Newest papers

From Word Counts to Context: Topic Models for Asset Pricing

Kevin Foley, Jonathan Hartadi, Shivesh Prakash, Swapnil Vatsal · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2609.34169 · cited by 0 works (OpenAlex) · open-access PDF

French

Nous cherchons à tester si l'utilisation d'un transformateur de phrase représente une amélioration par rapport aux techniques telles que l'allocation latente de dirichlet (LDA) dans la cohérence des listes de termes thématiques générées à partir de données textuelles non structurées. Pour tester cela, la même collection de données textuelles non structurées comprenant 394,661 articles et le même pipeline de construction de portefeuille financier en aval ont été appliqués avec la couche de texte différente, y compris la longueur du texte de l'article de chaque modèle utilisé et la façon dont les termes thématiques ont été classés: nous comparons LDA contre un transformateur de phrase gelé avec k-me clanstering. Nous constatons que la branche de transformateur de phrase avait observé des scores plus

Machine translation (Meta M2M100); the English original follows.

Abstract

News may reveal systematic risk, but whether its context enhances the construction of systematic risk factors is still unclear. We seek to test whether utilizing a sentence transformer represents an improvement over techniques such as Latent Dirichlet Allocation (LDA) in the coherence of topic term lists generated from unstructured text data. To test this, the same collection of unstructured text data comprising of 394,661 articles and the same downstream financial portfolio construction pipeline were applied with the text layer differing, including the length of article text each model used and how topic terms were ranked: we benchmark LDA against a frozen sentence transformer with k-means clustering. We find that the sentence transformer branch had higher observed scores both in terms of coherence (measured by NPMI) as well as financial performance (measured by Sharpe), although the available tests do not establish outperformance. Further exploratory specifications such as utilizing spherical clustering and multi-horizon exposures had an observed excess-return Sharpe of 1.03 for the combined model. We believe that there is some promise in applying context-aware techniques on unstructured news text, but stricter tests using only information available at each date and broader datasets may be required to enhance the confidence in the observed performance.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Latent Dirichlet allocation, Topic model, Computer science, Portfolio, Sentence, Transformer

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange