The fraud-on-the-market doctrine has been the backbone of securities class action litigation since Basic Inc. v. Levinson in 1988. It lets plaintiffs avoid the nearly impossible task of proving individual reliance – instead, they invoke a rebuttable presumption that anyone who traded in an efficient market effectively relied on the integrity of the market price, which already reflected the defendant’s alleged misstatements. For decades, that framework worked reasonably well.
AI is starting to break the system. Large investment firms and everyday investors now use computer programs to place their trades. This has raised a serious question in courts: when a computer placed the trade, who actually made the choice to rely on the information? The answer could reshape securities class action lawsuits, changing both how these cases get approved and how courts calculate the damages.
The Fraud-on-the-Market Presumption: A Quick Foundation
Under Basic, the reasoning works like this. In an efficient market, all public information about a company, including false or misleading statements, gets built into the stock price. Investors who trade at that price are assumed to have trusted it, even if they never personally read the company filings. Courts allow this assumption to be challenged, but for many years, defendants rarely managed to succeed.
That changed incrementally – Halliburton Co. v. Erica P. John Fund (2014) confirmed that defendants could challenge the presumption at the class certification stage by arguing the alleged misstatement had no price impact. Then, in Goldman Sachs Group v. Arkansas Teacher Retirement System (2021), the court sharpened that tool, allowing defendants to use the generic nature of a challenged statement as evidence of limited price impact.
The legal trend has been toward giving defendants more room to contest the presumption of reliance earlier in the case. AI-driven trading is the next front in that fight – and it’s a more complicated one.
Where AI Disrupts the Reliance Chain
The fraud-on-the-market theory assumes a human investor (or an investment process oriented around human judgment) that, at some level, looks to market price as a signal of fundamental value. The entire logic depends on the connection between price and trading decision.
AI trading systems don’t always work that way. Depending on the strategy, an algorithm might trade based on momentum signals, statistical arbitrage, order flow patterns, volatility surfaces, or machine-learned correlations that have nothing to do with the information content embedded in a stock’s price. Some high-frequency algorithms hold positions for milliseconds – far too short a time frame for fundamental information to be the operative input.
When a plaintiff’s trading activity is driven by one of these systems, the question isn’t whether the market was efficient. It’s whether the algorithm actually used the market price, as Basic contemplated, as the basis for the trade. If the algorithm was optimizing around a signal unrelated to company fundamentals, the causal chain between the defendant’s alleged fraud and the plaintiff’s loss becomes genuinely unclear.
This isn’t a hypothetical concern. It’s a structural feature of modern institutional trading, and it creates a real vulnerability for plaintiffs whose portfolios are managed algorithmically.
The Class Certification Battleground
The reliance issue becomes even more important during class certification. To get a damages class certified under Rule 23(b)(3), plaintiffs must show that the main legal questions are common to the whole class. The fraud-on-the-market presumption usually helps make reliance a shared issue. Without it, each investor would need to prove individual reliance, making class certification very difficult.
If a large number of investors in a proposed class used AI trading systems, defendants may argue that reliance is not the same for everyone in the class. An algorithm used by a quantitative hedge fund makes decisions very differently from a pension fund manager who studies earnings reports before investing. Treating both groups as if they relied on market prices in the same way may not reflect reality.
Courts have not yet reached a clear answer on this issue. Legal scholars have been discussing this issue for some time. A 2022 Texas Law Review article, “Artificially Intelligent Class Actions,” explored how AI-based trading challenges some of the basic assumptions behind securities fraud class actions. The real impact of these issues on litigation is only starting to emerge.
What Plaintiffs’ Counsel Needs to Consider
Plaintiffs who rely on AI reliance arguments in securities class actions to support class certification need to anticipate defense challenges with supporting economic analysis, not just legal arguments.
The key question is simple: what information was the algorithm actually using to make trading decisions? If the system used public price data that included the effects of the defendant’s alleged false statements, plaintiffs can argue that the algorithm effectively relied on an efficient market. However, making that argument requires technical and economic evidence showing how the system worked. It is not enough to simply say that the algorithm traded in the market and should automatically receive the benefit of the presumption.
Plaintiffs’ counsel should consider engaging financial economists early to conduct a technical analysis of the trading methodology. Understanding whether the algorithm was fundamentally- or technically-driven, how frequently it rebalanced in response to news events, and how its signals correlated with price movements during the alleged fraud period all become relevant. These are not simple questions, and they require quantitative analysis that goes beyond traditional economic expert work in securities litigation.
Defense Strategies: A New Rebuttal Tool
For defense lawyers, algorithmic trading may provide a new and potentially strong way to challenge the fraud-on-the-market presumption during class certification, before the case even reaches the main legal issues.
This approach involves hiring economic experts to closely examine the trading systems used by major class members, especially institutional investors. If the experts find that trading decisions were based on signals unrelated to fundamental price information, defendants can argue that those investors should not be part of a class that is based on reliance on the integrity of market prices.
This argument is particularly potent for proposed classes that include hedge funds, quantitative investment managers, and high-frequency traders. These investors are least likely to have traded based on the mechanism Basic assumed. Including them in a class alongside long-term retail investors creates internal inconsistency that defendants can exploit – both to defeat class certification and, where possible, to reduce the damages period.
The Role of Quantitative Financial Analysis in These Disputes
As AI reliance disputes in securities class actions become more common, the evidentiary demands will intensify on both sides. Courts won’t accept hand-waving about whether an algorithm “relied” on price information – they’ll require rigorous quantitative analysis of the trading system’s architecture, inputs, and behavioral outputs during the relevant period.
This is exactly the kind of technical and economic analysis I bring to securities litigation. With a PhD in economics, CFA designation, and over 16 years as a testifying expert witness in complex financial disputes – including multibillion-dollar securities cases – I have substantial experience applying econometric modeling and financial market analysis to the questions actually contested in court. Analyzing the relationship between market prices, information events, and trading decisions is core to my work, whether retained by plaintiff or defense counsel.
The AI reliance question is, at its core, about how financial systems process information and generate trading activity. That’s an economic and financial problem before it’s a legal one – and it requires an expert who can translate quantitative analysis into clear, defensible testimony.
What Comes Next
Courts and litigants are still in the early stages of working through these issues. Expect more targeted challenges to fraud-on-the-market at class certification, where algorithmic traders make up a meaningful portion of the proposed class. Expect more scrutiny of how plaintiffs’ damages experts define the class and establish reliance. And expect the underlying economics of AI-driven trading to become an increasingly important battleground in securities litigation.
The fraud-on-the-market presumption isn’t going away. But its application in a market where AI systems generate a growing share of trading volume is no longer straightforward – and the litigators who recognize that early will be better positioned on both sides of the case.
Work With an Expert Who Understands the Intersection of AI, Markets, and Litigation
If you’re litigating a securities class action where algorithmic trading is part of the picture – on either side – the economic analysis underlying your reliance arguments matters. Dr. Pavithra Kumar provides rigorous, court-tested analysis for counsel navigating complex securities disputes.