In part 3, I am going to define 1st, 2nd, and 3rd pair defensemen based on performance.
I explained the process in part 1, but I'll go over again what I did for D-man. Again, big props to Tyler Dellow for giving me this idea and overall having a awesome site.
I took every defensemen from the 2007-2008 season to the 2011-2012 season and sorted them by time on ice per game (highest to lowest). I then separated the defensemen from 5 seasons (note that defensemen are represented by individual seasons, thus a D-man who played all 5 years show up 5 different times in the data) by games played. Thus, the first pair defensemen played roughly the same amount of games as the third pair of D-man. This is of course why there are far more third line d-man in the below chart, since third pair d-man often rotate in and out of the lineup.
Remember this is all even strength data.
|Defensive pair||players||avg GP||TOI/60 even strength||Max minutes per game||Min minutes per game||Rel Corsi / 60 min||Corsi On / 60 min||Rel Corsi QOC||Corsi QOC||On ice SH %||On ice SV%||PDO||GF/60||GA/60||Rel GF/60||Rel GA/60 (negative is a good)||GD/60||Rel GD/60||O zone start %|
Quick notes on some of the columns. Each column is how the average 1st pair, 2nd pair, and 3rd pair D-man performed. All stats are weighted by minutes played, thus a player who played 10 games will not affect the averages as much as someone who played 70 games. Rel GF/60 and Rel GA/60 are the weighted averages of the players On ice GF (GA)/60 minus Off ice GF (GA)/60 (basically similar to Rel Corsi). For Rel GA/60 , a negative number is a good thing. This means you are letting up less goals while on the ice then the team is when that player is off the ice. A positive number is bad.
We see some pretty obvious trends. 1st pair d-men play more games, more minutes, play tougher competition (Rel Corsi QoC), drive possession forward more (rel Corsi), and have better Goal differential numbers than 2nd and 3rd pair defensemen.
So this data tells us what an average 1st, 2nd, and 3rd pair defensemen should look like (statistically).
This next chart breaks up each of my 3 groups into three smaller groups by Rel Corsi. Again, I tried to make the games played equal in each group. Thus, now we have a way of determining at what level a player is performing versus other 1st liners, 2nd liners, 3rd liners, or 4th liners.
|Defensive pair||players||avg GP||TOI/60 even strength||Rel Corsi / 60 min||Corsi On / 60 min||Rel Corsi QOC||Corsi QOC||On ice SH %||On ice SV%||PDO||GF/60||GA/60||Rel GF/60||Rel GA/60 (negative is a good)||GD/60||Rel GD/60||O zone start %|
So now we not only can determine what pair a defensemen is (in terms of minutes played), but we can now compare his performance to other guys who play similar minutes.
This concludes part 3. In part 4 I will use the above charts to look at the the Jets defensemen.