Time Seriestime-series-forecasting

Time Series Forecasting

Time-series forecasting exploded in 2023-2025 when foundation models crossed over from NLP. Nixtla's TimeGPT (2023), Google's TimesFM (2024), and Amazon's Chronos showed that a single pretrained model can zero-shot forecast diverse series, rivaling task-specific statistical models like ETS and ARIMA. Yet the Monash benchmark and M-competition lineage (M4, M5) reveal an uncomfortable truth: simple ensembles of statistical methods still win on many univariate tasks. The real battle now is multivariate long-horizon forecasting, where PatchTST and iTransformer compete with state-space models like Mamba.

6
Datasets
39
Results
smapi
Canonical metric
Canonical Benchmark

M4 Competition

100,000 time series from diverse domains (finance, demographic, macro, micro, industry, other). Competition ran in 2018. Lower sMAPE/MASE/OWA is better.

Primary metric: smapi
View full leaderboard

Top 10

Leading models on M4 Competition.

RankModelsmapiYearSource
1
TiDE
13.92026paper
2
DLinear
13.62026paper
3
PatchTST
13.22026paper
4
Autoformer
12.92026paper
5
FEDformer
12.82026paper
6
iTransformer
12.72026paper
7
N-HiTS
11.92026paper
8
LMS-AutoTSF
11.92026paper
9
N-BEATS
11.92026paper
10
TimesNet
11.82026paper

All datasets

6 datasets tracked for this task.

Related tasks

Other tasks in Time Series.

Run Inference

Looking to run a model? HuggingFace hosts inference for this task type.

HuggingFace