in hours rather than weeks. AI agents can then test those strategies against historical data, track markets continuously and modify portfolio allocations as conditions change.

For some investors, the appeal is not necessarily beating the market. It is removing emotion from the process.

Human investors frequently make poor decisions because of fear, greed or impatience. They may hold losing positions too long, sell during market panics or chase stocks after rapid increases.

Algorithms, by contrast, can follow predefined instructions without becoming nervous when markets fall.

That discipline has long been one of the attractions of quantitative investing.

But giving AI control over personal portfolios introduces a different set of risks.

One of the most important is that an artificial-intelligence system may appear far more confident than its actual understanding warrants.

Large language models can generate convincing explanations even when their underlying reasoning is incomplete or incorrect. In financial markets, a fabricated statistic or misunderstood earnings figure can have immediate monetary consequences.

Academic research has also raised doubts about whether AI-generated stock portfolios consistently outperform traditional strategies.

An April 2026 National Bureau of Economic Research study examining portfolios recommended by several large language models found that AI tended to favour relatively concentrated portfolios tilted toward momentum stocks, large companies and businesses receiving substantial media attention.

The researchers found no statistically significant abnormal returns from either buy-and-hold or actively managed AI portfolios under their methodology.

That finding highlights an important distinction between possessing sophisticated tools and possessing an investment advantage.

If millions of investors have access to the same AI models, the information advantage created by those models may quickly disappear.

Indeed, financial markets could become more crowded rather than more efficient.

The CFA Institute has warned that wider adoption of autonomous trading and large language models could cause markets to absorb information faster while simultaneously becoming more vulnerable to crowded positioning, reflexive behaviour and failures shared across similar models.

Imagine thousands of AI agents receiving similar economic data, interpreting it in comparable ways and reaching the same conclusion at almost the same moment.

What initially appears to be independent investment analysis could become algorithmic herd behaviour.

If the agents all decide to sell simultaneously, relatively small changes in market information could produce unusually large price movements.

Wall Street has experienced versions of this phenomenon before.

Quantitative funds suffered significant losses during episodes when strategies that appeared diversified turned out to be based on similar assumptions. More recently, systematic funds have again demonstrated how crowded algorithmic trades can reverse rapidly during volatile markets. Reuters reported in July that some systematic managers had surrendered roughly a quarter of their year-to-date gains during an abrupt market reversal.

Retail AI could eventually amplify similar dynamics on a much larger scale.

There is also the problem of software reliability.

A poorly written trading program might execute an order incorrectly, misunderstand position sizing or continue trading during unusual market conditions.

Professional quantitative firms spend enormous sums testing their systems, building risk controls and monitoring infrastructure.

A retail investor who creates an algorithm through a conversational AI tool may not realise that the underlying code contains vulnerabilities or logical errors.

The convenience of vibe coding could therefore create a dangerous illusion: that producing software is the same as understanding it.

Still, enthusiasm is growing rapidly.

A survey of 938 US retail investors conducted by Investing.com in March found that 62% had already used AI in some form to inform investment decisions. Among those users, 65% said they believed AI had improved their investment performance. Most respondents nevertheless reported continuing to verify AI-generated information before acting on it.

Professional finance is moving in the same direction.

A recent NBER study examining AI adoption among investment managers found that AI-driven investing has expanded steadily since the early 2010s, particularly among hedge funds. Early AI adopters initially generated stronger performance than non-AI competitors, although that advantage diminished over time.

That pattern may ultimately prove instructive for retail investors.

AI can provide an advantage when relatively few participants possess the technology. But once the same capabilities become universally available, the advantage shifts elsewhere — perhaps toward better data, better risk management or better judgement about when not to follow the machine.

The bigger transformation may therefore be less about whether AI can pick winning stocks and more about who can participate in quantitative finance.

Artificial intelligence is lowering the technical barrier that once separated professional trading desks from ordinary investors.

A person who previously bought a handful of index funds can now build an automated strategy, test it across historical markets and deploy agents that monitor hundreds of securities around the clock.

That is a profound change.

The smartphone turned millions of people into active market participants. Commission-free trading made frequent transactions economically viable. Social media created vast communities capable of moving individual stocks.

AI may represent the next step: turning those investors into amateur systematic traders.

The future retail investor may no longer sit in front of multiple screens watching prices.

Instead, they may supervise a collection of autonomous agents, each searching for opportunities, managing risk and executing instructions while the human investor decides the broader strategy.

That vision resembles a hedge fund more than a traditional brokerage account.

Yet the most important lesson from professional quantitative investing remains unchanged: automation does not eliminate risk. It simply changes where the risk resides.

When humans trade badly, emotion is often responsible.

When machines trade badly, the problem can be hidden inside code, data or assumptions — and thousands of identical systems can potentially make the same mistake at once.

AI may therefore give ordinary investors capabilities that once belonged exclusively to Wall Street’s most sophisticated firms.

Whether it also gives them Wall Street-level returns is another question entirely.

Trending