Skip navigation

 Login or Register | Member Centre

Here we go

ALAN RYDER

Globe and Mail Update

Latest comment posted at 6:37 PM EST 02/10/08

A look at who made the best contribution to their teams in 2007-08 ...Read the full article

This conversation is closed

  1. Mike M from Canada writes: I like the list and agree with your way of assessing a player's on-ice impact (it needs to be remembered though, that there are off-ice contributions that can matter).

    However, your methodology seems limited by the statistics that we currently collect. The impact of a goalie is mixed up in the defensive metrics of his teammates, since you're evaluating players on the number of goals scored against them. Essentially, players can influence the quality of the other team's opportunities, while goalies can influence whether the other team scores with those opportunities. If you measure players by the number of goals scored against, they're being evaluated on both metrics as opposed to just the quality of opportunity. Your goalie metric separates these two aspects, but the player metrics do not.

    It's too bad that the NHL doesn't keep track of who is on the ice for shots against, the same way it tracks who is on the ice for goals against. This would allow you to separate the quality of opportunity from the goal being scored. You could then evaluate players based on the quality of offensive opportunities the opposition has while they are on the ice.
  2. Alan Ryder from Canada writes: My PC math actually does separate goaltending and defence rather than leaving them intertwined. For instance both Detroit and San Jose were very good at goal prevention and I attributed most of that to defence rather than goaltending. How? I attribute to defence the responsibility for shots and shot quality. I attribute to goaltending a shot quality neutralized save percentage.

    The NHL's play by play files now give us the players on ice for every 'event' (including shots). The problem is that the NHL does not summarize this for us. This makes the data only available to those who have the time and energy to collect it from the game files. See, for instance, theglobeandmail.com
  3. Mike M from Canada writes: Very cool - thanks, I didn't know that and stand corrected.
  4. P McLean from Canada writes: alan, you may have explained this in a previous column. how do you determine shot quality? I'm really curious as to how, without watching every game and assigning a value to each shot. Not calling you out, just curious as to how you do that.
  5. Alan Ryder from Canada writes: There are a number of (similar) ways to figure shot quality. At my web site (www.hockeyanalytics.com) you can find a number of papers on the subject.

    In simple terms, one can do a regression analysis on the 70,000 shots taken every year to build a model that predicts the probability of a goal under the observed circumstances. By far the most important available predictors are shot distance, shot type and situation (EH, PP, SH). With this probability model one can determine 'expected goals' for a team (or goaltender). When you compare this to league average expected goals, normalized for the the number of shots taken, you can get a shot quality factor. This factor is typically between 90% and 110%. A factor of 95% means that the shots allowed were 5% less dangerous (expected goals 5% less) than league average.

Comments are closed

Thanks for your interest in commenting on this article, however we are no longer accepting submissions. If you would like, you may send a letter to the editor.

Report an abusive comment to our editorial staff

close

Alert us about this comment

Please let us know if this reader’s comment breaks the editor's rules and is obscene, abusive, threatening, unlawful, harassing, defamatory, profane or racially offensive by selecting the appropriate option to describe the problem.

Do not use this to complain about comments that don’t break the rules, for example those comments that you disagree with or contain spelling errors or multiple postings.

Back to top