Pith. sign in

REVIEW 3 major objections 6 minor 44 references

WorldCupArena sets out to show that evaluating football forecasts needs fine-grained, pre-match predictions: across 104 World Cup matches, models with nearly identical result accuracy separate sharply on scorelines, lineups, events, and sta

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

A pre-registered benchmark of 13 LLM systems on all 104 World Cup 2026 matches shows fine-grained predictions expose differences that result accuracy hides.

T0 review reviewed 2026-08-01 challenge →

load-bearing objection A genuinely reusable pre-registered benchmark for fine-grained football forecasting, with a headline claim that is a bit ahead of the evidence. the 3 major comments →

arxiv 2607.18084 v1 pith:5KF53T53 submitted 2026-07-20 cs.AI cs.CLcs.LG

WorldCupArena: Fine-Grained Evaluation of Language Models and Deep-Research Agents on Football Forecasting

Zhaokai Wang , Tianlin Gui , Jiayuan Rao , Shangzhe Di , Yihong Tang , Dingli Liang This is my paper
classification cs.AI cs.CLcs.LG
keywords football forecastinglarge language modelsdeep research agentsdynamic benchmarkpre-registered predictionscoreline metricWorld Cup 2026fine-grained evaluation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that a model's football forecasting skill cannot be read off result accuracy alone. To show this, it builds WorldCupArena, a benchmark that freezes each model's forecast—result, score, lineups, events, statistics, and full competition path—24 hours before kickoff and scores it only after the official record exists. Across all 104 matches of the 2026 World Cup and 13 systems, result accuracy differed little between many systems, while the finer layers separated them clearly; the top system's clearest edge over betting-market and human-fan baselines was a much higher scoreline score, meaning its misses were closer rather than its hits more frequent. The same locked-prediction protocol is reusable for future leagues and cups, so models released later can be tested on genuinely unknown outcomes. A sympathetic reader would care because it supplies a method—and evidence—for measuring whether AI forecasts are useful before an event, not just accurate after the fact.

Core claim

On the paper's own terms, the discovery is that fine-grained pre-match predictions carry signal that coarse result accuracy hides. Over all 104 matches and 13 systems, result accuracy fell in a narrow band while detailed scores varied widely; the leading system's clearest advantage over betting-market and human-fan baselines was not more strict hits but a much higher scoreline score, meaning its wrong scoreline predictions were consistently closer. At the competition level, four systems predicted champion Spain, and only two also produced the exact Spain–Argentina final pairing.

What carries the argument

The load-bearing mechanism is the prediction lock plus the five-layer scoring taxonomy. Every forecast is frozen 24 hours before kickoff together with the evidence snapshot and, for search-enabled agents, the retrieved source URLs; scoring happens only after the official record exists. The five layers—result and score, players and lineups, events, tactics and statistics, and competition outcome—are combined with fixed weights, and missing truth fields are excluded rather than scored as zeros. The scoreline score does the differentiating work: it awards 100 for an exact score and otherwise gives partial credit for a correct result class and small errors in goal difference, total goals, and te

Load-bearing premise

The entire pre-match validity rests on the assumption that no system saw the final outcome before its prediction was locked, which depends on imperfect publication-date checks and manual review of search sources.

