Friday, 18 December 2009

Player Effectiveness Measures

I think I've argued here before that hockey needs to undergo a sabermetrification of sorts, an analysis of the metrics that we use to determine how effective a player is being in particular situations, how well a team is doing as compared to earlier teams, how effective the power play unit is, etc. I've recently discovered On Goal Analysis, a site with an associated blog, that apparently shares my interest. Recently they've undertaken to design a metric that measure overall effectiveness. This metric works by measuring the number of points scored per shift and the number of "defensive actions" per shift.

I think the intent is laudable, but it's not clear to me that this is a better metric than the old fashioned +/- metric that simply calculates the difference between the number of even strength goals scored while you were on the ice and the number scored against your team. After all, not all defensive actions are equal. Some hits are light or result in the hitter falling out of the play, some blocked shots weren't headed for the net or fervent shot blockers may also have a tendency to screen their goaltenders. Similarly, not all shifts are equal. Some players spend more time on power plays, others on penalty kills. Some teams favour longer shifts, other short shifts. And considering only points scored per shift overlooks the offensive actions that can lead to goals but not count as points or "defensive actions". At the very least, I'd have liked to have seen this changed to points/defensive action/minute played, but even then we're ignoring power plays. The old +/- effectively addresses all these things without overly favouring power play units and hurting penalty kill units. Presumably, if your defensive actions are effective, the long term result is fewer goals scored against you. Show me a man with a high 'defensive action/shift' and a lousy +/- and I'll show you a poor hitter or shot blocker.

So, I like the old +/-, the main disadvantage being that it's not very effective at measuring player power play and penalty kill contribution. But here's a way to do that: for each player measure the power play goals/per power play minute played (or net goals, subtracting shorthanded goals) and number of goals scored against/per penalty kill minute. Then compare that to the team average. Is the player a positive or negative contributor to these situations?

No comments: