Multimodal Language Models with Modality-Specific Experts for Financial Forecasting from Interleaved Sequences of Text and Time Series

By Ross Koval, Nicholas Andrews, Xifeng Yan

Published 2025-09-23

Everscope rating

1435.4

Relevance to quantitative trading

9 / 10

Implementation complexity

7 / 10

Reproducibility

4 / 5

About this paper

Methodology: MSE-ITT (Modality-Specific Experts for Interleaved Text and Time Series) with SALMON (Salience-Aware Language Modeling over Interleaved Modalities). Problem types: Classification, Time Series Forecasting, Natural Language Processing, Portfolio Optimization, Transfer Learning.

arXiv:2509.19628 · Code · Paper rankings

Open the interactive Everscope explorer for full analysis, charts, and paper battles.