Hockey Player Rating Method

I have been writing a series of posts on sabermetrics and hockey. I wrote several posts about Pnep's Hall of Fame Monitor. This is by no means the only way to rank players from different eras. Another attempt recently posted on the net is Daryl Shilling's new hockey project rating system.

This system takes into account six things. Tenure, career value, rate stat performance, dominance, awards and undocumented seasons.

Tenure is calculated by measuring the career adjusted games played (these are the percentage of games played in each season multiplied by 82 games (a standard number of games played in an NHL season). This is multiplied by 400 and then divided by the all time leader in adjusted games played (Gordie Howe). Thus, Gordie will get a score of 400 and all other players will have a lower score.

Career value is normalized goals, assists, points and goals created. Normalized scoring stats are explained here. Goals created is another sabermetric stat Daryl Shilling uses that I will have to write about another day or you can read about it here. Each of these stats is normalized by multiplying it by 100 and dividing by the all time leader. A player who leads in a 4 of these stats will again have a total score of 400 points.

Rate stat performance is measured using adjusted goals, adjusted assists, adjusted points and goals created all per game. Again it is normalized to give a maximum possible score of 400.

Dominance is measured using stats called ink points per year, career ink points, league leader points and 30Ink points. At this point, Daryl has not fully posted exactly how he determines these stats. I will quote his explanations, although they are not complete enough that I could calculate them myself. Basically they attempt to measure how often the player was among the league leaders offensively and how high in the league leaders he was

Ink Points are derived by a players finish in a category in a given year. For example, if a players lead the league in NG in 2004, then he receives 10 Ink Points. The second leading goal scorer will receive 9, third will get 8, and so, until the 10th leading NG scorer receives 1 Ink Point. In early year of the NHL, fewer players wll receive Ink Points due to their being fewer players in the league, and as such, more chance to receive Ink Points.

30Ink refers to any season in which a player recorded 30 Ink Points, meaning that he lead the league in Normalized Goals, Assists and Points. That would represent an extremely dominant offensive season.

Again this is normalized to be out of 400 and the all time leader in all four stats would get 400 points.

Awards include Hart Trophies, Conn Smythe Trophies, Norris Trophies or Selke Trophies, and All-Star nominations. Before these trophies existed, Daryl uses projected winners picked by various experts for years when these trophies did not exist. This leaves 4 possible categories for each player (since defenders can only win Norris but not Selke trophies and forwards vice versa). Each of these four categories is normalized to 100 points for a maximum of 400 points in this category if a player leads all time in number of victories in all 4 award categories.

Finally, we come to one of the more unique things Daryl attempts. He attempts to account for undocumented seasons. These are: players whose careers had begun before the formation of major league hockey in 1910, players that missed time due to military service or players who were unable to play major league hockey due to the political atmosphere of the world. He calculates the number of points a players expected points under this system per season then multiplies by the percentage of seasons in their career the player played in these undocumented years.

I'll have to explain goals created in a future post. Then we can have some fun and look at the results of the HPR system and compare it with Pnep's system. When we see which players do better in the respective systems, we can gage the strengths and weaknesses of them.