What would settle it

Audit the publicly saved source URLs and snapshots for the search-enabled systems: if any page retrieved before the lock already contains the final score or a post-match report, the temporal guarantee is broken; a simple diff against the official match record would reveal it.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Result accuracy alone is an incomplete measure of forecasting skill: systems that pick the same number of winners can be far apart on scorelines, lineups, events, and statistics.
  • Exact-score prediction remains very hard for all systems (roughly 10 to 17 percent), and near-miss credit is what separates the leading systems from betting-market and fan baselines.
  • Adding web search did not consistently improve match forecasting in the systems compared; observed changes on shared matches ranged from slightly negative to moderately negative.
  • Competition-level evaluation adds signal: four systems predicted the champion, but only two also predicted the exact final pairing, distinguishing a correct champion reached through the right bracket from one reached via the wrong path.
  • The four-step protocol—save evidence and prediction before kickoff, score after the official record—is reusable for future leagues and cups and for models released after the 2026 World Cup.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Inference: If locked-forecast scoring generalizes, the same design could expose hidden differences in other domains where a coarse category is the usual metric—election outcomes, economic releases, or injury reports—where a near miss is informative.
  • Inference: The finding that search does not consistently help is a statement about current commercial systems, not a law; a controlled comparison holding the base model fixed while varying only the search tool would test whether better search evidence ever pays off.
  • Inference: The consensus failures on the same upsets suggest that model diversity, not just average quality, matters for forecasting; ensembling diverse models might avoid the misses that a single strong system cannot.
  • Inference: Because the scoreline metric deliberately rewards closeness, it could be used as a training or selection signal—optimizing for scoreline quality may yield forecasts more useful to bettors or planners, who care about the size of the miss, not just whether the favorite won.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. WorldCupArena introduces a dynamic, pre-event benchmark for evaluating language models and deep-research agents on football forecasting. For all 104 matches of the 2026 FIFA World Cup, models either receive a common evidence package (S1) or search for evidence themselves (S2); they predict result, score, lineups, events, statistics, and the full competition, with predictions locked 24 hours before kickoff. The paper reports result accuracy, exact-score accuracy, a Scoreline metric that rewards close misses, and layer-level scores, and compares 13 systems against betting-market and human-fan baselines. The central claim is that models with similar result accuracy differ more clearly on fine-grained predictions, and that the benchmark reveals capabilities not captured by coarse result accuracy. The paper also reports an in-play forecasting track and an open-source artifact release.

Significance. If the central claim is sustained, the benchmark is a useful and reusable contribution: it evaluates genuinely pre-event forecasting, distinguishes result accuracy from finer-grained predictive skill, and provides a protocol that can be applied to future leagues. The paper has substantial strengths: saved predictions and evidence, availability-aware aggregation that distinguishes missing data from zero outcomes, explicit leakage checks as part of the pipeline, a published scoring/evaluation codebase, and an honest discussion of implementation boundaries. These are real contributions to evaluation methodology for LLM forecasting. However, the headline quantitative claims are not yet fully supported: the main comparisons lack confidence intervals and raw Scoreline values, the display calibration is applied after observing the score distribution, and the temporal-leakage safeguard is acknowledged to be imperfect without a documented systematic audit. These issues are fixable and do not, in my view, invalidate the benchmark design.

