All That Glisters Is Not Gold: A Benchmark for Reference-Free Counterfactual Financial Misinformation Detection

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jiang, Yuechen, Liu, Zhiwei, Cao, Yupeng, He, Yueru, Xu, Ziyang, Xu, Chen, Deng, Zhiyang, Tiwari, Prayag, Chen, Xi, Lopez-Lira, Alejandro, Huang, Jimin, Tsujii, Junichi, Ananiadou, Sophia
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912810588438528
author Jiang, Yuechen
Liu, Zhiwei
Cao, Yupeng
He, Yueru
Xu, Ziyang
Xu, Chen
Deng, Zhiyang
Tiwari, Prayag
Chen, Xi
Lopez-Lira, Alejandro
Huang, Jimin
Tsujii, Junichi
Ananiadou, Sophia
author_facet Jiang, Yuechen
Liu, Zhiwei
Cao, Yupeng
He, Yueru
Xu, Ziyang
Xu, Chen
Deng, Zhiyang
Tiwari, Prayag
Chen, Xi
Lopez-Lira, Alejandro
Huang, Jimin
Tsujii, Junichi
Ananiadou, Sophia
contents We introduce RFC Bench, a benchmark for evaluating large language models on financial misinformation under realistic news. RFC Bench operates at the paragraph level and captures the contextual complexity of financial news where meaning emerges from dispersed cues. The benchmark defines two complementary tasks: reference free misinformation detection and comparison based diagnosis using paired original perturbed inputs. Experiments reveal a consistent pattern: performance is substantially stronger when comparative context is available, while reference free settings expose significant weaknesses, including unstable predictions and elevated invalid outputs. These results indicate that current models struggle to maintain coherent belief states without external grounding. By highlighting this gap, RFC Bench provides a structured testbed for studying reference free reasoning and advancing more reliable financial misinformation detection in real world settings.
format Preprint
id arxiv_https___arxiv_org_abs_2601_04160
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle All That Glisters Is Not Gold: A Benchmark for Reference-Free Counterfactual Financial Misinformation Detection
Jiang, Yuechen
Liu, Zhiwei
Cao, Yupeng
He, Yueru
Xu, Ziyang
Xu, Chen
Deng, Zhiyang
Tiwari, Prayag
Chen, Xi
Lopez-Lira, Alejandro
Huang, Jimin
Tsujii, Junichi
Ananiadou, Sophia
Computation and Language
Computational Engineering, Finance, and Science
Computational Finance
We introduce RFC Bench, a benchmark for evaluating large language models on financial misinformation under realistic news. RFC Bench operates at the paragraph level and captures the contextual complexity of financial news where meaning emerges from dispersed cues. The benchmark defines two complementary tasks: reference free misinformation detection and comparison based diagnosis using paired original perturbed inputs. Experiments reveal a consistent pattern: performance is substantially stronger when comparative context is available, while reference free settings expose significant weaknesses, including unstable predictions and elevated invalid outputs. These results indicate that current models struggle to maintain coherent belief states without external grounding. By highlighting this gap, RFC Bench provides a structured testbed for studying reference free reasoning and advancing more reliable financial misinformation detection in real world settings.
title All That Glisters Is Not Gold: A Benchmark for Reference-Free Counterfactual Financial Misinformation Detection
topic Computation and Language
Computational Engineering, Finance, and Science
Computational Finance
url https://arxiv.org/abs/2601.04160