Hockey Outsider
Registered User
- Jan 16, 2005
- 9,166
- 14,499
Most of you are aware of BM67’s adjusted scoring system (link). Basically, the player who finishes second in a scoring category receives one point, and all other statistics are compared to that amount.
The main benefit of this system is that it put all players from all seasons on an even playing field. Since all players are compared to their peers in a given year, any major variables (ie length of the NHL schedule, level of offense, average ice time per player) are automatically taken into account by the statistic. Still, I think there are two weaknesses with his approach, and I’ve tried to improve on them.
The first weakness is that the results are a bit confusing. For example, when I’m told that Stan Mikita scored 14.913 career points and averaged 3.26 points per season over his prime, it’s hard to intuitively understand the number. The second weakness is that not all second place finishes are created equal – sometimes there’s really only one player who’s ahead of the pack (ie Gretzky scoring 183 points when there were several players 100-110 points) and other times there are a few players well ahead of the pack (ie Lemieux & Jagr each scoring 149+ points, when nobody else exceeded 120 points). The numbers may be skewed if you’re comparing a season with a strong player who finishes 2nd in scoring, to a season with a weak 2nd place finish.
Essentially my “revised BM67 scoring” system is calculated as follows:
Why this method makes sense:
The disadvantages:
Let me know what you think about this method. Here are some statistical tables:
Example: Iginla 52 goals= 1.268293 points, Guerin, Sundin & Murray 41 goals= 1 point each, Naslund 40 goals= 0.97561 points
The main benefit of this system is that it put all players from all seasons on an even playing field. Since all players are compared to their peers in a given year, any major variables (ie length of the NHL schedule, level of offense, average ice time per player) are automatically taken into account by the statistic. Still, I think there are two weaknesses with his approach, and I’ve tried to improve on them.
The first weakness is that the results are a bit confusing. For example, when I’m told that Stan Mikita scored 14.913 career points and averaged 3.26 points per season over his prime, it’s hard to intuitively understand the number. The second weakness is that not all second place finishes are created equal – sometimes there’s really only one player who’s ahead of the pack (ie Gretzky scoring 183 points when there were several players 100-110 points) and other times there are a few players well ahead of the pack (ie Lemieux & Jagr each scoring 149+ points, when nobody else exceeded 120 points). The numbers may be skewed if you’re comparing a season with a strong player who finishes 2nd in scoring, to a season with a weak 2nd place finish.
Essentially my “revised BM67 scoring” system is calculated as follows:
- I take the average of second through fourth place in goals. My rationale is that if you average a few top spots, the result is unlikely to be skewed by a single really strong or weak performance from whoever finishes in 2nd. The average is then awarded 50 goals. Each player’s total goals in a season is compared to this number.
- Example: in 2002, the average of 2nd through 4th place was 41 goals. We pro-rate all goals by 50/41 = 1.219. So Iginla’s 52 goals are worth 63, Sundin, Guerin & Murray’s 41 goals are worth 50, Naslund’s 40 are worth 49, etc.
- I do the same calculation for assists (fixing the average of 2nd to 4th place as 75 assists). Then I add goals and assists together to get total adjusted points.
Why this method makes sense:
- Players are compared to their peers, so it doesn’t matter whether they played in a low- or high-scoring era, and other variables (ie length of schedule, average ice time, etc) are rendered irrelevant
- The results make sense intuitively since they are all benchmarked to 50 goal & 75 assist milestones.
- Since I use the average of 2nd through 4th place, it’s unlikely that a really good/bad performance by a single player will skew the results in any way
The disadvantages:
- There is no adjustment for quality of era, so even obviously weaker eras (ie pre-consolidation from 1918-1925 and WWII from 1943-1945) are treated equally to eras with strong talent pools
- Obviously this method looks at regular season scoring statistics only -- so playoff performances, defensive play, etc, are not included in this analysis
Let me know what you think about this method. Here are some statistical tables:
Last edited: