Login

***sve7en*** · 08-31-2020, 12:07 PM

[560 words, most from last season, plus sheets]

First delivered last year for the SMJHL, I present on behalf of the SJHN:

[note for graders, basically copy pasted from last season]
I initially began this work as a deep dive to determine both if there was an accurate way to measure individual contribution to a team in the manner that Wins Above Replacement (WAR) does for baseball. While @juke had done offensive and defensive point shares throughout the season, I wanted to refine one thing, open up transparency and discussion regarding the calculations, and tabulate the data for consumption by both the Awards Committee, and the SHL as a whole. As I did this, I discovered alternative candidates for MVP depending on what definition of MVP is used, and decided to follow up on the outliers that I found.

My process (and the fields you see in the spreadsheet are found and detailed more in this post on hockey-reference.com.

The only change from the process listed above is that FHM doesn't track team assists, the ways to get this value are:

Sum up the assists from the players from the index, which doesn't properly account for roster moves.
Sum up the goals and assists across the league, and derive a common assist per goals stat, which doesn't perfectly reflect differences in teams.
Manually calculate the values, which I don't have time to do.

While in a crunch for time during the season, the SHN has used the first method, I chose to base the team assists metric off the league average as I believe it most closely represents the actual values.

Beyond this, point shares analysis is inherently flawed in some ways, and there is no perfect metric for determining impact in a fluid sport such as hockey, but it does paint a solid picture. The core issues are the lack of representation of off-puck impact and rigidity in defining and allocating success between position groups. For much of the defensive off-puck impact plus-minus is used, which has it's own issues - namely that it imperfectly measures offensive and defensive off-puck impact. Offensive off-puck impact isn't recognized at all, and point shares only consider the goals and assists. Beyond that, defensemen and forwards have a rigid allocation of assumed contribution in point share analysis, which falls apart when considering the fluidity and roles of hockey players.

Because of the issues in point shares, they obviously should not be a definitive metric of ranking players, and I doubt there will ever be one. The best approach to measuring players will always be a holistic view of their play such as those I've done recently, considering all of their basic and advanced stats as well as the context for those numbers, including their situations that they preform in and the players around them.

DeletedAtUserRequest · 09-03-2020, 12:55 PM

Nice, Always a great read.

|

|

|

|

Navigation

Extra Menu

About us