In financial markets, effective trading decisions require accurately interpreting unstructured information, especially news articles. However, prior work overlooks the complex news and agent interactions when making trading decisions. To address these limitations, we propose FinRAQ, which enhances the news importance scoring by modeling semantic and temporal interactions between news articles. FinRAQ incorporates a Socratic questioning stage into the debate framework, prompting agents to challenge each other’s reasoning through Socratic questions. Extensive experiments across six company stocks show that FinRAQ outperforms state-of-the-art baselines, achieving up to 38% improvement in annualized returns.

Existing evaluations of large language models primarily rely on single-agent dilemmas or static binary-choice tasks, offering limited insight into how cooperation contexts influence LLM behavior. We introduce CoopValue, a multi-agent evaluation framework that assesses LLMs’ value preferences through cooperative scenarios. CoopValue includes 1,778 scenarios spanning all pairwise conflicts among the 10 Schwartz values and three cooperation types: reciprocal, coopetitive, and altruistic. We evaluate 24 LLMs across 8 model families and examine how their value preferences vary across different cooperative contexts, showing the importance of assessing LLM value preferences in interactive, context-sensitive settings to guide the selection and deployment of LLMs aligned with desired cooperative behavior.