Uni-FinLLM: A Unified Multimodal Large Language Model with Modular Task Heads for Micro-Level Stock Prediction and Macro-Level Systemic Risk Assessment

TL;DR

Uni-FinLLM is a unified multimodal large language model that integrates diverse financial data types to improve micro- and macro-level risk predictions, outperforming existing methods in accuracy.

Contribution

It introduces a shared Transformer backbone with modular task heads for joint processing of multimodal financial data, capturing cross-scale dependencies for comprehensive risk assessment.

Findings

Stock prediction accuracy improved to 67.4%.

Credit-risk assessment accuracy increased to 84.1%.

Systemic risk early-warning accuracy reached 82.3%.

Abstract

Financial institutions and regulators require systems that integrate heterogeneous data to assess risks from stock fluctuations to systemic vulnerabilities. Existing approaches often treat these tasks in isolation, failing to capture cross-scale dependencies. We propose Uni-FinLLM, a unified multimodal large language model that uses a shared Transformer backbone and modular task heads to jointly process financial text, numerical time series, fundamentals, and visual data. Through cross-modal attention and multi-task optimization, it learns a coherent representation for micro-, meso-, and macro-level predictions. Evaluated on stock forecasting, credit-risk assessment, and systemic-risk detection, Uni-FinLLM significantly outperforms baselines. It raises stock directional accuracy to 67.4% (from 61.7%), credit-risk accuracy to 84.1% (from 79.6%), and macro early-warning accuracy to 82.3%. Results validate that a unified multimodal LLM can jointly model asset behavior and systemic vulnerabilities, offering a scalable decision-support engine for finance.