FinZero: Launching Multi-modal Financial Time Series Forecasting with Large Reasoning Model

Yanlong Wang, Jian Xu, Fei Ma, Hongkang Zhang, Hang Yu, Tiantian Gao, Wang Yu, Haochen You, Shao-Lun Huang, Danny Dongning Sun, Xiao-Ping Zhang

Abstract: Financial time series forecasting is both highly significant and challenging. Previous approaches typically standardized time series data before feeding it into forecasting models, which inherently leads to a loss of important information during the encoding process. Moreover, past time series models generally require fixed numbers of variables or lookback window lengths, which further limits the scalability of time series forecasting. In addition, interpretability and uncertainty in forecasting remain underexplored, as these factors directly impact the reliability and practical value of predictions. To address these issues, we first constructed a diverse financial image-text dataset (FVLDB) and developed the Uncertainty-adjusted Group Relative Policy Optimization (UARPO) method to enable the model not only output predictions but also analyze the uncertainty of those predictions. We then launched FinZero, a multi-modal pre-trained model finetuned by UARPO to perform reasoning, prediction, and analytical understanding on the FVLDB financial time series. Extensive experiments validate that FinZero exhibits strong adaptability and scalability. After fine-tuning with UARPO, FinZero achieves an approximate 13.48% improvement in prediction accuracy over GPT-4o in the high-confidence group, demonstrating the effectiveness of reinforcement learning fine-tuning of multi-modal large model for financial time series forecasting tasks.