Only two of seven LLMs produce positive returns on live Polymarket data, with MiMo-V2-Flash at 17.6% CWR and Gemini-3-Flash at 6.2% CWR while the other five lose money.
hub Mixed citations
A decoder-only foundation model for time-series forecasting
Mixed citation behavior. Most common role is background (60%).
44
Pith papers citing it
42
external citations · Pith
Background
60% of classified citations
abstract
Motivated by recent advances in large language models for Natural Language Processing (NLP), we design a time-series foundation model for forecasting whose out-of-the-box zero-shot performance on a variety of public datasets comes close to the accuracy of state-of-the-art supervised forecasting models for each individual dataset. Our model is based on pretraining a patched-decoder style attention model on a large time-series corpus, and can work well across different forecasting history lengths, prediction lengths and temporal granularities.
hub tools
citation-role summary
background 6
baseline 3
method 1
citation-polarity summary
representative citing papers
B[FM]^2 pretrains an EEG foundation model on raw signals with flow matching and SplitUNet, reaching SOTA on 7 of 9 tasks using ~30x less data and generating neurologist-indistinguishable synthetic EEG.
CloudCons benchmark shows foundation models' superior zero-shot forecasting does not automatically yield better resource consolidation decisions, with predictive quantile choice acting as a key lever for efficiency-reliability trade-offs.
GNSS-FM is a self-supervised foundation model for GNSS displacement time series that outperforms task-specific baselines on 90-day forecasting and seismic step localization after pretraining on global station data.
Hybrid TimesFM plus ridge regression on covariates forecasts 1-MeV electron flux with average R² of 0.9 on out-of-sample 2024 data, outperforming linear regression, CNN, LSTM and Transformer models.
SurF applies the Time Rescaling Theorem as a learnable bijection to create a single generative model for forecasting irregular multivariate event streams that outperforms or matches baselines on six benchmarks.
TimeClaw is an exploratory execution learning system that turns multiple valid tool-use paths into hierarchical distilled experience for improved time-series reasoning without test-time adaptation.
FactoryBench reveals that frontier LLMs achieve under 50% on structured causal questions and under 18% on decision-making in industrial robotic telemetry.
Time series foundation models match the performance of specialized models for day-ahead load forecasting while providing explanations that match domain knowledge on weather and calendar effects.
Chronos pretrains transformer models on tokenized time series to deliver strong zero-shot forecasting across diverse domains.
An agentic forecaster with linguistic belief states, logit-space multi-trial shrinkage, and hierarchical Platt calibration achieves SOTA Brier Index on ForecastBench binary questions.
An LLM-based agentic sampler over building knowledge graphs selects target-specific exogenous variables for zero-shot IoT forecasting, matching or beating trained baselines on three real buildings.
Compares foundation models for probabilistic low-voltage load forecasting on 200 real feeders and introduces a grid-planning metric that scores peak prediction by its effect on asset cost-risk decisions.
Aionoscope shows that time-series representations recover coarse signal types reliably but expose dense latent states like phase and amplitude much less reliably, with best dense-probe R² at 0.689 versus oracle 0.999.
A JEPA-based model with domain-informed multi-view self-distillation learns light-curve representations that outperform hand-crafted features on 15 of 16 StarEmbed metrics and adapts competitively to other irregular time-series datasets.
Presents a fail-closed certification protocol for determining when forecasting leaderboard winners are deployment-actionable, using a traffic dataset to show friction-induced reversals and an audit to prevent overclaiming.
Tyan-WP is a pretrained wind power foundation model that outperforms site-specific TSMs and generic LTSMs in zero-shot ultra-short-term probabilistic forecasting on U.S. and U.K. sites via static embeddings and PAMF module.
GeoGNN is a two-tower GNN that learns geographic cell embeddings from adjacency graphs and matches them to temporal representations via dot-product similarity plus classification, improving geolocalization accuracy by ~27% on electricity datasets.
GITCO delivers +1.95% average MASE reduction on TimesFM 2.5 across 53 datasets by gated inference-time suppression of anomalous patches, capturing 89.9% of the improvement upper bound.
Attentive Neural Processes outperform Gaussian Processes and neural networks on light curve interpolation quality, feature recovery, calibration, and speed for 15 transient classes under realistic Rubin cadences.
