How they work and what we should take away from them – The Athletic

When it comes to gauging offense, we know that shots aren’t created evenly – that’s a criticism of focusing on just sharing the shot, though there is predictive potential for Corsi. Scoring opportunities are, in theory, a solution to improving the measure of quality as well as quantity. But there are loopholes there as well, from consistency in how scoring opportunities are considered, to the filming location being just an extra piece to the puzzle.

When assessing “risk” for a shot, there are quite a few factors to consider in addition to the location of the shot. This is what expected goal models try to do.

But in the public domain, there is more than one model floating around – and more than one way to calculate the scale. The four “basic” models are the HockeyViz, MoneyPuck, Evolution-Hockey, and NaturalStatTrick models.

So, let’s take a closer look at some of the primary goal models expected in the public domain, and delve into what these factors are and why they feature certain metrics.

How did they come to be

Peter Tanner drew inspiration from football in his model building at MoneyPuck.com. But what caught his eye was the opportunity. No other website has been updating this metric directly during games.

For Josh Younggren of Evolution-Hockey.com, a website he created and maintains with his twin brother Luke, this was the next step in hockey modeling.

Similarly, Brad Timmins of NaturalStatTrick.com created his site because he felt you “should” have this advantage because “the analysis was turning into [expected goals] of scoring chances.

.

Leave a Comment