Hockey Outsider
Registered User
- Jan 16, 2005
- 9,314
- 15,152
I wanted to revisit an idea that @matnor presented about a decade ago. (He was a great poster and, unfortunately, no longer appears to be active). matnor was trying to find a way to measure the number of games that each player played at a scoring pace that was high to lead the league (or give them a 5th place finish - or any other rank).
It's always been tough to deal with players who scored at a high rate in partial seasons. For example, it feels unfair to give Peter Forsberg credit for leading the league in scoring in 2004. He was by far the league's leader in points per game, but he only played in 39 games. It doesn't feel right to give him credit for an Art Ross trophy (taking it away from Martin St. Louis, who wasn't quite as good, but who played in all 82 games). On the other hand, it also feels misleading to treat Forsberg as finishing tied for 47th in scoring, as if he were no different than Shawn McEachern, Nils Ekman, or Tyler Arnason.
matnor's method (link) says, essentially - let's recognize that Forsberg scored at the highest rate in 2004. But he'll only get credit for doing so for 39 games. (The purpose of the method isn't to crown a new scoring champion each season - it's to look at how many games a player played at a certain pace during the entire career).
There are two important modifications that I've made to the method. The first one is simple - all seasons are adjusted in length to 82 games. Jean Beliveau gets credit for leading the league in PPG for 82 games in 1956 (because it's not his fault the schedule was only 70 games back then). Then in 1959, he gets credit for 64/70*82 = approximately 75 games at this pace. (I treated the 2020 season as being exactly 70 games in length, even though the actual length varied from 68 to 71 games - thanks to COVID).
The second modification is I didn't have any minimum thresholds for games played. If a player played at least half the games, PPG is calculated normally. If a player played in less than half the games, I take his actual point total, and add enough scoreless games to get him to the 50% threshold. (Otherwise, in 2004 as an example, Peter Forsberg wouldn't get credit for anything whatsoever with his 39 games, but Marc Savard's 45 games would qualify - it didn't feel right to disqualify Forsberg entirely when there's only a difference of six games). So Forsberg's adjusted PPG is 55 (actual points) / 41 (39 actual games + 2 scoreless games) = 1.34 PPG. In this case, it's still enough to lead the league (and Forsberg only gets credited for the 39 actual games he played). Sometimes, the order changes. For example, in 1952, Dickie Moore played in 33 of the 70 games, and scored more PPG than Ted Lindsay. However, after adding two scoreless games to get Moore up to the 35 game mark (half the schedule), he fell behind Terrible Ted. So Moore was 2nd in PPG in real life, but he gets knocked down to 3rd in this method (but at least he gets some credit, rather than having the season disqualified outright).
All data has been compiled in good faith. There's a lot of numbers here (more than 1 million cells of data), so let me know if anything jumps out as being inaccurate.
It's always been tough to deal with players who scored at a high rate in partial seasons. For example, it feels unfair to give Peter Forsberg credit for leading the league in scoring in 2004. He was by far the league's leader in points per game, but he only played in 39 games. It doesn't feel right to give him credit for an Art Ross trophy (taking it away from Martin St. Louis, who wasn't quite as good, but who played in all 82 games). On the other hand, it also feels misleading to treat Forsberg as finishing tied for 47th in scoring, as if he were no different than Shawn McEachern, Nils Ekman, or Tyler Arnason.
matnor's method (link) says, essentially - let's recognize that Forsberg scored at the highest rate in 2004. But he'll only get credit for doing so for 39 games. (The purpose of the method isn't to crown a new scoring champion each season - it's to look at how many games a player played at a certain pace during the entire career).
There are two important modifications that I've made to the method. The first one is simple - all seasons are adjusted in length to 82 games. Jean Beliveau gets credit for leading the league in PPG for 82 games in 1956 (because it's not his fault the schedule was only 70 games back then). Then in 1959, he gets credit for 64/70*82 = approximately 75 games at this pace. (I treated the 2020 season as being exactly 70 games in length, even though the actual length varied from 68 to 71 games - thanks to COVID).
The second modification is I didn't have any minimum thresholds for games played. If a player played at least half the games, PPG is calculated normally. If a player played in less than half the games, I take his actual point total, and add enough scoreless games to get him to the 50% threshold. (Otherwise, in 2004 as an example, Peter Forsberg wouldn't get credit for anything whatsoever with his 39 games, but Marc Savard's 45 games would qualify - it didn't feel right to disqualify Forsberg entirely when there's only a difference of six games). So Forsberg's adjusted PPG is 55 (actual points) / 41 (39 actual games + 2 scoreless games) = 1.34 PPG. In this case, it's still enough to lead the league (and Forsberg only gets credited for the 39 actual games he played). Sometimes, the order changes. For example, in 1952, Dickie Moore played in 33 of the 70 games, and scored more PPG than Ted Lindsay. However, after adding two scoreless games to get Moore up to the 35 game mark (half the schedule), he fell behind Terrible Ted. So Moore was 2nd in PPG in real life, but he gets knocked down to 3rd in this method (but at least he gets some credit, rather than having the season disqualified outright).
All data has been compiled in good faith. There's a lot of numbers here (more than 1 million cells of data), so let me know if anything jumps out as being inaccurate.