ChronoVAE-HOPE proposes a VAE foundation model for time series classification that replaces attention with a HOPE Block dual-memory system and uses disentangled trend-seasonal latent representations, pre-trained on Monash and evaluated on UCR datasets.
MILM fine-tunes LLMs on XML-encoded multimodal irregular time series via a two-stage process that exploits informative sampling patterns to achieve top performance on EHR classification datasets.
RareCP improves interval efficiency for time series conformal prediction by retrieving and weighting regime-specific calibration examples while adapting to drift and maintaining coverage.
S4 models exhibit stable time-continuity unlike sensitive S6 models, with task continuity predicting performance and enabling temporal subsampling for better efficiency.
citing papers explorer
-
PolyBench: Benchmarking LLM Forecasting and Trading Capabilities on Live Prediction Market Data
Only two of seven LLMs produce positive returns on live Polymarket data, with MiMo-V2-Flash at 17.6% CWR and Gemini-3-Flash at 6.2% CWR while the other five lose money.
-
B[FM]$^2$: Brain Foundation Model via Flow Matching with SplitUNet
B[FM]^2 pretrains an EEG foundation model on raw signals with flow matching and SplitUNet, reaching SOTA on 7 of 9 tasks using ~30x less data and generating neurologist-indistinguishable synthetic EEG.
-
CloudCons: A Comprehensive End-to-End Benchmark for Cloud Resource Consolidation
CloudCons benchmark shows foundation models' superior zero-shot forecasting does not automatically yield better resource consolidation decisions, with predictive quantile choice acting as a key lever for efficiency-reliability trade-offs.
-
GNSS-FM: A Self-Supervised Foundation Model for Daily GNSS Displacement Time Series
GNSS-FM is a self-supervised foundation model for GNSS displacement time series that outperforms task-specific baselines on 90-day forecasting and seismic step localization after pretraining on global station data.
-
Forecasting megaelectron-volt electron flux in the Earth's outer radiation belt using supervised machine learning algorithms and a timeseries foundation model
Hybrid TimesFM plus ridge regression on covariates forecasts 1-MeV electron flux with average R² of 0.9 on out-of-sample 2024 data, outperforming linear regression, CNN, LSTM and Transformer models.
-
SurF: A Generative Model for Multivariate Irregular Time Series Forecasting
SurF applies the Time Rescaling Theorem as a learnable bijection to create a single generative model for forecasting irregular multivariate event streams that outperforms or matches baselines on six benchmarks.
-
TimeClaw: A Time-Series AI Agent with Exploratory Execution Learning
TimeClaw is an exploratory execution learning system that turns multiple valid tool-use paths into hierarchical distilled experience for improved time-series reasoning without test-time adaptation.
-
FactoryBench: Evaluating Industrial Machine Understanding
FactoryBench reveals that frontier LLMs achieve under 50% on structured causal questions and under 18% on decision-making in industrial robotic telemetry.
-
Explainable Load Forecasting with Covariate-Informed Time Series Foundation Models
Time series foundation models match the performance of specialized models for day-ahead load forecasting while providing explanations that match domain knowledge on weather and calendar effects.
-
Chronos: Learning the Language of Time Series
Chronos pretrains transformer models on tokenized time series to deliver strong zero-shot forecasting across diverse domains.
-
Agentic Forecasting using Sequential Bayesian Updating of Linguistic Beliefs
An agentic forecaster with linguistic belief states, logit-space multi-trial shrinkage, and hierarchical Platt calibration achieves SOTA Brier Index on ForecastBench binary questions.
-
TopoBrick: Agentic Topology Sampling of Exogenous Variables for Zero-Shot Building IoT Forecasting
An LLM-based agentic sampler over building knowledge graphs selects target-specific exogenous variables for zero-shot IoT forecasting, matching or beating trained baselines on three real buildings.
-
Probabilistic Low-Voltage Peak Load Forecasting with Time Series Foundation Models Evaluated on Application-Oriented Metrics
Compares foundation models for probabilistic low-voltage load forecasting on 200 real feeders and introduces a grid-planning metric that scores peak prediction by its effect on asset cost-risk decisions.