major comments (3)
  1. [§3.1 and §A.5] The benchmark's central validity guarantee is that all predictions were made before kickoff and contain no outcome-revealing information. The paper states in §3.1 that a prediction is excluded if it contains information published after the deadline, but §A.5 concedes that 'search-source publication times are also imperfect, so automatic leakage checks need periodic manual review.' The manuscript does not document any systematic manual audit of the 104-match, 13-system corpus. For S2 systems, a single missed post-deadline source could inflate Scoreline, player, or event scores and create the kind of fine-grained separation the headline claims. I am not asserting that leakage occurred; I am saying the central claim rests on an unverified assumption. Please provide a leak-audit trail: number of runs flagged and excluded per system, the manual-review protocol actually followed, and a sensiti
  2. [§5.2, Table 2] The headline comparisons are reported without confidence intervals, raw Scoreline values, or paired tests. For example, the 15.14 display-point Scoreline gap between Claude Opus 4.7 (Thinking + Search) and BetVictor compares a 58-match system with a 104-match baseline; the result-accuracy gap is only 2.4 points. Similarly, several systems in Table 2 have result accuracies of 65–68% and Scoreline values spanning 36.95–68.49, but no uncertainty is attached to these numbers. The paper could be over-reading noise. The authors state that comparisons are 'descriptive rather than paired,' but the phrase 'models with similar result accuracy differ more clearly on detailed predictions' is a comparative claim that needs statistical support. Please report raw Scoreline means and per-system standard errors or confidence intervals, and perform paired shared-match analyses for the S1/S2 contrasts and
  3. [§4.2, Eq (3)] The display calibration in Eq (3) is chosen after observing the distribution of raw scores and is applied to the Scoreline values in Table 2. The example in the text shows that raw scores 44 and 47 become displayed scores 23.15 and 35.43: a 3-point raw gap becomes a 12.28-point displayed gap. This nonlinear expansion can make a small raw difference look like a 'clearer gain' in Scoreline. The transformation is fixed, but the center and temperature are free parameters selected on the basis of the observed result cluster. Because the central 'clearer gain' claim is expressed in display-calibrated Scoreline values, please report the raw Scoreline numbers throughout the leaderboard and provide a sensitivity analysis showing that the main ordering and the headline gaps are stable under reasonable alternative calibration choices.
minor comments (6)
  1. [§4.1, Eq (1)] The variables e_d, e_t, and e_team are used in Eq (1) but not formally defined. Please define them explicitly (e.g., absolute error in goal difference, total goals, and team-wise goals) in one sentence before or after the equation.
  2. [§4.2 and Table 2] It is easy to misread Table 2's Scoreline column as a raw mean rather than a display-calibrated value. Please add a note under the table or in the column header stating that Scoreline is reported after the display calibration in Eq (3), and that raw values are available in the artifact.
  3. [§5.2, RQ5] The comparison with the human-fan baseline uses 94 matches for fans and 58 for the best Claude system, and the Polymarket baseline has no Scoreline value in 39 fixtures because of 'Any Other Score.' Please make these coverage differences more prominent, since they affect the interpretation of the baseline gaps.
  4. [§5.2, RQ3] The shared-match analysis is reported only in the text as three numbers. Consider a small table or figure showing the shared-match comparison for the three models, including the number of shared matches, so readers can verify the direction and magnitude of the search effect.
  5. [§A.2] The sentence 'Champion, runner-up, third place, and top scorers are derived from the saved match list. rather than accepted as unconstrained independent claims.' contains a misplaced period before 'rather' and reads awkwardly. Please rephrase.
  6. [§5.1] The Polymarket baseline retrieves prices in the window [T−24h, T), while other baselines use the latest stored pre-match odds. Please clarify whether this timing difference matters for the reported result-accuracy comparison, especially for matches with late news.

Circularity Check

0 steps flagged

No significant circularity: model predictions are external, scored against official records and external baselines; the central comparison does not reduce to its inputs.

full rationale

The paper's derivation chain is not circular by construction. Every model prediction is saved before kickoff (Section 3.1) and scored against independently collected official match and competition records. Result accuracy, exact-score accuracy, and Scoreline are computed from the same saved predictions, but they are not derived from one another: Eq. (1) defines Scoreline as a function of the distance between the predicted and actual score, while result accuracy and exact-score accuracy are separate hit-based metrics. The headline claim that models with similar result accuracy differ more clearly on detailed predictions (RQ1, Section 5.2) is an empirical observation about the separately computed columns in Table 2, not a fitted or definitional relation. Baselines (Polymarket, BetVictor, human fans) are external to the paper's model outputs, so the clearest claimed gain (Scoreline 68.49 vs 53.35 for BetVictor and 53.40 for human fans, Section 5.2 RQ5) is anchored outside the paper's own prediction pipeline. The fixed contrast calibration in Eq. (3) is a disclosed, order-preserving display transformation; even if its parameters were chosen after observing raw scores cluster near 50, it changes displayed spacing but not ranking or the direction of any gain. The authors' self-citations [33, 34] appear only in Related Work as examples of completed-event sports benchmarks and are not load-bearing for the central claims. Appendix A.5 states 'Search-source publication times are also imperfect', which is a temporal-contamination and data-quality risk, not circularity: it bears on whether predictions were truly pre-match, not whether any claimed result is equivalent to its inputs. No fitted parameter is renamed as a prediction, no uniqueness or self-citation chain forces the conclusions, and no known empirical pattern is merely relabeled as a new result. Verdict: no significant circularity.

