arXiv subjects

Andrew W. Lo

Publications and source records attributed to Andrew W. Lo.

10 recordsLinked to original sources

arXiv2026

Why Better Models Can Create Riskier Systems: Evidence from LLM Agents in Financial Markets

Large language models (LLMs) are being deployed at scale in consequential real-world systems, from financial markets to content moderation to hiring. We show that improving individual model capability can degrade rather than improve system-level outcomes. We hypothesize that shared training and architectures can lead more capable LLMs to behave more similarly, creating correlated actions that do not diversify away. We develop a general framework showing how this correlation creates a non-diversifiable risk floor and test its predictions in financial markets using an agent-based simulation with LLM traders of varying general-purpose capability. We find that: (1) frontier LLMs exhibit significantly correlated behavior that increases with capability; (2) when their shared reasoning is accurate, increasing agent participation reduces market-level risk; and (3) when agents share a common misinformation environment, the same correlated behavior becomes a liability. Together, these results identify a capability paradox: improving individual models does not necessarily produce better system-level outcomes. Whether the same dynamics arise in other domains is an open empirical question.

cs.AI↗

arXiv2026

Breaking Bad Financial Habits: How LLM Conversations Correct Financial Misconceptions

Financial misconceptions carry direct economic costs, from panic selling to equity market avoidance, yet they are notoriously resistant to correction. Traditional financial literacy interventions are constrained by cost, reach, and a persistent gap between knowledge and behavioral change. Across three pre-registered studies, we find that purposefully designed LLMs can durably correct financial misconceptions. Critically, two factors are necessary for this effect. First, corrective intent: LLMs prompted only to discuss a misconception produce corrections no better than unassisted self-reflection, and undirected LLM conversations can actively entrench misconceptions. Second, recipient receptivity: financial concepts are often foreign to the investors who misapply them, and LLM responses pitched below a participant's financial sophistication are judged as less credible and produce substantially weaker corrections. LLMs thus offer a scalable alternative to traditional financial literacy intervention, but only when designed with both factors in mind.

cs.HC↗

arXiv2026

One Size Fits None: Heuristic Collapse in LLM Investment Advice

Large language models are increasingly deployed as advisors in high-stakes domains -- answering medical questions, interpreting legal documents, recommending financial products -- where good advice requires integrating a user's full context rather than responding to salient surface features. We investigate whether frontier LLMs actually do this, or whether they instead exhibit heuristic collapse: a systematic reduction of complex, multi-factor decisions to a small number of dominant inputs. We study the phenomenon in investment advice, where legal standards explicitly require individualized reasoning over a client's full circumstances. Applying interpretable surrogate models to LLM outputs, we find systematic heuristic collapse: investment allocation decisions are largely determined by self-reported risk tolerance, while other relevant factors contribute minimally. We further find that web search partially attenuates heuristic collapse but does not resolve it. These findings suggest that heuristic collapse is not resolved by web search augmentation or model scale alone, and that deploying LLMs as advisors requires auditing input sensitivity, not just output quality.

cs.CL↗

arXiv2025

AUTOCT: Automating Interpretable Clinical Trial Prediction with LLM Agents

Clinical trials are critical for advancing medical treatments but remain prohibitively expensive and time-consuming. Accurate prediction of clinical trial outcomes can significantly reduce research and development costs and accelerate drug discovery. While recent deep learning models have shown promise by leveraging unstructured data, their black-box nature, lack of interpretability, and vulnerability to label leakage limit their practical use in high-stakes biomedical contexts. In this work, we propose AutoCT, a novel framework that combines the reasoning capabilities of large language models with the explainability of classical machine learning. AutoCT autonomously generates, evaluates, and refines tabular features based on public information without human input. Our method uses Monte Carlo Tree Search to iteratively optimize predictive performance. Experimental results show that AutoCT performs on par with or better than SOTA methods on clinical trial prediction tasks within only a limited number of self-refinement iterations, establishing a new paradigm for scalable, interpretable, and cost-efficient clinical trial prediction.

cs.LG↗

arXiv2024

LLM economicus? Mapping the Behavioral Biases of LLMs via Utility Theory

Humans are not homo economicus (i.e., rational economic beings). As humans, we exhibit systematic behavioral biases such as loss aversion, anchoring, framing, etc., which lead us to make suboptimal economic decisions. Insofar as such biases may be embedded in text data on which large language models (LLMs) are trained, to what extent are LLMs prone to the same behavioral biases? Understanding these biases in LLMs is crucial for deploying LLMs to support human decision-making. We propose utility theory-a paradigm at the core of modern economic theory-as an approach to evaluate the economic biases of LLMs. Utility theory enables the quantification and comparison of economic behavior against benchmarks such as perfect rationality or human behavior. To demonstrate our approach, we quantify and compare the economic behavior of a variety of open- and closed-source LLMs. We find that the economic behavior of current LLMs is neither entirely human-like nor entirely economicus-like. We also find that most current LLMs struggle to maintain consistent economic behavior across settings. Finally, we illustrate how our approach can measure the effect of interventions such as prompting on economic biases.

