Kukla's Korner

The Puck Stops Here

Stats Which Tell Us Little

A few days ago Scott Cullen of TSN published win loss records for individual players who missed considerable numbers of games this season.  For the most part this is not a very good method to learn anything statistically.  To show this, here are the ten players who had the biggest improvement to their team’s win/loss percentage from Cullen’s stats.

Top 10 Players By Increase In Win-Loss Percentages With Them In LineUp
Player  
Team   
Record In LineUp   
Record Out Of LineUp   
Difference In Winning Percentage
Chris CampoliOtt42-19-52-12-1

.507

Tyler KennedyPit39-16-67-11-1

.394

Marc-Edouard VlasicSJ42-11-108-9-1

.374

Rod PelleyNJ41-16-45-11-3

.363

Colin FraserChi47-15-64-7-1

.360

Darroll PowePhi36-21-45-13-1

.360

Blair BettsPhi35-21-55-13-1

.326

Evgeni MalkinPit41-18-65-9-1

.310

Brett CarsonCar29-19-56-17-5

.308

Brett ClarkCol37-19-86-10-0

.286



This list is basically a random list of players who were out of the line-up for not significantly more than 11 games (the minimum number to be included in this study).  When a player misses a small number of games, randomly some will will happen to have a poor team record when they are out of the line-up.  There is little correlation between this and the player’s contribution to his team.  The only one of these players who could arguably be called the most important player on his team is Evgeni Malkin (though I would argue Sidney Crosby is better).  For the most part, these are players who are bit players on their teams.  Clearly, any signal that may be in these numbers is so deeply lost in random noise that it cannot be extracted.

Here are the ten worst players by this method.

Worst 10 Players By Increase In Win-Loss Percentages With Them In LineUp
Player  
Team   
Record In LineUp   
Record Out Of LineUp   
Difference In Winning Percentage
Ryan ParentPhi18-24-522-10-1

-.246

James SheppardMin26-31-712-5-0

-.245

Sean AveryNYR29-30-108-3-0

-.234

Ted PurcellTB5-11-127-25-11

-.192

Fernando PisaniEdm8-25-518-21-3

-.188

Andrew PetersNJ13-13-333-14-4

-.186

Fredrik ModinLA6-7-439-20-4

-.180

Rob NiedermayerNJ38-25-68-2-1

-.179

Robert LangPhx37-22-513-3-1

-.177

Mike CammalleriMon28-28-811-5-1

-.176



This is another group of bit players.  They also have small sample sizes either of games missed or of games in the line-up and random statistical fluctuations are the majority (if not all) of the signal.  In some cases, the dropoff in team success can be explained by other effects (for example - Edmonton suffered their most serious injuries in time for Fernando Pisani to make his comeback to the line-up).  These players do not play enough to be the important player who drives the difference in their team’s winning percentage.

This study shows that there is little value to comparing team winning percentages with and without a given player in the lineup.  Hockey is a team game and one player only has a limited effect on his team (in almost all cases).  The change in the winning percentage is rarely due to the player in question.  When a player is good enough that he makes a significant difference to his team, it is clear that random fluctuations distort any signal.

The problem is that people often try to use numbers like this to try to make determinations of who should win awards.  For example, it is common to use the fact that Pittsburgh did not have as good a record with Evgeni Malkin out of the lineup as an argument against Sidney Crosby’s Hart Trophy case, when this shows it is likely just statistical fluctuations.  This is an extremely poor method to learn anything about how valuable a player is to his team.  It generally leads to incorrect reasoning and unsupported assumptions that are often believed as being statistically valid and used as arguments for trophy.  As this shows, it proves little.

Filed in: | The Puck Stops Here | Permalink
 

Comments

PaulinMiamiBeach's avatar

The only one of these players who could arguably be called the most important player on his team is Evgeni Malkin (though I would argue Sidney Crosby is better).

did anybody argue that this particular stat means these players are the most important players on their team?

For the most part, these are players who are bit players on their teams.

bit players are important in hockey.  a team full of Malkins and Crosbys won’t win.

Clearly, any signal that may be in these numbers is so deeply lost in random noise that it cannot be extracted.

that’s an interesting hypothesis.  one worth exploring - by looking at these players’ and their teams’ past history of win % when these players were out of the lineup.

although it would be difficult to quantify statistically, it’s also important to consider who filled their roster spots while they were gone.  as usual with hockey it all boils down to “you’ve got to watch the games.”  I’ll bet the opinion of experienced fans of these various teams would give us more insight into the true value of these players.

Posted by PaulinMiamiBeach on 04/19/10 at 10:06 PM ET

Avatar

Paul - PuckStopsHere is completely right on this one.  You need hundreds of games in order to use the “With or Without You” method.  Saku Koivu over his entire career in Montreal is a pretty good example. 

One thing you can figure out over a single season is what the average “good player” (say a D who gets 21+ minutes of ice time) is worth to his team.  When he’s out, everybody’s ice time goes up a little and it’s like you replaced #1 with #7.

Posted by Hawerchuk on 04/20/10 at 11:41 AM ET

Avatar

@Paul: A team full of Crosby’s and Malkin’s would be an unstoppable juggernaught, and would shatter the single-season team point total record. It would also abolish world hunger (by feeding everyone crisp, cross-ice passes into the slot), settle all religious disputes (there would be only one God, and it would be the man who put together that team), and would solve the energy needs for countless future generations (and that would just be from the 4th line)! See, isn’t it fun to speculate on purely hypothetical situations?

Now, for the original purpose of my comment… to add on to just how meaningless these kinds of stats are.. last year, there were TONS of claims in the MSM that the Rangers couldn’t win without Avery in the line-up. He added that intangible desire to win, or provided that spark the team needed so desperately, or was the <insert tired hockey cliche> that his team needed to get them “over the edge”, blah blah blah… This year, the Rangers were 8-3 without him in the line-up… if they kept up that kind of (insanely small sample-size alert) play for an entire season, they’d… well, they’d finish a depressing second place to the All-Crosby-Malkins… but everyone would be pretty happy with the effort, and would probably talk about how “gritty” they were… or something like that.

Posted by Hilleraj from LeafLand on 04/20/10 at 12:03 PM ET

Avatar

I would have to say these stats are not exactly the bets indicators to a players worth to a team. Outside of Malkin this a poor litmus test. Modin is injured half the season anyway so you can throw that stat out the window since playing 17 games by that count hardly counts as an adequate test. I agree with

You need hundreds of games in order to use the “With or Without You” method.

because most of these players were either out of the lineup due to being benched etc. Now i dont know the ins and outs of every player on this list and their contributions to the teams pk or what not but it seems pretty worthless to draw conclusions for one season. Chart a players worth over 5 seasons and then draw conclusions throwing out time missed due to sitting up in the press box blah blah blah.

Just my opinion

Posted by Zach from Canada on 04/20/10 at 02:07 PM ET

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.

Smileys

Notify me of follow-up comments?

Feed

Most Recent Blog Posts

About The Puck Stops Here

imageThe Puck Stops Here was founded during the 2004/05 lockout as a place to rant about hockey. The original site contains over 1000 posts, some of which were also published on FoxSports.com.

Who am I? A diehard hockey fan.

Why am I blogging? I want to.

Why are you reading it? ???

Email: y2kfhl@hotmail.com