How Are Player Cards Determined

[Glenn-Philadelphia]

These are the ingredients for determining offensive (GREEN) values

An offensive composite is calculated based on goals and assists (with goals being weighted heavier) divided by total minutes played. This composite score is then used as a reference in a chart I created to come up with the offensive score. This means that a player who has 10 goals and 10 assists while playing a total of 800 minutes in a year will have a worse offensive number compared to a player who had 10 goals and 10 assists in 600 minutes of play.


These are the ingredients for determining defensive (RED) values:

The main ingredients (with varying amounts used if F or D) are:

seconds played between blocked shots
seconds played between takeaways
seconds played between giveaways
percentage of time played shorthanded
Corsi Rating
and lastly
player plus minus compared to teams plus minus

Neutral (BLUE) rating is determined by an algorithm that uses both the OFF and DEF derived scores.

The sub-scores for each of these 3 areas are assigned in a way that the sub score aggregate equals the total score for the category.

Also, if a player for one reason or another played less than a certain number of games, the card used will be the better of a comparison between an average of the players cards (last 3 years I do believe) versus the nerfed version of the card generated. Now that being said, if the average is used, the players Fortitude will be very low and he will be prone to injury.

edit: GB 7/5/2018 (added Corsi rating verbiage and average of scores to OP)

[Phil-Cornwall]

This thread/post only has 33 views ?!?!

Glenn, where is the data to make all this happen mined ? THN website ? NHL.com ?

[Glenn-Philadelphia]

www.nhl.com/ice/playerstats.htm?navid=nav-sts-indiv#www.nhl.com/ice/playerstats.htm?navid=nav-sts-indiv#

That is where the Lions-share of the stats are grabbed. I use about 5 or 6 different views (change the report drop down box) to grab the data. Positions come from TheHockeyNews.com

[Lance-Pittsburgh]

Is this original post still accurate? Has this been expanded upon in another thread somewhere that I’ve missed?

[Scott-New York]

As far as I know, this is accurate

[Glenn-Philadelphia]

I added corsi ratings to the defensive mix 2 or 3 years ago. Other than that, it has been static

[Matt-Syracuse]

What about this explosion of hockey sabermetrics? Any of it useful?

[Glenn-Philadelphia]

I am not familiar with it. The offensive numbers are rock solid as they are directly derived from readily available offensive stats and minutes played. Thay being said I am always looking for metrics that can be used to further tighten how d ratings created. The fact that a good defensive player loses some d score if he doesn't play on the PK unit is a known necessary evil at present

[Scott-New York]

There is, however, realism in having penalty killers scored a little higher than their counterparts who do not kill penalties. It stands to reason that NHL coaches are using their best defensive players when killing penalties.

[Glenn-Philadelphia]

I agree 100% but this actually reduces the defensive value of cards for guys who are good or at least average defensively but don't get the PK ice time because of heavy minutes on the PP and 5 on 5. Just because they dont npkay doesn't mean they are sub par defensively. It's just one aspect of d rating and that's 2hy I would love to find some additipnal stats to use in conjunction with the current mix to tighten those up

[Scott-New York]

I absolutely agree, I just don't think we want to diminish the value of penalty killers at the same time or put all defenseman on a level playing field