by PuckStopsHere on 09/10/10 at 02:51 AM ET
My sabermetrics and hockey post today is a look at the 20 best players (goalies will be ranked separately in an upcoming post) in the NHL last season by shootout goals versus threshold. Since shootouts are reasonably clear statistically, i.e. the player with the most goals on the least shots will be the top player, this portion of Tom Awad’s goals versus threshold system it quite reliable. The only question is if the goals valuations are correct.
Here are the top 20 players by shootout goals versus threshold:
These statistics were calculated by behind the net.
I have little doubt that this is a very reliable list based on last season’s stats. My problem comes in the valuation of shootouts. First it is for reasons of the way hockey should be. It is alarming that 10 staged breakaways by Sidney Crosby should have so much value. Sidney Crosby played 1778 minutes of hockey in the 2009/10 regular season. He is credited with 29.4 goals versus threshold on the entire season. In over 29 hours of regular season ice time, almost 13% of Crosby’s value comes from ten staged breakaways. That is far too much emphasis given to the shootout. This is a fundamental criticism of shootouts in general. They play too large a value in the NHL results.
For some players, their only value is shootout. Mike Santorelli was a net negative player in this system neglecting shootouts. His lack of offence (3 points in 25 games) are worth -0.9 goals versus threshold. His defence is worth -0.2 more goals. Santorelli was worse than a replacement player last year, except for three staged breakaways in which he scored. They are enough to by themselves make him a positive player worth 0.5 goals versus threshold.
On a statistical level there is room to argue that the valuation of shootouts might be wrong, because the valuations of offence and defence (especially defence) are wrong in this system. For example, if defence is more important than this system can capture, then the percentage of goals versus threshold on shootouts captured in this system will be less.
I dislike the fact that Sidney Crosby is more valuable in this system on his ten shootout shots than he is in all of his defensive play in the entire season (he gets 3.8 goals versus threshold on the shootout and 3.6 for defensive play). Although there are plenty of reasons why defensive goals versus threshold is suspect, the conclusion that ten breakaway shots are more valuable than Crosby’s defensive play is quite plausible.
Shootouts are far too valuable in the NHL today. That is why it is useful to measure shootout performance when evaluating players. One question that needs to be addressed is the repeatability. Crosby went 8/10 in 2009/10, but went 3/10 in 2008/09 in the shootout. There are not enough shootouts to establish reliable statistics. The player with the best numbers in any given season is lucky. His few chances that year are above his more established success rate that years of shootout data would suggest. Basically, the idea is that in ten flips of a fair coin, it is not that unlikely that you will see 8 heads of 8 tails a reasonable amount of time, even though 5/10 is the most probable number. The player who has luck on his side will put up the best number in any given season, when his longterm average level is well below it. In short, shootout success can be measured, but it has limited predictive value.
I wish the NHL did not have shootouts. Having games decided by a skills competition is not an ideal situation. It is basically an exciting coin toss to decide games. Nevertheless, if they exist, measuring effectiveness in them is useful and Tom Awad has done a good job of this.
Be the first to comment.
Add a Comment
Please limit embedded image or media size to 575 pixels wide.
Most Recent Blog Posts
About The Puck Stops Here
Who am I? A diehard hockey fan.
Why am I blogging? I want to.
Why are you reading it? ???