Axiom & Free-Parameter Ledger

3 free parameters · 5 axioms · 0 invented entities

The central claims rest on external football rules, single-source truth data, temporal leakage integrity, stable model identities, and the authors' hand-tuned scoring metrics. The last two are acknowledged in the paper as imperfect or arbitrary, and the scoring metrics directly shape the headline Scoreline comparison.

free parameters (3)
  • Scoreline metric coefficients (Eq 1) = w_r=45, w_d=25, w_t=20, w_team=10; decay denominators 5, 6, 8
    Hand-chosen weights/decay rates determine how much partial credit near-miss scorelines receive; the paper's headline 'clearer gain in Scoreline' is measured with this specific metric, and raw Scoreline values are not reported.
  • Display contrast calibration (Eq 3) = center=50, temperature=5 for composite; center=70, temperature=5 for Scoreline display
    Chosen after observing raw scores cluster around 50 (Section 4.2); the transformation is rank-preserving but the reported 15.14-point Scoreline gap between Claude and BetVictor is a display value, not raw, and depends on these parameters.
  • Layer and task weights (Table 1) = T1=0.40, T2=0.20, T3=0.15, T4=0.15, T5=0.10; internal task weights as in Table 1
    Hand-set to emphasize match result; composite scores and rankings depend on these weights.
axioms (5)
  • domain assumption FIFA 2026 World Cup schedule, format (72 group + 32 knockout matches), and tie-break/best-third-place rules as given in references [10, 11].
    The benchmark's 104-match structure and tournament generation depend on these external rules.
  • domain assumption Official match records (score, lineups, events, statistics, standings) are accurate ground truth; the paper notes the truth adapter depends mainly on one football data provider and is not cross-checked against a second source (Appendix A.5).
    All reported accuracies are computed against these records.
  • domain assumption The 24-hour pre-kickoff lock prevents leakage of outcome information; the paper concedes publication dates for web sources are imperfect and automatic leakage checks require manual review (Appendix A.5).
    The validity of 'pre-match' prediction depends on this temporal integrity.
  • domain assumption Commercial model versions are stable under the same public names; the paper acknowledges providers may silently update models (Section 6).
    The comparison treats a named system as a fixed artifact across the tournament.
  • ad hoc to paper The hand-designed scoring equations (Eq 1, 3, 4; Table 1) are accepted as the evaluation definition without external justification.
    No argument is provided that these weights/calibrations are canonical or robust to reasonable variations; the Scoreline and composite numbers are defined by these choices.

reviewed 2026-08-01 · how reviews work

0 comments
Cite this review

Pith. "Pith review of WorldCupArena: Fine-Grained Evaluation of Language Models and Deep-Research Agents on Football Forecasting." pith.science (2026). https://pith.science/paper/5KF53T53

@misc{pith2026260718084,
  author       = {Pith},
  title        = {Pith review of: WorldCupArena: Fine-Grained Evaluation of Language Models and Deep-Research Agents on Football Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5KF53T53}},
  note         = {Machine review of arXiv:2607.18084}
}
Share X Bluesky LinkedIn Reddit HN
read the original 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.

Figures

Figures reproduced from arXiv: 2607.18084 by Dingli Liang, Jiayuan Rao, Shangzhe Di, Tianlin Gui, Yihong Tang, Zhaokai Wang.

Figure 1
Figure 1. Figure 1: Visual taxonomy of the five evaluation layers, moving from a match result and score through players and lineups, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of WorldCupArena. Match information and model predictions are saved before kickoff; the final match and [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Scores of leading systems for result and score (T1), players and lineups (T2), events and timeline (T3), tactics and [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Claude Opus 4.7 (Thinking)’s prediction for Argentina–Switzerland beside the recorded result. The example shows [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Summary of match and competition-level results over all 104 World Cup matches. The right panel includes the [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

44 extracted references · 4 canonical work pages

  1. [1]

    Anthropic. 2026. Claude Opus 4.7. https://www.anthropic.com/news/claude- opus-4-7. Accessed 2026-07-16

    2026

  2. [2]

    Gianluca Baio and Marta Blangiardo. 2010. Bayesian Hierarchical Model for the Prediction of Football Results.Journal of Applied Statistics37, 2 (2010), 253–264. doi:10.1080/02664760802684177

    work page doi:10.1080/02664760802684177 2010

  3. [3]

    Middleton, Timothy J

    Ryan Beal, Stuart E. Middleton, Timothy J. Norman, and Sarvapali D. Ramchurn

  4. [4]

