2026-10-11T20:09:55Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34035662 2026-09-30T17:36:49Z category_7 category_45 category_734 category_931 category_106 category_132 portal_5 item_type_1 month_year_09_2026

<p>Temporal alignment of semantic and numerical signals in an STFA prediction case from the FNSPID dataset.</p> Hong He (221185) Xinhai Li (191514) Medicine Sociology Biological Sciences not elsewhere classified Information Systems not elsewhere classified Science Policy Infectious Diseases three key innovations strictly chronological train shape investor behavior scale temporal fusion relevant historical patterns random walk benchmarks effect size analysis bootstrap confidence intervals aware temporal alignment article without using ablation studies reveal 2022 &# 8211 2013 &# 8211 time fusion architecture preventing information leakage future market information price autocorrelation vanishes scale financial news using large models time series forecasting financial news series models advanced time market reaction level forecasting financial narratives series prediction series data work demonstrates text publication testing ). test splits stfa ), sentiment annotations semantic information return prediction predictive accuracy multimodal sources multimodal baselines liquidity stocks generated independently experiments demonstrate deep learning based sentiment balance accuracy <p>The model predicts a two-day delay between news release and market reaction using only historical information.</p> 2026-09-30T17:36:43Z Image Figure 10.1371/journal.pone.0356139.g006 https://figshare.com/articles/figure/_p_Temporal_alignment_of_semantic_and_numerical_signals_in_an_STFA_prediction_case_from_the_FNSPID_dataset_p_/34035662 CC BY 4.0