This isn't really original, or even half-clever. It's a cruder, stupider variant of a clever idea called shot quality.
But here goes.
While I was trying to avoid watching the OT tonight, I did this:
If you don't need many shots to score a goal, we'll call that "shot efficiency." Other analysts refer to it (or related measures) as "shot quality."
How many shots does a team have to take to score a goal? Here's the
First number is team rank (by point percentage, so it accounts for games in hand, OTL, etc.; when all teams have played all their games, point percentage ranking is the same as their final standings ranking).
The thing about "shot efficiency" is that it doesn't seem to correlate tightly with team goodness or badness. Partly this is because it doesn't measure defense at all, except in the sense that it is composed of a good offense :).
So this is a lopsided measure to start, and made more lopsided by the thought that the other way to score goals is to just shoot a lot, even if any one chance isn't a "quality" shot, and you don't score a lot of goals.
I also think, without testing this thought, that this is a fairly volatile measure.
Enough couching. Analysis.
I predicted in The Room that the Canucks were an efficient team (Puck control team, many passes, few shots, good shots). They are the 6th most efficient-shooting team. Coincidentally, they're the 6th-lowest at shots on goal per game.
NYR are notable as the least efficient team in hockey, but also the 5th best team in hockey. They shoot a lot. Fifth most in the league. Again, that's a coincidence.
The variable outcomes of shot efficiency are highlighted by the 9th and 10th most efficient teams. 9 is 29-ranked STL, 10 is 1-ranked DET.
Detroit shoots the most in the league and scores the most in the league and SCREW shot efficiency! STL has far fewer shots and far fewer goals and nearly identical shot efficiency to DET.
I extracted all the numbers from NHL.com's pretty good stats pages.
Comments
metrics in general
This is a problem with statistics and defining metrics in general. Before defining a metric you want to qualitatively figure out what you are trying to measure. After defining the metric you need to test the predictability of the metric. Qualitatively Detroit has a very good offense and the metric does not reflect that. So this is a useless metric. Period. End of story. Good read something else.
Hee hee!
I appreciate the comment and the sentiment, Mr. Smartypants, but I knew exactly what I was trying to measure when I started: The Canucks appear to play a very possession-oriented game in which they shoot rarely but effectively. I searched for a metric that would extract an answer to that question.
It appears to have been answered (they rank highly on that metric). I knew (heck, I even said explicitly) that this was not a measure of a team's performance or even their offensive capability. And this exploration supported the null hypothesis of what I insist was an interesting question to ask: does shot quality correlate with team performance?
I consider it quite interesting that, at a first cut, teams can score a lot of goals either by shooting a lot but ineffectively, or by shooting rarely but effectively. Detroit shoots a lot, fairly effectively. I have further questions to work out...
Post new comment