    Bosse, Jon Evans, Robert G

    Nikos I. Bosse, Jon Evans, Robert G. Gambee, Daniel Hnyk, Peter Mühlbacher, Lawrence Phillips, Dan Schwarz, and Jack Wildman. 2025. Deep Research Bench: Evaluating AI Web Research Agents. arXiv:2506.06287 https://arxiv.org/abs/ 2506.06287

    Pith/arXiv arXiv 2025

  5. [5]

    Glenn W. Brier. 1950. Verification of Forecasts Expressed in Terms of Probability. Monthly Weather Review78, 1 (1950), 1–3. doi:10.1175/1520-0493(1950)078<0001: VOFEIT>2.0.CO;2

    doi:10.1175/1520-0493(1950)078 1950

  6. [6]

    ByteDance Seed. 2026. Seed 2.0. https://seed.bytedance.com/en/seed2. Accessed 2026-07-16

    2026

  7. [7]

    DeepSeek-AI et al. 2026. DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence.arXiv preprint arXiv:2606.19348(2026). doi:10.48550/arXiv. 2606.19348

    doi:10.48550/arxiv.2606.19348 2026

  8. [8]

    Seikavandi, Jacob V

    Adrien Deliège, Anthony Cioppa, Silvio Giancola, Meisam J. Seikavandi, Jacob V. Dueholm, Kamal Nasrollahi, Bernard Ghanem, Thomas B. Moeslund, and Marc Van Droogenbroeck. 2021. SoccerNet-v2: A Dataset and Benchmarks for Holis- tic Understanding of Broadcast Soccer Videos. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ...

    arXiv 2021

  9. [9]

    Dixon and Stuart G

    Mark J. Dixon and Stuart G. Coles. 1997. Modelling Association Football Scores and Inefficiencies in the Football Betting Market.Journal of the Royal Statistical Society: Series C (Applied Statistics)46, 2 (1997), 265–280. doi:10.1111/1467-9876. 00065

    doi:10.1111/1467-9876 1997

  10. [10]

    FIFA. 2026. World Cup 2026 Match Schedule, Fixtures, Results, Teams and Stadiums. https://www.fifa.com/en/tournaments/mens/worldcup/ canadamexicousa2026/articles/match-schedule-fixtures-results-teams- stadiums. Accessed 2026-07-16

    2026

  11. [11]

    FIFA Council. 2023. FIFA Council Approves International Match Cal- endars. https://inside.fifa.com/organisation/fifa-council/media-releases/fifa- council-approves-international-match-calendars

    2023

  12. [12]

    David Forrest, John Goddard, and Robert Simmons. 2005. Odds-Setters as Fore- casters: The Case of English Football.International Journal of Forecasting21, 3 (2005), 551–564. doi:10.1016/j.ijforecast.2005.03.003

    work page doi:10.1016/j.ijforecast.2005.03.003 2005

  13. [13]

    Tilmann Gneiting and Adrian E. Raftery. 2007. Strictly Proper Scoring Rules, Prediction, and Estimation.J. Amer. Statist. Assoc.102, 477 (2007), 359–378. doi:10.1198/016214506000001437

    doi:10.1198/016214506000001437 2007

  14. [14]

    Google. 2026. Gemini 3.1 Pro. https://blog.google/innovation-and-ai/models- and-research/gemini-models/gemini-3-1-pro/. Accessed 2026-07-16

    2026

  15. [15]

    Google. 2026. Gemini Deep Research. https://gemini.google/overview/deep- research/. Accessed 2026-07-16

    2026

  16. [16]

    Andreas Groll, Christophe Ley, Gunther Schauberger, and Hans Van Eetvelde

  17. [17]

    Lars Magnus Hvattum and Halvard Arntzen. 2010. Using ELO Ratings for Match Result Prediction in Association Football.International Journal of Forecasting26, 3 (2010), 460–470. doi:10.1016/j.ijforecast.2009.10.002

    doi:10.1016/j.ijforecast.2009.10.002 2010

  18. [18]

    Hyndman and Anne B

    Rob J. Hyndman and Anne B. Koehler. 2006. Another Look at Measures of Forecast Accuracy.International Journal of Forecasting22, 4 (2006), 679–688. doi:10.1016/j.ijforecast.2006.03.001

    doi:10.1016/j.ijforecast.2006.03.001 2006

  19. [19]

