About TS-Arena

A live benchmarking platform for Time Series Foundation Models using forecast pre-registration.

Research Paper

TS-Arena: A Live Forecast Pre-Registration Platform

Marcel Meyer, Sascha Kaltenpoth, Henrik Albers, Kevin Zalipski, Oliver Müller — Paderborn University, Data Analytics Group

Abstract. TS-Arena is a live benchmarking platform that evaluates Time Series Foundation Models (TSFMs) by requiring forecast submissions before ground-truth data exists — a “forecast pre-registration protocol.” This design eliminates test-set contamination and information leakage, since the evaluation target physically does not exist at submission time. The platform continuously collects forecasts from models across 186 energy-sector time series in 14 challenge definitions, scores them with MASE, and ranks them using an ELO rating system with confidence intervals.
Read on arXiv

How It Works

TS-Arena runs continuously scheduled forecasting challenges on real-world energy data. When a new challenge round opens, models have a registration window to submit their forecasts for a future time period. Once the ground truth becomes available, submitted forecasts are automatically evaluated and rankings are updated.

Pre-Registration

Forecasts must be submitted before ground truth exists, making data leakage structurally impossible.

MASE Scoring

Mean Absolute Scaled Error provides scale-independent accuracy scores comparable across time series.

ELO Ranking

Pairwise ELO ratings with confidence intervals enable fair comparison between models over time.

Data & Challenges

The benchmark covers 186 energy-sector time series from multiple European and North American grid operators (SMARD, EIA, Fingrid, ENTSO-E, GridStatus), organized into 14 challenge definitions with varying forecast frequencies (15 min, 1 h) and horizons (1 day, 1 week). Challenges span electricity consumption and generation, providing diverse conditions for a thorough model evaluation.

Team

TS-Arena is developed by the Data Analytics Group at Paderborn University, Germany. Learn more on the team page.

We are open to collaboration, for example to integrate additional live time series into TS-Arena. If you are interested, please contact us at DataAnalytics@wiwi.uni-paderborn.de.

Marcel MeyerSascha KaltenpothHenrik AlbersKevin ZalipskiProf. Dr. Oliver Müller