Comparison of automated and manual performance analysis tools in amateur padel: a descriptive case study

Background Automated match-analysis systems increasingly provide technical-tactical reports in padel, but their agreement with manual observation remains unclear. The aim was to describe discrepancies between aggregate match statistics displayed by Padmi and manual observational coding in amateur padel. Methods Three amateur male matches involving one focal player were retrospectively analysed from video using the Padmi app (version 1.3). Six shot types (forehand, backhand, forehand volley, backhand volley, bandeja, smash) and three outcome categories were coded manually and compared with Padmi's aggregate reports. Signed and absolute differences were calculated for shot frequencies and percentage-point differences for outcomes. Shot-level agreement metrics were not calculated because raw Padmi event-level data were unavailable. Results Manual coding identified 891 shots, whereas Padmi displayed 757 (−15.0%). Padmi displayed fewer shots in all matches (−11.4% to −19.2%). Discrepancies varied by shot type: forehand −28.3%, backhand −20.0%, bandeja −3.5%, smash −44.8%, forehand volley +27.2%, backhand volley +5.9%. Overall outcome distributions were similar, but larger shot-specific discrepancies emerged for the smash, bandeja, and forehand volley. Conclusions Padmi's user-facing aggregate statistics differed from manual coding in this small case study. Automated reports may be useful for broad feedback but require cautious interpretation and further independent validation for detailed technical-tactical decisions.