RAM AI is pleased to share a new research paper by Tian Guo, Senior Quant, and Emmanuel Hauptmann, CIO & Head of Systematic. The paper examines how to combine quantitative factors with newsflow representations derived from Large Language Models (LLMs) to improve stock return prediction and selection.
It studies two main approaches:
(1) fusion learning, across architectures of varying complexity
(2) mixture models, under different training schemes
Building on our EMNLP 2024 paper, ‘Fine-Tuning Large Language Models for Stock Return Prediction Using Newsflow,’ this work provides empirical and theoretical insights into multimodal data fusion, mixture modelling, and training strategies for systematic investing.
