WorldCupArena: Fine-Grained Evaluation of Language Models and Deep-Research Agents on Football Forecasting
2026-07-20 • Artificial Intelligence
Artificial IntelligenceComputation and LanguageMachine Learning
AI summaryⓘ
The authors created WorldCupArena, a new tool to test how well language models can predict football match outcomes before the games start. Their system uses current information to forecast match results, scores, key players, events, and stats, then compares predictions to actual outcomes. They tested 13 models over 104 World Cup matches and found that while result accuracy was similar, detailed predictions varied more. The best models only slightly outperformed betting markets and fans on predicting exact scores but did better when close guesses counted. The setup can be reused for future tournaments and is openly available online.
language modelsfootball predictionWorld Cup 2026benchmarkexact-score accuracybetting marketsmodel evaluationdynamic predictionmatch statisticsopen source
Authors
Zhaokai Wang, Tianlin Gui, Jiayuan Rao, Shangzhe Di, Yihong Tang, Dingli Liang
Abstract
Predicting a football match before kickoff requires more than knowing past results: a model must use changing information and make a clear prediction before the answer is available. We present WorldCupArena, a dynamic benchmark for language models and deep-research agents. The 2026 FIFA World Cup is its first evaluation, and the same process can be reused for future leagues and cups. Before each match, a model either receives a common evidence package or searches for information itself. It predicts the result and score, likely players and events, match statistics, and the outcome of the competition. After the match, these predictions are compared with the recorded result. We report result accuracy, exact-score accuracy, and a scoreline score that gives some credit when a predicted score is close but not exact, together with scores for the other prediction tasks. Across 104 matches and 13 systems, models with similar result accuracy differ more clearly on detailed predictions. Compared with betting-market and human-fan baselines, the best system shows only small gains in result and exact-score accuracy, but a clearer gain in Scoreline. New schedules can be added as they begin, allowing the benchmark to evaluate future models without using outcomes that are already known. Code, prompts, predictions, and evaluation scripts are open sourced at https://github.com/wzk1015/WorldCupArena.