-
Aionoscope: Debugging Latent-State Accessibility in Time-Series Representations
Aionoscope shows that time-series representations recover coarse signal types reliably but expose dense latent states like phase and amplitude much less reliably, with best dense-probe R² at 0.689 versus oracle 0.999.
-
Domain-Informed Multi-View Self-Distillation for Astronomical Light-Curve Representation Learning with JEPA
A JEPA-based model with domain-informed multi-view self-distillation learns light-curve representations that outperform hand-crafted features on 15 of 16 StarEmbed metrics and adapts competitively to other irregular time-series datasets.
-
From Forecasting Leaderboards to Deployment Decisions: A Fail-Closed Certification Protocol
Presents a fail-closed certification protocol for determining when forecasting leaderboard winners are deployment-actionable, using a traffic dataset to show friction-induced reversals and an audit to prevent overclaiming.
-
Tyan-WP: A Wind Power Foundation Model for Ultra-Short-Term Probabilistic Forecasting
Tyan-WP is a pretrained wind power foundation model that outperforms site-specific TSMs and generic LTSMs in zero-shot ultra-short-term probabilistic forecasting on U.S. and U.K. sites via static embeddings and PAMF module.
-
GeoGNN: Time Series Geo-Localization using Two-Tower Graph Neural Networks
GeoGNN is a two-tower GNN that learns geographic cell embeddings from adjacency graphs and matches them to temporal representations via dot-product similarity plus classification, improving geolocalization accuracy by ~27% on electricity datasets.
-
GITCO: Gated Inference-Time Context Optimization in TSFMs
GITCO delivers +1.95% average MASE reduction on TimesFM 2.5 across 53 datasets by gated inference-time suppression of anomalous patches, capturing 89.9% of the improvement upper bound.
-
Probabilistic Data-Driven Modelling of Astrophysical Transients: The Neural Process Family for Ultrafast and Class-Agnostic Light Curve Reconstruction with NightLANP
Attentive Neural Processes outperform Gaussian Processes and neural networks on light curve interpolation quality, feature recovery, calibration, and speed for 15 transient classes under realistic Rubin cadences.
-
ChronoVAE-HOPE: Beyond Attention -- A Next-Generation VAE Foundation Model for Specialized Time Series Classification
ChronoVAE-HOPE proposes a VAE foundation model for time series classification that replaces attention with a HOPE Block dual-memory system and uses disentangled trend-seasonal latent representations, pre-trained on Monash and evaluated on UCR datasets.
-
MILM: Large Language Models for Multimodal Irregular Time Series with Informative Sampling
MILM fine-tunes LLMs on XML-encoded multimodal irregular time series via a two-stage process that exploits informative sampling patterns to achieve top performance on EHR classification datasets.
-
RareCP: Regime-Aware Retrieval for Efficient Conformal Prediction
RareCP improves interval efficiency for time series conformal prediction by retrieving and weighting regime-specific calibration examples while adapting to drift and maintaining coverage.
-
Continuity Laws for Sequential Models
S4 models exhibit stable time-continuity unlike sensitive S6 models, with task continuity predicting performance and enabling temporal subsampling for better efficiency.
-
FETS Benchmark: Foundation Models Enable Scalable and Generalizable Energy Time Series Forecasting
Covariate-informed zero-shot time-series foundation models beat task-specifically tuned XGBoost and random forests in aggregate on a 54-dataset energy forecasting benchmark.
-
Predicting Power-System Dynamic Trajectories with Foundation Models
LASS-ODE-Power is a pretrained model that predicts power-system dynamic trajectories across regimes in a zero-shot manner after large-scale ODE pretraining and targeted fine-tuning.
-
MICA: Multivariate Infini Compressive Attention for Time Series Forecasting
MICA adapts infini compressive attention to the channel dimension, enabling scalable cross-channel dependencies in Transformers and cutting forecast error by 5.4% on average versus channel-independent baselines.
-
Dynamic Linear Coregionalization for Realistic Synthetic Multivariate Time Series
DynLMC creates synthetic time series data with dynamic inter-channel correlations that improve zero-shot forecasting in foundation models across multiple benchmarks.
-
Timer-S1: A Billion-Scale Time Series Foundation Model with Serial Scaling
Timer-S1 is a released 8.3B-parameter MoE time series model that achieves state-of-the-art MASE and CRPS scores on GIFT-Eval using serial scaling and Serial-Token Prediction.
