While conducting research for the Statistical Analysis in the NHL blog which was posted in July, I became acutely interested in the potential of this “science” as it pertains to player evaluation. Statistical Analysis is now fairly prevalent in the baseball world as a number of General Managers either have backgrounds in the field and/or have started departments within their respected front offices dedicated to statistical research.
As I stated in the Statistical Analysis blog, there was far more activity in the NHL in this field than I expected. I mentioned several sites in that post but have since found many more. Included in these other sites are some new “advanced statistics” that I plan on dissecting to see how they play out versus more traditional player evaluation methods.
First off, I’d like to issue a quick disclaimer to the folks reading my posts; I am NOT advocating for statistical analysis to replace traditional scouting methods as the primary tool in player evaluation. I do believe that statistical analysis can become a valuable tool to help augment the traditional methods of player evaluation though.
Statistical analysis IS a science. As of yet, no one has either developed or claimed to have developed, a stat or a formula which will correctly predict or evaluate any player’s value or performance to an accuracy level of 100%. Everyday intelligent people carefully consider and debate the merits of certain stats and their actual relevance to on-ice success. By no means is the “science” perfect; however that doesn’t mean that the research should be dismissed out-of-hand because it doesn’t conform to the more traditional methods of player evaluation. Hockey is a game; the NHL is a business. Operating a successful organization in the NHL (especially in today’s salary cap landscape) requires the willingness to use modern business techniques. Statistical analyses can be one of those techniques.
Statistical analysis is not only about evaluating stats like goals and assists and On-Ice/Off-Ice +/- rating, it’s also about properly managing assets. A lot of Managers seem to just pluck numbers out of the air when it comes to offering contracts to free agents. With the amount of money that we’ve seen thrown around recklessly at free agents it is clear that NHL Managers could use every tool at their disposal to properly value players and determine what they should be paid.
Some old-school hockey fans believe that statistics don’t make winning teams and they are unwilling to even consider the possible value of sabremetrics to the NHL and its teams. They point to team chemistry and heart as intangible elements that go into building a Stanley Cup contender. To that I have to agree whole-heartedly. But just the fact that those qualities are “intangible” and can not be quantified makes the art of building a winning team that much more difficult if you don’t look at some of the advanced statistics as well. Statistical analysis only evaluates the tangible aspects of players and teams and what makes them successful on the ice.
The chances of building a Stanley Cup winning team around 20 nice, hard working guys that all get along is just about nil. Look, the Rangers team of last season reportedly got along famously and liked each other and was willing to go to “battle” for one another. Where exactly did that get them? Booted out of the first round of the playoffs after a lackluster, up-and-down regular season, that’s where.
Don’t get me wrong, chemistry is a key component of on-ice success. Unfortunately there is no book or manual that tells you which players will combine to form the perfect “chemistry” that will lead to success on the ice. It’s a crap shoot and 29 Managers every year get it wrong. Statistical analysis is math and math doesn’t lie; it can only be misinterpreted.
Alright, I guess that was a little longer than I anticipated. I only ask of those that read these pieces to keep an open mind as I will also. I expect that we will find holes and flaws with most, if not all, of the existing “advanced statistics. Though this science is very new to the sport of hockey and is somewhat controversial, I feel that this topic has the ability to inspire a lot of interesting discussion and that’s my main goal here.
That’s it for my disclaimer; now it’s on to the introduction. Today I’d like to introduce an “advanced statistic” referred to as a “Corsi Rating” or a Corsi Number”. I found this stat on BehindtheNet.com. The Corsi Rating or Number is named after current Buffalo Sabres Goaltending Coach, Jim Corsi. Apparently he devised his rating system as an alternative to +/- and the Sabres use it in its evaluation of players. I can’t tell you exactly how long the Sabres have used this rating but the earliest references I have found were from November of 2007 on hockeynumbers.blogspot.com and an October 2008 post on Japersrink.com.
Basically Corsi’s rating is the difference between the number of shots DIRECTED toward the offensive goal versus the number of shots DIRECTED toward the defensive goal while in 5-on-5 skating situations and excluding empty net shots. The key word in that definition is DIRECTED and it includes all shots that are on net, miss the net or are blocked. I would guess that the rationale behind this stat is that the larger the discrepancy between shots directed for and against, the more that team is controlling the play.
Let’s look at the top five skaters with at least 75 games played from 2008-2009 in terms of highest Corsi Rating per 60 minutes of ice time:
David Moss CGY +23.8
Pavel Datsyuk DET +23
Henrik Zetterberg DET +20.1
Alexander Ovechkin WSH +19
Mikael Samuelsson DET +18.9
The interesting thing I notice is that there are 3 Red Wings players in the top 5. In fact 7 of the top 9 are Red Wings. Not surprisingly the Red Wings led the NHL in Shots on Goal during the regular season with an average of 36.2 per game. Washington was second with 33.5 while Calgary was eighth finishing with an average of 32.2 Shots on Goal per game. While Shots on Goal varies some from the criteria of Shots Directed on Goal used by the Corsi rating it still shows to be a good indicator of which teams have players finish with higher Corsi Ratings.
The entire Corsi Rating System is more complex than the simple Corsi Number. According to an article posted on the Irreverent Oiler Fans page on vhockey.blogspot.com, it actually goes into more detail than just the number itself. It also includes information about which part of the net the puck is shot into and how many crossbars or goal posts are hit. I don’t know how the Sabres factor that information into an equation or how that helps them evaluate a player but apparently they do.
I guess if I had to judge the merits of the Corsi Number my first complaint would be that not all shots (whether they end up on net or not) are created equal. Most hockey analysts prefer scoring chances to shots on goal to determine which team controlled the play offensively. Scoring chances are of course, shots or opportunities in which an offensive player has a better than even chance of scoring a goal. They can also account for shots that hit the cross bar or the goal post or just miss the net in some cases. Those shots aren’t recognized as shots on goal but they can still count as quality scoring chances.
Another part of this stat that can be misleading is the fact that it doesn’t take into account the strength of the opponents or the weaknesses of your teammates when used to evaluate individual players. Granted, the idea is that over the course of a full season, all teams will have faced a fairly equal quality of opponents. The problem lies with the players who miss time due to injury or what have you. They may play a much different quality of opponents if they’ve missed a tough or easy stretch of opponents than someone else. This statistic, like most, loses some of its luster unless applied to players who play a minimum number of games which is why I only analyzed players with a minimum of 75 games played.
All-in-all, while I think this stat is better than the traditional +/- figure, I don’t think it is nearly enough in and of itself to properly evaluate the value of an individual player’s contributions. Hockey is a team sport and a player can be affected either positively or negatively by their teammate’s performance too much. I also prefer scoring chances as an indicator of offensive performance better than shots on goal or shots directed toward goal.
That’s it for today folks. Join me next time for another exciting piece breaking down another advanced statistic or comparing players using an advanced statistic. Please let me know your thoughts on the Corsi Number specifically or advanced statistics in general.