Mike Chen's Hockey Blog
Next entry: All-Star Game Live Blog & Other Things
Previous entry: Off Topic: U2’s Get On Your Boots
Calculating The Worst Player
by Mike Chen on 01/21/09 at 12:37 PM ET
Comments (1)
On XM Home Ice this morning, there was a discussion about the league’s worst player. People called and emailed in, and one of the common themes was being the guy who turned over the puck most. There were also discussions of wasted talent (Maxim Afinogenov was tossed around).
What makes up the worst player in the league? Is it something you could tackle statistically? I got to thinking about this and I came up with a little formula. The key statistics here are ice time, giveaways, and +/-, but there’s a little bit of math involved. First, some explanations.
It’s important to note that statistics are relative. Wayne Gretzky was always near the top of the giveaway column, but that’s cause he had the puck so damn much and he tried to set up so many plays. There’s a difference between a soft defenseman turning over the puck and The Great One behind the opponent’s net sending out a pass that was intercepted. With +/-, it’s a reasonable tracker of effectiveness but it’s kind of tainted by overall team abilities. A guy on a kick-ass team will simply have a better +/- through roster osmosis.
So what can we do to break this down? I think you can look put some statistics in a different context to get a better sense of their weight. I only took one stats class in college so I don’t know all the technical terms for this, but it seems reasonably logical to me. Here are some breakdowns:
Weighted Giveaways
[ (average team ice time) / (player’s ice time) ] * [ (average team points) / (player’s points) ] * giveaways
What this does is it takes into account how much a coach uses a guy, then figures out how often he turns the puck over based on that. As mentioned above, strong offensive players might turn the puck over more when they are trying to set up plays. Those guys often have more ice time, so if you weight the amount of giveaways based the effectiveness of the player, their giveaway factor becomes greater.
You might ask why I’m dividing an average value by a player value rather than the other way around. It’s to put a brighter spotlight on giveaways for a bad player. In other words, the less effective a player is, the greater the number will be. For example:
Team average has 14 minutes of ice time, 40 points, and 70 giveaways
Good Player X has 17 minutes of ice time, 80 points, and 100 giveaways
Bad Player X has 11 minutes of ice time, 6 points, and 70 giveaways
Good player giveaway score: [ (14 / 17) ] * [ (40 / 80) ] * 100
Bad player giveaway score: [ (14 / 11) ] * [ ( 40 / 6 ) ] * 70
You can see that the ratios favor the good player in that they diminish the overall value of the final score while the bad player’s get multiplied several times over.
Weighted +/-
(player +/-) - (averaged team +/-)
Essentially here, you want to take away any additional bump that the performance of a team might give to a player’s +/-. For example, if a guy is +2 but the average on the team is +10, then you know his true +/- value is worse. Thus, removing the average of +10 gives you a value of -8. If a guy is -25 but the team average is -6, then he might not be as awful as hit numbers indicate and the weighted value is -19 (which is still pretty bad).
Combined special teams time
(PK ice time) + (PP ice time)
The best players are used in all situations. If you’re hardly used in either, then you’re just a filler player for the coach. Pretty self-explanatory.
Average number of 3rd-period shifts per game
(Total number of 3rd-period shifts / Games played)
When teams are winning, they often continue to roll lines. When teams are losing, teams shorten benches, particularly in the latter stages of the third period. Also, some coaches love to do what I call Bad Turnover Detention. That is, sitting a guy for long chunks of the game after a bonehead play (hello Ron Wilson). Inversely, reliable key players could see their ice time increase in key situations when trying to come from behind.
I’m sure there’s some formula you could put together with these to make one mega Bad Player equation, even if some of these are positive values (special teams time) and some highlight negatives (weighted giveaways). And for those professional number crunchers out there (hello Forechecker!), feel free to poke holes in my ideas.
If anyone wants to put together a massive spreadsheet with this stuff, I’d love to see it. Otherwise, it remains theoretical until I sit down with a calculator and Excel (which won’t be until after I do my taxes).
Filed in: NHL | Mike Chen's Hockey Blog | Permalink
Tags: Statistics, Worst+Player,
Comments
Add a Comment
Please limit embedded image or media size to 575 pixels wide.
Add your own avatar by joining Kukla's Korner, or logging in and uploading one in your member control panel.
Captchas bug you? Join KK or log in and you won't have to bother.
Commenting is not available in this weblog entry.
Most Recent Blog Posts
Intent To Blow vs. In The Crease
What Do You Expect From A Rookie Season?
About Kukla’s Korner
Kukla’s Korner is updated around the clock with the work of our own talented bloggers, plus links to the best hockey writing around the internet. We strive to bring you all the breaking hockey news as it happens.
The home page allows you to see the latest postings from every blog on the site. Subscribe here. For general inquiries and more, please contact us anytime.
Get the top online sports betting bonuses available to sports betters!
When learning from experts it’s best to learn personally from them, or from their blog. We can provide that with poker lessons blog, your home to learn poker personally.
Do you get shocked from the luck in the game of poker? Stop getting shocked and start being a Poker Shoker.
As well as reading about hockey games, you can also find info about poker like which poker sites accept American Express or which are the best Canadian poker sites and also find the top rakeback sites at rakeback.net.
Kukla’s Korner is always a free service for readers, but it costs some money to maintain. If you’re ever in a position to donate a few dollars to help out, we’d be very appreciative.
I would argue that this (like most other NHL issues) cannot be oslved with the application of any statistical formula. Statistics are valuable, but they all come in a complex context that we cannot remove from the numbers.
As you probably know, I thy to pick a worst player so far this seaons and I made my most recent pick early this morning. I picked Brendan Witt of the New York Islanders. Essentially what I am looking for is a player who does not score and has a +/- that is significantly worse than his teammates. There is no statistical formula for this and in the end it is a judgment call, but I feel the selection of Witt is a good one.
Unlike picking a best player in the NHL, this selection is never very clean cut. If a player is really bad, he won’t stay in the NHL very long and we likely won’t notice him. I pick the worst player who is playing a regular role with his team. I imagine that at any given time, the actual worst player in the NHL is somebody who is not significant who will only play a couple NHL games in his career.
As for the numbers you look at, I think giveaways is a poor start. A player who gives away the puck a lot musat possess the puck a lot - and that in and of itself keeps him from being the worst in the league. The NHL giveaway leaders are creative talented players who make a lot happen when they have the puck - and also cough it up when they fail. The leaders in the NHL right now are Alexander Ovechkin, Andrei Markov, Mike Green, Mike Richards and Derek Roy. One could argue that they all are all star calibre players. Trying to make sense of a number as a bad thing - when the leaders are star players doesn’t make a lot of sense.
+/- is interesting and can be adjusted in a couple different ways. Both as a rate stat and as a counting stat. Both give an interesting look, but neither are perfect. Last year, Radek Bonk was worst as a counting stat and Craig Adams was worst as a rate stat (however the counting stat adjustment fails when a player changes teams in mid-season as Adams did). That said, I would be more inclinded to pick Colton Orr as having been the worst NHL player last season. He is a man who put up 2 points in 74 games with the NY Rangers with a team worst -13 +/-.
My study over a few seasons shows me that generally the worst player in the league falls into one of two classes.
1) An energy player who the coach likes for his hard work - despite the fact he isnt very good. - Colton Orr
2) An aging “name” player who has been a longtime NHL player but is no longer good enough—Brendan Witt
Posted by PuckStopsHere on 01/21/09 at 02:32 PM ET