-
TelecomTS: A Multi-Modal Observability Dataset for Time Series and Language Analysis
TelecomTS is a de-anonymized 5G observability benchmark with scale information, anomalies, and multi-modal QA tasks, on which current foundation models perform poorly.
-
General Geospatial Inference with a Population Dynamics Foundation Model
A GNN-based foundation model on aggregated US geospatial data produces embeddings achieving SOTA on all 27 interpolation tasks and 25/27 extrapolation/super-resolution tasks across health, socioeconomic and environmental domains, plus improved forecasting when combined with TimesFM.
-
Zeus: Towards Tuning-Free Foundation Model for Time Series Analysis
Zeus proposes a multi-scale Transformer with point-wise tokenization and Multi-Objective Temporal Masking to enable tuning-free performance on forecasting, interpolation, and other time series tasks.
-
A Three-Phase Foundation Model for Tax-Aware Personalized Portfolio Management
A deep-RL portfolio system with ticker-agnostic encoding, six-objective MoE policy, and LoRA personalization reports +2.93% 14-day alpha vs equal-weight, but the evidence is a short, in-sample, cost-free backtest.
-
Pretrained Time-Series Foundation Models for Financial Return Forecasting
Pretrained TSFMs achieve top ranks on equity return tasks but show sparse, minimal improvements over random walk, serving as practical priors without reliable alpha generation.
-
Public transit gains and spatially uneven travel demand changes after NYC congestion pricing
Time series foundation models generate counterfactual forecasts showing increased transit ridership and reduced aggregate travel after NYC congestion pricing, with spatial and demographic heterogeneity.
-
When Do Autoregressive Sequence Models Forecast Physical Wavefields? A Controlled Study on Synthetic Seismograms
Multi-token prediction accounts for nearly all rollout stability gains on synthetic three-component seismograms, with sharp dependence on context covering the full P-S interval and magnitude-based losses unable to prevent polarity flips.
-
Toto 2.0: Time Series Forecasting Enters the Scaling Era
Time series foundation models scale under a single training recipe, with forecast quality improving from 4M to 2.5B parameters and new SOTA results on BOOM, GIFT-Eval, and TIME benchmarks.
-
Quantifying the Pre-training Dividend: Generative versus Latent Self-Supervised Learning for Time Series Foundation Models
Self-supervised pre-training delivers large gains up to 375% on time series anomaly detection and classification but only marginal benefits for forecasting, driven by a precision-invariance trade-off in the learned representations.
-
A Quantum Inspired Variational Kernel and Explainable AI Framework for Cross Region Solar and Wind Energy Forecasting
A hybrid classical-plus-quantum-inspired framework for cross-region renewable energy forecasting matches top baselines within 1% accuracy and separates calm versus stormy conditions with a 15-fold higher Fisher discriminant ratio than a tuned radial basis kernel.
-
Degradation-aware Predictive Energy Management for Fuel Cell-Battery Ship Power System with Data-driven Load Forecasting
A degradation-aware predictive controller for hybrid ship power systems reduces hydrogen consumption by up to 5.8% and fuel cell degradation by up to 36.4% versus a filter-based benchmark on real harbor tug data.
-
Does Normalization Choice Matter for Causal Large Time-Series Models?
Normalization choice significantly influences training convergence and forecasting performance in causal large time-series models.
-
Assessing the Operational Viability of Foundation Models for Time Series Forecasting
Foundation models match or approach supervised performance in periodic and cold-start domains but lag in physically constrained systems, while a feature-based router improves accuracy and cuts inference cost versus always using one model class.
-
KairosHope: A Next-Generation Time-Series Foundation Model for Specialized Classification via Dual-Memory Architecture
KairosHope proposes a HOPE block with dual-memory (Titans + CMS) and hybrid statistical-deep decision head for TSFM classification, pre-trained via MTSM and InfoNCE on Monash then adapted via LP-FT to UCR, claiming superior results on causal domains.
-
Combining Statistical Features and Deep Encodings for Rehearsal-Based Class-Incremental Time Series Classification
Dual-stream statistical plus deep-embedding pipeline with rehearsal yields competitive accuracy and low forgetting on five time-series benchmarks for class-incremental classification.