LLM 投资者模拟研究：行为准确性之外，个体级还原度仍存差距
Beyond Behavioral Accuracy: Evaluating Individual-Level Fidelity in LLM Investor Simulation
AI 导读
一项为期纵向的纸面交易研究以 80 名参与者为对象，评估大语言模型能否忠实再现个体的交易行为与决策逻辑。结果显示，没有任何被评估的 LLM 在预测交易发生上稳定优于简单的近期活动持续性基线，买卖结构与资产选择也难以复现，相似的活动表现还会产生明显不同的组合轨迹。消融实验表明交易活动预测主要依赖近期交易，资产选择则受多源证据影响，研究呼吁将行为准确性、解释对齐与证据依赖分开评估。
正文 · 原文
Abstract:Large language models (LLMs) are increasingly used as user simulators, but can they faithfully reproduce an individual's behavior and decision logic over time? We investigate this question in a longitudinal paper-trading study with 80 participants. Using rolling next-day predictions, we evaluate trading activity, buy-sell structure, asset selection, and portfolio outcomes, complemented by controlled input ablations and participant-rated explanation alignment. No evaluated LLM reliably outperforms a simple recent-activity persistence baseline for predicting trade occurrence, while buy-sell structure and asset selection remain difficult to reproduce. Similar activity-level performance can also yield substantially different portfolio trajectories. Ablations show that activity predictions depend primarily on recent trades, whereas asset choices are sensitive to multiple evidence sources. Participants rate explanations as more aligned with their recurring strategies than with their decision-specific rationales. These findings reveal a gap between coarse behavioral agreement and individual-level fidelity, motivating separate evaluation of behavioral accuracy, explanation alignment, and evidence dependence in personalized user simulation.
| Comments: | This paper is under review in ARR |
| Subjects: | Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Information Retrieval (cs.IR) |
| Cite as: | arXiv:2610.26524 [cs.AI] |
| (or arXiv:2610.26524v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.26524 arXiv-issued DOI via DataCite |
Submission history
From: Jiajie He [view email]
[v1]
Mon, 19 Oct 2026 14:54:08 UTC (3,924 KB)
[v2]
Sat, 24 Oct 2026 01:10:19 UTC (3,924 KB)
来源：arXiv cs.AI · arxiv.org