Beyond Polarity: Multi-Dimensional LLM Sentiment Signals for WTI Crude Oil Futures Return Prediction

Dehao Dai; Ding Ma; Dou Liu; Kerui Geng; Yiqing Wang

q-fin.STMar 12, 2026

Beyond Polarity: Multi-Dimensional LLM Sentiment Signals for WTI Crude Oil Futures Return Prediction

Authors: Dehao Dai, Ding Ma, Dou Liu, Kerui Geng, Yiqing Wang

Organizations: University of California San Diego, La Jolla, California, United States · Georgia Institute of Technology, Atlanta, Georgia, United States · New York University, New York, New York, United States · Tulane University, New Orleans, Louisiana, United States · Southern Methodist University, Dallas, Texas, United States

Abstract

Forecasting crude oil prices remains challenging because market-relevant information is embedded in large volumes of unstructured news and is not fully captured by traditional polarity-based sentiment measures. This paper examines whether multi-dimensional sentiment signals extracted by large language models improve the prediction of weekly WTI crude oil futures returns. Using energy-sector news articles from 2020 to 2025, we construct five sentiment dimensions covering relevance, polarity, intensity, uncertainty, and forwardness based on GPT-4o, Llama 3.2-3b, and two benchmark models, FinBERT and AlphaVantage. We aggregate article-level signals to the weekly level and evaluate their predictive performance in a classification framework. The best results are achieved by combining GPT-4o and FinBERT, suggesting that LLM-based and conventional financial sentiment models provide complementary predictive information. SHAP analysis further shows that intensity- and uncertainty-related features are among the most important predictors, indicating that the predictive value of news sentiment extends beyond simple polarity. Overall, the results suggest that multi-dimensional LLM-based sentiment measures can improve commodity return forecasting and support energy-market risk monitoring.

Figures & tables

	Full Sample	Sentiment Sample (20%)
Total number of articles	29,153	8,639
Date range	01/01/2020–12/31/2025
Avg. articles/week	93	28

Table 1: News Article Corpus: Summary Statistics

Model	Relevance	Polarity	Intensity	Uncertainty	Forwardness
Range	[0, 1]	[-1, 1]	[0, 1]	[0, 1]	[0, 1]
GPT-4o	$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$
Llama 3.2-3b	$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$
FinBERT		$\checkmark$	$\checkmark$
AlphaVantage		$\checkmark$

Table 2: Sentiment Dimensions by Model

	GPT-4o	Llama 3.2	FinBERT	AlphaVantage
Mean	0.2952	0.1109	0.3050	0.2177
Stand Deviation	0.4323	0.3458	0.6801	0.2818
Min	-0.9000	-1.0000	-0.9772	-0.9144
$Q25\%$	0.0000	0.0000	0.0000	0.0368
Median	0.3000	0.0000	0.6491	0.2758
$Q75\%$	0.7000	0.2000	0.9000	0.4218

Table 3: Distributional Properties of Polarity Scores Across Models

Figure 1: Pairwise Polarity Correlation Across Models

Figure 2: Sentiment Dimension Distributions: GPT-4o vs Llama 3.2

Model	AUC		Accuracy		IC
	Mean	Std	Mean	Std	Mean	Std
AV Baseline	0.5694	0.0349	0.4902	0.0328	0.0910	0.0894
Tradition	0.6200	0.0666	0.5216	0.0505	0.1571	0.1398
GPT-4o	0.6336	0.0563	0.5373	0.0490	0.2490	0.0372
Llama 3.2	0.5892	0.1013	0.5137	0.0454	0.1147	0.1586
LLM Ensemble	0.5974	0.0439	0.5804	0.0697	0.2118	0.0620

