RMISC: A Large-scale Real-world Multivariate Corpus for Time Series Foundation Models

2026-07-07

AI Summary

This work addresses the limitations of existing time series foundation models, which predominantly rely on synthetic multivariate data for pretraining and struggle to capture the complex temporal dynamics and inter-variable relationships inherent in real-world scenarios. To bridge this gap, we introduce RMISC—the first large-scale, high-quality, and openly available corpus of real multivariate time series—comprising approximately 200 datasets and 14.2 billion time points. Leveraging this corpus, we pretrain four state-of-the-art foundation models on univariate, synthetic multivariate, and real multivariate data, respectively, and systematically evaluate their zero-shot generalization across in-distribution and out-of-distribution benchmarks. Our experiments demonstrate that incorporating real multivariate data substantially enhances zero-shot performance on both univariate and multivariate tasks, underscoring the critical role of authentic data in improving model generalization.

Abstract

Recent years have witnessed the emergence of multivariate modeling using time series foundation models (TSFMs), which achieve advanced zero-shot generalization. Modern multivariate TSFMs are predominantly pretrained on multivariate synthetic data, which is easier to scale but may fail to capture the complex temporal dynamics and cross-variable relationships present in real-world time series. This raises a key question: Whether and to what extent the leading TSFMs trained with the real-world corpus perform better than those trained with synthetic data? To answer this, we establish the RMISC corpus, a considerably large-scale, high-quality, openly accessible, real-world, and multivariate time series archive that contains around 200 datasets and 142 billion time points across diverse domains. Furthermore, we pretrain four advanced TSFMs on univariate, synthetic multivariate, and real-world multivariate data and evaluate their zero-shot generalization capabilities on standard in-distribution and out-of-distribution benchmarks. Experimental results show that incorporating real-world multivariate data predominantly improves the generalization performance for both univariate and multivariate TSFMs. These results provide a deeper understanding of how real-world multivariate data contributes to the development of stronger TSFMs.

Problem

• time series foundation models
• real-world multivariate data
• synthetic data
• zero-shot generalization
• temporal dynamics

Innovation

• time series foundation models
• real-world multivariate data
• zero-shot generalization
• large-scale corpus
• pretraining