cs.CL↗

arXiv2011

Privacy-Preserving Methods for Sharing Financial Risk Exposures

Unlike other industries in which intellectual property is patentable, the financial industry relies on trade secrecy to protect its business processes and methods, which can obscure critical financial risk exposures from regulators and the public. We develop methods for sharing and aggregating such risk exposures that protect the privacy of all parties involved and without the need for a trusted third party. Our approach employs secure multi-party computation techniques from cryptography in which multiple parties are able to compute joint functions without revealing their individual inputs. In our framework, individual financial institutions evaluate a protocol on their proprietary data which cannot be inverted, leading to secure computations of real-valued statistics such a concentration indexes, pairwise correlations, and other single- and multi-point statistics. The proposed protocols are computationally tractable on realistic sample sizes. Potential financial applications include: the construction of privacy-preserving real-time indexes of bank capital and leverage ratios; the monitoring of delegated portfolio investments; financial audits; and the publication of new indexes of proprietary trading strategies.

q-fin.RM↗

arXiv2010

WARNING: Physics Envy May Be Hazardous To Your Wealth!

The quantitative aspirations of economists and financial analysts have for many years been based on the belief that it should be possible to build models of economic systems - and financial markets in particular - that are as predictive as those in physics. While this perspective has led to a number of important breakthroughs in economics, "physics envy" has also created a false sense of mathematical precision in some cases. We speculate on the origins of physics envy, and then describe an alternate perspective of economic behavior based on a new taxonomy of uncertainty. We illustrate the relevance of this taxonomy with two concrete examples: the classical harmonic oscillator with some new twists that make physics look more like economics, and a quantitative equity market-neutral strategy. We conclude by offering a new interpretation of tail events, proposing an "uncertainty checklist" with which our taxonomy can be implemented, and considering the role that quants played in the current financial crisis.

q-fin.RM↗

arXiv2010

Is It Real, or Is It Randomized?: A Financial Turing Test

We construct a financial "Turing test" to determine whether human subjects can differentiate between actual vs. randomized financial returns. The experiment consists of an online video-game (http://arora.ccs.neu.edu) where players are challenged to distinguish actual financial market returns from random temporal permutations of those returns. We find overwhelming statistical evidence (p-values no greater than 0.5%) that subjects can consistently distinguish between the two types of time series, thereby refuting the widespread belief that financial markets "look random." A key feature of the experiment is that subjects are given immediate feedback regarding the validity of their choices, allowing them to learn and adapt. We suggest that such novel interfaces can harness human capabilities to process and extract information from financial data in ways that computers cannot.

q-fin.GN↗

arXiv2009

A Computational View of Market Efficiency

We propose to study market efficiency from a computational viewpoint. Borrowing from theoretical computer science, we define a market to be \emph{efficient with respect to resources $S$} (e.g., time, memory) if no strategy using resources $S$ can make a profit. As a first step, we consider memory-$m$ strategies whose action at time $t$ depends only on the $m$ previous observations at times $t-m,...,t-1$. We introduce and study a simple model of market evolution, where strategies impact the market by their decision to buy or sell. We show that the effect of optimal strategies using memory $m$ can lead to "market conditions" that were not present initially, such as (1) market bubbles and (2) the possibility for a strategy using memory $m' > m$ to make a bigger profit than was initially possible. We suggest ours as a framework to rationalize the technological arms race of quantitative trading firms.

cs.CE↗

arXiv1999

Frontiers of finance: Evolution and efficient markets

In this review article we explore several recent advances in the quantitative modeling of financial markets. We begin with the Efficient Markets Hypothesis and describe how this controversial idea has stimulated a number of new directions of research, some focusing on more elaborate mathematical models that are captable of rationalizing the empirical facrts, others taking a completely different different tack in rejecting rationality altogether. One of the most promising directions is to view financial markets from a biological perspective and, specifically, with an evolutionary framework in which markets, instruments, institutions, and investors interact and evolve dynamically according to the "law" of economic selection. Under this view, financial agents compete and adapt, but they do not necessarily do so in an optimal fashion. Evolutionary and ecological models of financial markets is truly a new frontier whose exploration has just begun.

adap-org↗