    Kalervo Järvelin and Jaana Kekäläinen. 2002. Cumulated Gain-Based Evaluation of IR Techniques.ACM Transactions on Information Systems20, 4 (2002), 422–446. doi:10.1145/582415.582418

    arXiv 2002

  20. [20]

    Woojeong Jin, Rahul Khanna, Suji Kim, Dong-Ho Lee, Fred Morstatter, Aram Galstyan, and Xiang Ren. 2021. ForecastQA: A Question Answering Challenge for Event Forecasting with Temporal Text Data. InProceedings of ACL-IJCNLP. 4636–4650. doi:10.18653/v1/2021.acl-long.357

    work page doi:10.18653/v1/2021.acl-long.357 2021

  21. [21]

    Ezra Karger, Houtan Bastani, Yueh-Han Chen, Zachary Jacobs, Danny Halawi, Fred Zhang, and Philip E. Tetlock. 2025. ForecastBench: A Dynamic Bench- mark of AI Forecasting Capabilities. InInternational Conference on Learning Representations. arXiv:2409.19839 https://arxiv.org/abs/2409.19839

    Pith/arXiv arXiv 2025

  22. [22]

    Smith, Yejin Choi, and Kentaro Inui

    Jungo Kasai, Keisuke Sakaguchi, Yoichi Takahashi, Ronan Le Bras, Akari Asai, Xinyan Yu, Dragomir Radev, Noah A. Smith, Yejin Choi, and Kentaro Inui. 2023. RealTime QA: What’s the Answer Right Now?. InAdvances in Neural Information Processing Systems

    2023

  23. [23]

    Maurice G. Kendall. 1938. A New Measure of Rank Correlation.Biometrika30, 1–2 (1938), 81–93. doi:10.1093/biomet/30.1-2.81

    doi:10.1093/biomet/30.1-2.81 1938

  24. [24]

    Kimi. 2026. Kimi K2.6. https://www.kimi.com/ai-models/kimi-k2-6. Accessed 2026-07-16

    2026

  25. [25]

    Harold W. Kuhn. 1955. The Hungarian Method for the Assignment Problem. Naval Research Logistics Quarterly2 (1955), 83–97. doi:10.1002/nav.3800020109

    doi:10.1002/nav.3800020109 1955

  26. [26]

    Haopeng Li, Andong Deng, Jun Liu, Hossein Rahmani, Yulan Guo, Bernt Schiele, Mohammed Bennamoun, and Qiuhong Ke. 2026. Sports-QA: A Large-Scale Video Question Answering Benchmark for Complex and Professional Sports. International Journal of Computer Vision134, 5 (2026), 196. doi:10.1007/s11263- 026-02734-1

    doi:10.1007/s11263- 2026

  27. [27]

    M. J. Maher. 1982. Modelling Association Football Scores.Statistica Neerlandica 36, 3 (1982), 109–118. doi:10.1111/j.1467-9574.1982.tb00782.x

    arXiv 1982

  28. [28]

    Grégoire Mialon, Clémentine Fourrier, Thomas Wolf, Yann LeCun, and Thomas Scialom. 2024. GAIA: A Benchmark for General AI Assistants. In International Conference on Learning Representations. https://proceedings.iclr. cc/paper_files/paper/2024/hash/25ae35b5b1738d80f1f03a8713e405ec-Abstract- Conference.html

    2024

  29. [29]

    MiniMax. 2026. MiniMax M2.7. https://www.minimax.io/news/minimax-m27-en. Accessed 2026-07-16

    2026

  30. [30]

    OpenAI. 2026. Introducing GPT-5.4. https://openai.com/zh-Hans-CN/index/ introducing-gpt-5-4/. Accessed 2026-07-16

    2026

  31. [31]

    Polymarket. 2026. Market Data Overview and Historical Prices API. https: //docs.polymarket.com/market-data/overview. Accessed 2026-07-16. Historical prices: https://docs.polymarket.com/api-reference/markets/get-prices-history

    2026

  32. [32]

    Qwen Team. 2026. Qwen3.7. https://qwen.ai/blog?id=qwen3.7. Accessed 2026- 07-16

    2026

  33. [33]

