Player scouting is a critical component of hockey,
enabling teams to identify standout players, devise
strategies, and assess performance. However, traditional
scouting methods are resource-intensive, requiring
substantial time and travel expenses for organizations.
An approach to performance assessment
has been video footage analysis; however, in
the context of ice hockey, challenges arise primarily
due to the fast-paced nature of the sport and
the limited field of view in broadcast footage.This
makes it difficult to consistently track all players
on the ice simultaneously. A solid performance
metric that would make the most out of the information
at hand is thus required. We present a
1v1 Success Metric Pipeline, an autonomous system
designed to analyze video footage and calculate
a success metric to evaluate defensive plays.
By automating the tracking and evaluation process,
this tool not only streamlines scouting but also provides
coaches with valuable insights for player performance,
which may come into play during line
matching and other strategic decision-making during
games. Results show that the ISM metric properly
reflects the danger level of a one on one encounter
and that despite the sample size being small,
the model achieves low root mean square error.
PDF