Table 4: Cross-Validation Prediction Performance Across Feature Sets

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Feature Name	Model	Range	Description
gpt_article_count	GPT-4o	(4, 394)	Number of sampled articles in the week
gpt_relevance_mean	GPT-4o	(0.18, 0.8)	Mean relevance to energy markets, used as aggregation weight
gpt_polarity_mean	GPT-4o	(-0.35, 0.71)	Relevance-weighted mean sentiment polarity
gpt_intensity_mean	GPT-4o	(0.42, 0.72)	Relevance-weighted mean sentiment intensity
gpt_uncertainty_mean	GPT-4o	(0.13, 0.41)	Relevance-weighted mean degree of uncertainty
gpt_forwardness_mean	GPT-4o	(0.33, 0.83)	Relevance-weighted mean forward-looking orientation

Appendix

Table 5: Full Feature List with Descriptions and Ranges

Explore similar work

Beyond Sentiment: Structured Information Extraction from Financial News
Jul 30, 2026Daohan Zhu, Sitong Ge, Ruofei Wang +4Financial Sentiment Analysis Document Information Extraction
School of ASEE, Beihang University, Beijing, China · School of BME, Beihang University, Beijing, China · CAIR and CECS, VinUniversity, Hanoi, Vietnam
Not All News Is Equal: Topic- and Event-Conditional Sentiment from Finetuned LLMs for Aluminum Price Forecasting
Mar 10, 2026Alvaro Paredes Amorin, Andre Python, Christoph WeisserFinancial Forecasting Time Series Forecasting
International Business School, Zhejiang University · Center for Data Science, Zhejiang University · Centre for Human Genetics, Nuffield Department of Medicine, Oxford University +2
PolyBench: Benchmarking LLM Forecasting and Trading Capabilities on Live Prediction Market Data
Apr 3, 2026Pu Cheng, Juncheng Liu, Yunshen LongForecasting Benchmarks LLM Evaluation
College of Electrical Engineering, Sichuan University, Chengdu, Sichuan, China · School of Economics, Sichuan University, Chengdu, Sichuan, China

Jul 30, 2026J/K move · Enter open · S save

Beyond Sentiment: Structured Information Extraction from Financial News

Daohan Zhu, Sitong Ge, Ruofei Wang, Honggu Chen+3

School of ASEE, Beihang University, Beijing, China · School of BME, Beihang University, Beijing, China · CAIR and CECS, VinUniversity, Hanoi, Vietnam

Financial Sentiment Analysis Document Information Extraction LLM Information Extraction Finance

Open paper AskPDF arXiv

Financial sentiment analysis has become a standard component in news-driven stock prediction, yet it reduces rich, multi-dimensional news articles to a single polarity score. We hypothesize that financial news encodes multiple orthogonal information dimensions---event type, impact scope, temporal horizon, and semantic confidence---that sentiment alone cannot capture, and that these dimensions carry independent predictive value. To test this hypothesis, we propose a structured information extraction framework that leverages LLaMA-3.1-70B to extract six semantic dimensions from financial news. Through large-scale experiments on 41,618 news--stock pairs from the FNSPID dataset, we find that (i) FinBERT sentiment features exhibit strong predictive power under nonlinear models (F1=0.576) but substantially weaker performance under linear models (F1=0.230), revealing a highly nonlinear sentiment--return relationship; (ii) LLM-extracted structured features, while individually weaker, capture information orthogonal to sentiment, as evidenced by a 53.5% systematic disagreement rate between the two approaches; and (iii) combining both signal sources yields F1=0.600, significantly outperforming either alone (

p < 0.0001

), with consistent improvements across all seven event types. Ablation experiments confirm that non-sentiment structural dimensions (event type, impact subject, time horizon, confidence) independently contribute

Δ\text{F1} = +0.019

beyond FinBERT alone. Feature importance analysis reveals balanced contributions from all six extracted dimensions (14--21%), demonstrating that compressing news into a single sentiment score incurs substantial information loss. Our results suggest that the sentiment--semantics decoupling in financial text is systematic and exploitable, opening a new direction for multi-dimensional financial NLP.

Abstract

Explore similar work

Beyond Sentiment: Structured Information Extraction from Financial News

Not All News Is Equal: Topic- and Event-Conditional Sentiment from Finetuned LLMs for Aluminum Price Forecasting

PolyBench: Benchmarking LLM Forecasting and Trading Capabilities on Live Prediction Market Data