One of the big achievements in baseball sabermetrics is Bill James win shares system. This is a system that assigns wins to the individual players on baseball teams. It works very well to rate individual players. One goal in hockey sabermetrics is to come up with a similar system. Of course there are differences between hockey and baseball which make building a hockey system more difficult, but that hasn`t stopped Justin Kubatko at hockey-reference.com from trying. I am going to look at his system and its results. I think it is an important goal, but we are not able to do it properly yet. As a result, he has built an approximate system which can give unreasonable results in some cases.
Kubatko makes three conceptual changes from the Bill James system. First in the baseball system, a win is worth 3 win shares. In the hockey system one point in the standings is worth one point share. The hockey version of this is problematic because of point scheme that the NHL uses.
For my top 50 players list, I use the question: Which 50 players would I most want on my team for the upcoming season? As a result this list is more of a projection toward next season than a ranking of last year’s results. That is not to say that a player who played well last year will not be ranked well, but that will only happen if I think they will keep up their results.
Usually, I like to compare my list to that of The Hockey News, but for the first time in years, they failed to publish a list this year.
Here is my top 50 player list:
According to Tom Awad’s goals versus threshold system, Brian Elliott was the worst player in the league. He split the season between the Ottawa Senators and Colorado Avalanche and posted a .893 saves percentage, in a league where the average saves percentage was .913. Nobody played more games than Elliott (who played 55 games) and posted a worse saves percentage. He also allowed six of eight shots against in shootouts. This is a small sample size, but it directly cost his team points. Brian Elliott had a poor season.
Elliott has been an NHL starter for the last two years. He played 55 games a season each year. He has played himself out of the NHL starters pool. As a free agent this summer, Elliott signed as a free agent with the St. Louis Blues for only $600,000 for one season. This is a last chance for Elliott. If he fails, he is likely out of the NHL.
This week I posted the top 20 players by goals versus threshold. Now I am looking at the flip side. Which players cost their team the most goals when compared to a replacement talent. I am using Tom Awad’s goals versus threshold system which attempts to rank players by one number, which is the number of goals a player is alleged to have been worth beyond that which a replacement level player could have produced.
It is easiest to rank goalies because saves percentage can be relatively easily converted into a number of goals prevented. For position players, offence (or lack of it) is easiest to rank and defence is far more difficult. Those position players who will appear on this list are players who had a lot of ice time without scoring.
Here are the worst 20 players in 2010/11 by goals versus threshold:
The Brendan Shanahan research and development camp that occurred last week is a potentially significant event in NHL history. It is where the NHL tests all sorts of gimmicks to change the game. The problem is that they are largely gimmicks and that they attempt to solve the wrong problem. The main thrust of the NHL’s gimmicks has been to increase scoring. In fact Mike Brophy writes that the NHL should have bigger nets. That is a problematic idea that may lead to an asterisk season where old scoring records are beaten and goals mean very little.
The main problem is that the NHL has been full of gimmicks designed to increase scoring. There has been the trapezoid, points for losing, shootouts, phantom penalties in “obstruction crackdowns” and they haven’t worked so well and they haven’t significantly increased scoring. Most of all they have not made hockey any better.
When I looked at goals versus threshold, we saw Sidney Crosby was ranked 16th in the league. This is despite the fact that he only played 41 games before a season ending concussion. This is despite the fact that he was 32nd overall in scoring in the NHL. He ranks quite a bit higher in goals versus threshold than he does in the actual point race; when goals versus threshold preferentially ranks goaltenders near the top. How did this happen?
The biggest reason Crosby ranks so high in the goals versus threshold system is that offence is ranked in part by the ice time it receives. A player’s offensive contribution is essentially the number of goals he creates less the number of goals that would be created by a replacement player playing the same ice time.
NOTE: This post is updated with correct numbers. The original post had incomplete statistics
With all of the sabermetrics I have done so far this summer, nothing attempts to rate the total value of players relative to one another. Corsi looks only at puck possession in 5 on 5 situations and though that is an important repeatable part of hockey it is not a full measure of any player. The only relatively common statistical measure out there that attempts to provide one number to rank players relative to one another is Tom Awad’s goals versus threshold system. It produces at best an approximate value for players and as long as one keeps in mind that there are significant error bars on any number it is pretty good. Last summer, I wrote several pieces looking at the results and the mechanics of the system. I think it is valuable and interesting to look at its results in the 2010/11 season and take a further look at the successes and problems that it gives us.
This is my second attempt to list the top 20 players in the 2010/11 season by goals versus threshold, because the original numbers as calculated by behind the net are incomplete and are missing the end of the season:
One of the more interesting players last year in terms of sabermetrics was Zenon Konopka who played with the New York Islanders and signed this summer with the Ottawa Senators. He winds up near the bottom of raw Corsi lists but played a tough role on a poor team.
Commenter steviestevie saw this and commented It’ll be interesting to see what Konopka’s performance looks like when more numbers are taken into account. He’s not exactly a world-beater, but he was pretty much thrown to the wolves taking tons of d-zone faceoff for a club whose transition game last year can best be described as “Is Grabner on the ice? No? *#&$%^!” in the worst team adjusted Corsi list.
Last week I published the worst 20 players by team and zone adjusted Corsi ratings in 2010/11. Perhaps the most significant player on the list is Robyn Regehr, then of the Calgary Flames and now a Buffalo Sabre. Regehr has represented Canada internationally in both the Olympics and the World Cup, so he must have been a top player at one time. Regehr is 31 years old, so perhaps he is starting to slow down - though many players remain at all star level at much higher ages than that.
Regehr had a relatively poor raw Corsi, which was the worst on his team. Calgary had a better than average team Corsi so this makes his team adjusted rating worse. Regehr played against a tough quality of opposition. He had the third toughest quality of opposition of any defenceman who played a significant number of games last season (behind Nicklas Lidstrom and Brent Seabrook), but he was not used preferentially in his own zone.
In an effort to put Corsi ratings into context to make sense of them, I have corrected the raw ratings for both team and zone start effects. Team effects are important because good teams will tend to attempt more shots and have fewer shots taken against them than weaker teams. Zone start effects are important because players who are on the ice for offensive zone faceoffs will tend to take more shots and players who are on the ice for defensive zone faceoffs will tend to have more shots against. This makes a pretty good individual indicator of puck possession abilities. The top players had good seasons and the worst players had weak seasons.
One look at the worst player list shows that Cam Fowler of the Anaheim Ducks was the worst adjusted Corsi player in the league. His -227.12 team and zone adjusted Corsi is the worst among players who played 50 or more games with one team in 2010/11. It is entirely possible that Paul Mara and Brent Sopel did even worse in terms of puck possession, but as they played with two different NHL teams in 10/11, this makes their team adjustments less certain so they are removed from the study.
About The Puck Stops Here
Who am I? A diehard hockey fan.
Why am I blogging? I want to.
Why are you reading it? ???