Validate football models by the misses, not the average
01 July 2026
Pass detection reached 0.92 F-score on dependent data and 0.71 on independent data. Staff should see per-class misses, timing errors, and context splits before using any model score in a report.
Bischofberger, Baca and Schikuta tested pass and shot detection on four football datasets. 1 The independent Euro sample gave the stress test because its positional feed came from Tracab and its event feed came from Wyscout after manual timestamp correction from broadcast footage.
Get ahead of the game
Sign in to continue reading.
Sign in to read during the free access period. No credit card required. By continuing, you agree to our Privacy Policy and Terms of Service.
Free access period
Read every Research report free during the free access period. Plans will be $144/year or $15/month. Annual saves 20%. Opinion posts and the Weekly Recap remain free.