Interpretable Machine Learning for Football Performance Analysis: Evidence of Limited Transferability from Elite Leagues to University Competition
Yu-Fang Tsai a *, Yu-Jen Chen b *, Kok-Hua Tan c , Sheng-Chieh Huang d , You-Ying Ji e , Yu-Lun Chen f , Chun-Yi Wang g and Chien-Ming Hsu a · 2026
Read Interpretable Machine Learning for Football Performance Analysis: Evidence of Limited Transferability from Elite Leagues to University Competition by Yu-Fang Tsai a *, Yu-Jen Chen b *, Kok-Hua Tan c , Sheng-Chieh Huang d , You-Ying Ji e , Yu-Lun Chen f , Chun-Yi Wang g and Chien-Ming Hsu a (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-10796.txt.
Description
Yu-Fang Tsai and colleagues ask whether game traits learned from top football carry to college games, using clear machine learning. Models trained on large-scale event data from Europe's top five leagues were run on college football data from National Tsing Hua University with the same trait set. Random Forest and Multilayer Perceptron models were read through SHAP and Counterfactual Impact Score methods.
Across five tests, top football showed a firm steady rank order of game traits across leagues, models, and read methods. College football, by contrast, showed much reordering of key signs, less steady reads, weaker frame match with top leagues, and more swing with read method. The results hint that read strength shifts with the field. Wobble in reads under a field shift may flag frame haze in the new field, not just method flaws on their own.
Catalog record
| CREATOR | Yu-Fang Tsai a *, Yu-Jen Chen b *, Kok-Hua Tan c , Sheng-Chieh Huang d , You-Ying Ji e , Yu-Lun Chen f , Chun-Yi Wang g and Chien-Ming Hsu a |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42252 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.10796