In late June, algorithmic trading funds experienced what the Financial Times called a “quant tremor”, a downward fluctuation affecting quantitative traders, which the newspaper predicted is becoming more common.
While quantitative trading firms are doing well this year, Goldman Sachs’ prime brokerage suffered their worst five-day performance since December 2023.
This is not a new phenomenon. Any finance-head would struggle to forget the 2007 quant crunch, and hedge funds and big financial institutions have employed algorithmic trading for decades.
But the barriers to algorithmic trading tools have dropped dramatically in recent years, driven by rapid advances in artificial intelligence.
Dozens of AI trading applications, from code-free automation tools such as Capitalise.ai, to quantitative research and strategy building platforms like QuantConnect, have opened the field to millions of individual retail investors who need little more than an internet connection and the budget for a platform subscription.
AI products can process large amounts of data, track market movements, spot patterns and even execute trades. They can also enhance the speed, responsiveness, perception and information level of every trader using them.
Rotem Farkash: AI can improve market access, but it may amplify systemic risks
Rotem Farkash, an AI expert and trader, who has founded algorithmic trading companies, is clear that AI in trading should be understood in two ways: “as a tool that can broaden access and improve pricing, but also as one that introduces new forms of risk”.
The first risk is as more traders rely on similar AI-driven signals, market movements may become more synchronised, magnifying swings and increasing volatility across the financial system.
The risk of converging trades has been around for a long time. Long Term Capital Management’s 1998 collapse and the 2010 Flash Crash showed how quantitative strategies can unravel rapidly. But now AI is amplifying this risk by increasing the frequency of automated trades, making even deeper crashes possible.
The second risk is machine error. While AI may reduce some human mistakes, it is not perfect. If AI misinterprets a word or does not understand the context of a particular piece of information, it could trigger purchases or sales that have significant consequences for the trader.
Ken Griffin: AI “is profoundly more powerful than it was just nine months ago”
Ken Griffin’s Citadel, perhaps the world’s best-known quantitative hedge fund, initially approached AI as a tool for operational efficiency, to accelerate research, automate workflows and improve internal processes.
Yet in May, Griffin acknowledged how quickly the technology had advanced, saying it was “profoundly more powerful than it was just nine months ago”. That shift, he argued, had allowed Citadel to “unleash a much broader array of use cases for AI”, with work that would once have required people with masters and PhDs in finance weeks or months being completed by AI agents in hours or days.
Some firms are pushing this even further. Minotaur Capital, an Australia-based firm, has built its investment process around a proprietary AI platform called Taurient, which is designed to identify global stock opportunities.
Its strategy, which focuses on under-researched equities, delivered a 13.7% return for its flagship fund in the six months to January 2025, outperforming the MSCI All-Country World Index.
AI will likely complement humans, but is not without risk
For now, AI stock pickers may be capable of outperforming some index funds, but they remain some distance from displacing human expertise when that expertise is itself enhanced by AI tools.
The more likely outcome is that leading firms combine artificial and human intelligence, rather than replacing one with the other.
But both institutional and retail investors should remain cautious. AI may be a powerful tool in trading, but it is not risk-free. It poses a threat to market stability and can replicate the same errors that have long undermined human traders.
