Next up is the defensemen. I will bring up a few lists, and discuss them a little bit. If you want to take a deeper look into LS-GAA, feel free to download the LS-GAA file under Data. Then you can do your own research.
Here are the top 20 defenders since 2007:
Not a bad list, but there are some obvious players missing. Drew Dougthy is ranked 44th and Duncan Keith is all the way down to 89th. I will look at these two players specifically later on.
But for now let’s look at LS-GAA per 60 minutes. The minimum time on ice is 4000 minutes.
I won’t discuss the numbers here. Instead I will leave it up to the readers, to make their own evaluations of the lists.
Here are the best defensive defensemen at even strength (TOI_ALL>4000 minutes):
The list is a combination of defensive defensemen (bad offense – good defense) and great two way defenders (good offense – good defense).
And here come the best offensive defenders at even strength (TOI_ALL>4000 minutes)
|Erik Gustafsson (D)||3774||20.83||-12.30||0.331|
The only player to make both lists is Charlie McAvoy.
Before we turn our attention towards this season, I would also like to look at the worst players. There are some fairly big names on that list. I think some of these players may be better than the numbers suggest, but they have either played in too big a role or played for a really bad team. Most are just bad players though.
The model really likes Ryan Ellis, while I think most hockey experts would rank Roman Josi higher this season.
Drew Doughty and Duncan Keith
These two players are interesting to take a deeper look at. Both players (especially Doughty) are considered to be amongst the very best defenders in the last 10+ years, so why doesn’t the model agree? Some of that can be answered by looking at the development graphs of the two. It’s a graph of LS-GAA total over time – the x-axis represents the years, so year 8 is season 2007/2008, year 9 is season 2008/2009 and so on.
I will discuss Duncan Keith first. From 2007/2008 to 2015/2016 he was a very good defender, and if you only look at these years he was the 14th best defender. That still seems a bit too low to me considering how dominant Chicago was in this time frame. Since then he has been really bad. In fact, he was the worst defender in the entire league in 2017/2018 according to the model. How much of this decline is because of age and how much is because Chicago has declined? Keith is probably not a top player anymore, but he is probably not league worst either.
The story about Drew Doughty is similar although the decline has not been as extreme, but he’s also a bit younger. In general, the model seems to be a lot lower on Doughty than the consensus thinking. I think most people considered him a top 5 defender a few years back, but the model never had him that high. So either he’s overvalued by experts or he’s undervalued by the model or maybe truth lies somewhere in between.
Have Doughty and Keith declined because their teams have declined? Or have their teams declined because Doughty and Keith have declined?
That’s it for now. Next blog will be about the goaltenders.
- Ryan Ellis is apparently really good according to the model – Drew Doughty not so much.
- Any statistical analysis should always be combined with actual scouting, but the stats can show you things you would otherwise miss.
Stay safe and remember to be kind
All raw data from evolving-hockey.com