The paper introduces SimpleTimeBench, a diagnostic suite testing basic temporal primitives like monotonic trends, periodic signals, and leading indicator covariates. It finds that prominent multivariate Time Series Foundation Models (Chronos‑2, Moirai, and Toto) often produce suboptimal zero‑shot forecasts for these simple patterns, and that fine‑tuning can improve specific tasks while harming performance on other fundamentals. These failures persist in real‑world sensor forecasting, indicating that current TSFMs may lack the inductive biases needed to capture straightforward relationships, thereby limiting their practical reliability.
By Nafiseh Ghoroghchian, Haipeng Zhang, Shuyi Han, Alex Labach, George Stein
arXiv:2503. 24007v4 Announce Type: replace-cross Abstract: In time series forecasting, covariates represent external factors that influence target variables.
By Yosuke Yamaguchi, Issei Suemitsu, Wenpeng Wei
arXiv:2606. 01289v1 Announce Type: new Abstract: Zero-shot time series forecasting aims to predict future values for previously unseen series, requiring models to generalize temporal dynamics beyond the training distribution.
By Yifan Wu, Junjie Wu, Kai Wu, Xiaoyu Zhang, Jian Lou
arXiv:2607. 01204v1 Announce Type: new Abstract: We introduce TiRex-2, a recurrent xLSTM-based time series foundation model that generalizes the univariate TiRex to multivariate forecasting with both past and future covariates.
By Patrick Podest, Marco Pichler, Elias B\"urger, Levente Z\'olyomi, Bernhard Voggenberger, Wilhelm Berghammer, Daniel Klotz, Sebastian B\"ock, G\"unter Klambauer, Sepp Hochreiter
arXiv:2609.06656v1 Announce Type: cross
Abstract: Modern power systems are growing increasingly complex as they integrate diverse generation sources to meet rising demand, making accurate load foreca...
By Varsha Pendyala, Yiwei Fu, Weizhong Yan, Nurali Virani
The paper introduces a hybrid attention model that learns a unified time‑aware patch representation for irregular multivariate time series (IMTS) forecasting. It employs a time‑aware patch encoding to embed variable‑length intra‑patch timestamps, a time bias attention mechanism to adjust for temporal misalignment and asynchronous cross‑channel dependencies, and a hybrid causal mask on a decoder‑only Transformer to balance historical context with autoregressive forecasting. The authors also curate VersaTSA, a 30 B‑observation dataset preserving native sampling sparsity, and demonstrate state‑of‑the‑art zero‑shot performance on three IMTS benchmarks while remaining competitive on regular MTS tasks.
By Zhihao Lin, Li Lin, Qi Zhang, Kaiwen Xia, Shuai Wang, Jialin Qiao
arXiv:2509. 26468v3 Announce Type: replace Abstract: Benchmark quality is critical for meaningful evaluation and sustained progress in time series forecasting, particularly with the rise of pretrained models.
By Oleksandr Shchur, Abdul Fatir Ansari, Caner Turkmen, Lorenzo Stella, Nick Erickson, Pablo Guerron, Michael Bohlke-Schneider, Yuyang Wang
arXiv:2602. 17634v2 Announce Type: replace-cross Abstract: Learning time series foundation models has been shown to be a promising approach for zero-shot time series forecasting across diverse time series domains.
By Xinghong Fu, Yanhong Li, Georgios Papaioannou, Yoon Kim
arXiv:2607. 06504v1 Announce Type: new Abstract: Recent years have witnessed the emergence of multivariate modeling using time series foundation models (TSFMs), which achieve advanced zero-shot generalization.
By Qian Sun, Yong-Ming Tian, Jia-Wei Huang, Cheng Feng, Shao-Qun Zhang
arXiv:2608. 13262v1 Announce Type: cross Abstract: Time series foundation models (TSFMs) have advanced primarily through architectural innovation, while training regimes for large-scale heterogeneous corpora remain under-explored.
By Hongjie Xia, Yiding Liu, Yifan Hu, Peiyuan Liu, Zewei Dong
Aurora‑X is a billion‑parameter time‑series foundation model designed for extreme forecasting tasks. It employs a progressive curriculum that starts with channel‑independent pretraining, then adds cross‑variable dependencies, variable context and horizon lengths, and optional future covariates during mid‑training. A variable‑resolution post‑training stage allows adjustable temporal spans per token at inference, while a pattern‑guided mixture‑of‑experts expands capacity through sparse activation and expert specialization. An implicit quantile network head predicts arbitrary quantiles, enhancing probabilistic forecasting flexibility. Experiments on GIFT‑Eval, TIME, FEV‑Bench, TFB, and DAG‑Bench show state‑of‑the‑art performance against both pretrained TSFMs and task‑specific supervised models.
By Xingjian Wu, Chenjuan Guo, Xiangfei Qiu, Zhigang Hu, Hanyin Cheng, Peng Chen, Yang Shu, Jilin Hu, Bin Yang
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.