@inproceedings{xue-etal-2026-temporal,
title = "Temporal Leakage in Financial News {NLP}: A Multi-Architecture Audit with a Regime-Specific {M}{\&}A Signal",
author = "Xue, Chenhao and
Guesmi, Raslen and
Feng, Siwei and
Gong, Yucheng and
Sundram, Jacob Xavier and
Pang, Jordan and
Wang, Lan and
Kaljuvee, Julian",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.2386/",
pages = "54209--54242",
ISBN = "979-8-89176-459-0",
abstract = "Financial-news direction prediction has become a popular NLP benchmark, yet reported gains depend critically on whether the train{--}test split is chronological or random, i.e., on temporal leakage. We audit this dependence on a 49,799-article corpus across 16 feature{--}model combinations spanning TF-IDF, MiniLM, FinBERT, and fine-tuned RoBERTa-large / DeBERTa-v3-large, plus separate zero/few-shot and LoRA probes of Llama-3 and Qwen2.5 LLMs: random splits inflate MCC by 1.1{\texttimes} to 6.5{\texttimes}, tracking model capacity and feature richness, and end-to-end FinBERT fine-tuning re-amplifies rather than closes the gap (size-matched ratio 1.75{\texttimes}). Conditioning on event type, mergers and acquisitions (M A) is the only audited category with a positive locked-test signal under near-temporal chronological evaluation (TF-IDF MCC = 0.138 train-only, 0.068 under train{\ensuremath{\cup}}val refit; 10,000-permutation p {\ensuremath{<}} 10{\ensuremath{^-}}{\textthreesuperior}); the signal does not transfer to FNSPID{'}s 2009{--}2020 U.S. corpus, localising the headline to our 2024{--}2025 European-tilted M A semantics rather than a universal predictor. Three independent role labellers converge on acquirer-tagged articles as the signal locus, a power-limited qualitative convergence rather than a hypothesis-tested asymmetry. Chronological splitting plays for financial NLP the role characteristics-purging plays for asset pricing: it strips the predictable, stale component of news and leaves a residual that is small, event-localized, and lexically shallow. We advocate leakage audits as a required disclosure for financial-NLP benchmarks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="xue-etal-2026-temporal">
<titleInfo>
<title>Temporal Leakage in Financial News NLP: A Multi-Architecture Audit with a Regime-Specific M&A Signal</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chenhao</namePart>
<namePart type="family">Xue</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Raslen</namePart>
<namePart type="family">Guesmi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siwei</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yucheng</namePart>
<namePart type="family">Gong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jacob</namePart>
<namePart type="given">Xavier</namePart>
<namePart type="family">Sundram</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jordan</namePart>
<namePart type="family">Pang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Julian</namePart>
<namePart type="family">Kaljuvee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Financial-news direction prediction has become a popular NLP benchmark, yet reported gains depend critically on whether the train–test split is chronological or random, i.e., on temporal leakage. We audit this dependence on a 49,799-article corpus across 16 feature–model combinations spanning TF-IDF, MiniLM, FinBERT, and fine-tuned RoBERTa-large / DeBERTa-v3-large, plus separate zero/few-shot and LoRA probes of Llama-3 and Qwen2.5 LLMs: random splits inflate MCC by 1.1× to 6.5×, tracking model capacity and feature richness, and end-to-end FinBERT fine-tuning re-amplifies rather than closes the gap (size-matched ratio 1.75×). Conditioning on event type, mergers and acquisitions (M A) is the only audited category with a positive locked-test signal under near-temporal chronological evaluation (TF-IDF MCC = 0.138 train-only, 0.068 under train\ensuremath\cupval refit; 10,000-permutation p \ensuremath< 10\ensuremath⁻³); the signal does not transfer to FNSPID’s 2009–2020 U.S. corpus, localising the headline to our 2024–2025 European-tilted M A semantics rather than a universal predictor. Three independent role labellers converge on acquirer-tagged articles as the signal locus, a power-limited qualitative convergence rather than a hypothesis-tested asymmetry. Chronological splitting plays for financial NLP the role characteristics-purging plays for asset pricing: it strips the predictable, stale component of news and leaves a residual that is small, event-localized, and lexically shallow. We advocate leakage audits as a required disclosure for financial-NLP benchmarks.</abstract>
<identifier type="citekey">xue-etal-2026-temporal</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2386/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>54209</start>
<end>54242</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Temporal Leakage in Financial News NLP: A Multi-Architecture Audit with a Regime-Specific M&A Signal
%A Xue, Chenhao
%A Guesmi, Raslen
%A Feng, Siwei
%A Gong, Yucheng
%A Sundram, Jacob Xavier
%A Pang, Jordan
%A Wang, Lan
%A Kaljuvee, Julian
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F xue-etal-2026-temporal
%X Financial-news direction prediction has become a popular NLP benchmark, yet reported gains depend critically on whether the train–test split is chronological or random, i.e., on temporal leakage. We audit this dependence on a 49,799-article corpus across 16 feature–model combinations spanning TF-IDF, MiniLM, FinBERT, and fine-tuned RoBERTa-large / DeBERTa-v3-large, plus separate zero/few-shot and LoRA probes of Llama-3 and Qwen2.5 LLMs: random splits inflate MCC by 1.1× to 6.5×, tracking model capacity and feature richness, and end-to-end FinBERT fine-tuning re-amplifies rather than closes the gap (size-matched ratio 1.75×). Conditioning on event type, mergers and acquisitions (M A) is the only audited category with a positive locked-test signal under near-temporal chronological evaluation (TF-IDF MCC = 0.138 train-only, 0.068 under train\ensuremath\cupval refit; 10,000-permutation p \ensuremath< 10\ensuremath⁻³); the signal does not transfer to FNSPID’s 2009–2020 U.S. corpus, localising the headline to our 2024–2025 European-tilted M A semantics rather than a universal predictor. Three independent role labellers converge on acquirer-tagged articles as the signal locus, a power-limited qualitative convergence rather than a hypothesis-tested asymmetry. Chronological splitting plays for financial NLP the role characteristics-purging plays for asset pricing: it strips the predictable, stale component of news and leaves a residual that is small, event-localized, and lexically shallow. We advocate leakage audits as a required disclosure for financial-NLP benchmarks.
%U https://aclanthology.org/2026.emnlp-main.2386/
%P 54209-54242
Markdown (Informal)
[Temporal Leakage in Financial News NLP: A Multi-Architecture Audit with a Regime-Specific M&A Signal](https://aclanthology.org/2026.emnlp-main.2386/) (Xue et al., EMNLP 2026)
ACL
- Chenhao Xue, Raslen Guesmi, Siwei Feng, Yucheng Gong, Jacob Xavier Sundram, Jordan Pang, Lan Wang, and Julian Kaljuvee. 2026. Temporal Leakage in Financial News NLP: A Multi-Architecture Audit with a Regime-Specific M&A Signal. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 54209–54242, Budapest, Hungary. Association for Computational Linguistics.