    Jiayuan Rao, Haoning Wu, Hao Jiang, Ya Zhang, Yanfeng Wang, and Weidi Xie

  34. [34]

    Jiayuan Rao, Haoning Wu, Chang Liu, Yanfeng Wang, and Weidi Xie. 2024. MatchTime: Towards Automatic Soccer Game Commentary Generation. InPro- ceedings of EMNLP. 1671–1685. doi:10.18653/v1/2024.emnlp-main.99

    doi:10.18653/v1/2024.emnlp-main.99 2024

  35. [35]

    Sportmonks. 2026. Football API 3.0: Standard Pre-match Odds Feed. https://docs.sportmonks.com/v3/endpoints-and-entities/endpoints/standard- odds-feed/pre-match-odds/get-all-odds. Accessed 2026-07-16

    2026

  36. [36]

    Colin White, Samuel Dooley, Manley Roberts, Arka Pal, Benjamin Feuer, Sid- dhartha Jain, Ravid Shwartz-Ziv, Neel Jain, Khalid Saifullah, Sreemanti Dey, Shubh-Agrawal, Sandeep Singh Sandha, Siddartha Naidu, Chinmay Hegde, Yann LeCun, Tom Goldstein, Willie Neiswanger, and Micah Goldblum. 2025. LiveBench: A Challenging, Contamination-Limited LLM Benchmark. I...

    2025

  37. [37]

    Haotian Xia, Zhengbang Yang, Junbo Zou, Rhys Tracy, Yuqing Wang, Chi Lu, Christopher Lai, Yanjun He, Xun Shao, Zhuoqing Xie, Yuan-fang Wang, Weining Shen, and Hanjie Chen. 2025. SPORTU: A Comprehensive Sports Understanding Benchmark for Multimodal Large Language Models. InInternational Conference on Learning Representations. https://proceedings.iclr.cc/pa...

    2025

  38. [38]

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2023. ReAct: Synergizing Reasoning and Acting in Language Mod- els. InInternational Conference on Learning Representations. https://openreview. net/forum?id=WE_vluYUL-X

    2023

  39. [39]

    Ori Yoran, Samuel Joseph Amouyal, Chaitanya Malaviya, Ben Bogin, Ofir Press, and Jonathan Berant. 2024. AssistantBench: Can Web Agents Solve Realistic and Time-Consuming Tasks?. InProceedings of EMNLP. 8938–8968. doi:10.18653/v1/ 2024.emnlp-main.505

    doi:10.18653/v1/ 2024

  40. [40]

    Z.ai. 2026. GLM-5.1: From Vibe Coding to Agentic Engineering. https://z.ai/blog/ glm-5.1. Accessed 2026-07-16

    2026

  41. [41]

    Andy Zou, Tristan Xiao, Ryan Jia, Joe Kwon, Mantas Mazeika, Richard Li, Dawn Song, Jacob Steinhardt, Owain Evans, and Dan Hendrycks. 2022. Forecasting Future World Events with Neural Networks. InAdvances in Neural Information Processing Systems. https://proceedings.neurips.cc/paper_files/paper/2022/hash/ aec870a6772336c15dac992c16f2e7c9-Abstract-Datasets_...

    2022

  42. [2019]

    doi:10.1515/jqas-2018-0060

    A Hybrid Random Forest to Predict Soccer Matches in International Tournaments.Journal of Quantitative Analysis in Sports15, 4 (2019), 271–287. doi:10.1515/jqas-2018-0060

    doi:10.1515/jqas-2018-0060 2019

  43. [2021]

    InProceedings of the AAAI Conference on Artificial Intelligence, Vol

    Combining Machine Learning and Human Experts to Predict Match Out- comes in Football: A Baseline Model. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 35. 15447–15451. doi:10.1609/aaai.v35i17.17815

    work page doi:10.1609/aaai.v35i17.17815

  44. [2025]

    InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Towards Universal Soccer Video Understanding. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 8384–8394. https://cvpr.thecvf.com/virtual/2025/poster/32544

    2025

This paper was first reviewed by deepseek-v4-flash on August 1, 2026.