In this series of articles, we describe a Bayesian approach to analyzing goalies that has these benefits over using Even Strength SV% (ESSV%):
  1. It is easier to compare goalies who have faced a different number of shots.  
  2. We get error bounds for our estimates.
  3. Our approach allows us to answer a variety of questions like "What is the probability that Goalie A is better than Goalie B?" and "What is the probability that Goalie A's true ESSV% is above .920?"
For now, it is best for comparing goalies on the same team, since we don't account for the quality of shots that the defense allows.  We'll account for shot quality in another series of articles. 

Links to a series of articles on A Bayesian approach to analyzing goalies:

Part 1 - Introduction
Part 2 - An example using only 10 shots
Part 3 - Updating estimates with more information
Part 4 - Regression to the mean, Luongo vs Schneider, Thomas vs Rask, Hiller vs Fasth
Part 5 - coming soon

A nice summary of the series and the subject by Eric T here.

Finally, an animation discussed in the series:
This series is based on joint work with Nick Clark.  

1 comment:

  1. Very well written post. Thanks for sharing this, I really appreciate you taking the time to share with everyone. Best PMP Certification Online