The Biggest story in professional hockey this week was the unveiling of the rosters for the 2016 World Cup of Hockey.
As Oilers fans are well aware, Taylor Hall did not make the cut for Team Canada.
Balderdash! Poppycock! Horse Hockey!
These were all words that I would use to describe that decision. As Oilers fans (or at least most of us) are well-aware, Taylor Hall has an offensive skill set which allows him to take over a game, even when playing against the best competition the league has to offer. Furthermore, Hall's defensive game has evolved by leaps and bounds in recent years. Not only can Hall score, he is also among the world's best in the neutral zone, and those abilities make him one of the league's best drivers of play.
But rather than tear down one of the Team's weaker picks, while advocating for Hall, I thought it would be a good idea to construct a statistical model to identify other Team Canada snubs.
The model I created was based on four stats.
- Adjusted CF%
- Adjusted xGF%
- 5 on 5 points/60 minutes
- 5 on 5 primary points/60 minutes
I outlined the method for the adjusted stats in this article.
I pulled data for every player with at least 1500 even-strength minutes since the start of the 2013-14 season (529 skaters). I found the rank of each skater in each in the five aforementioned statistical categories. I then took the average of these four ranks as the model's output.
All data was found at corsica.hockey.
Here are the results:
Hall makes the list!
But, he does so as the 13th forward. There are certainly a few players I didn't expect to see in front of him. Stone, Hoffman and Gallagher were not only subbed from the team, they were snubbed from the category of notable subs. That's a lot of snubbing!
These are the seven defencemen who my model picked out. I actually wrote another article, using a different model to pick Canada's blueline, and got different results. Barrie and Ellis were the surprises here, for me. Doughty not making the cut is probably my biggest surprise. He and Kris Letang did much better in my other model, which included power play production.
I went with adjusted on-ice percentage stats, rather than modified on-ice percentage stats (also explained in the link I posted in the 'model' section). This may be the reason why a few players from terrible teams ranked very high. I suspect that it's tougher to outperform good teammates than bad ones. Even still, the players who ranked high on these lists are all extremely good.
Offence gets double-counted to a certain extent. Players who score a lot also tend to drive offensive play, so they'll rank highly in every category, whereas a player who predominantly excels at shot suppression only performs well in two categories.
Special teams aren't accounted for at all.
All 529 players were ranked against each other. Obviously defenders ranked much lower in the scoring categories. This provides a possible advantage to high-scoring defencemen, because there are more players in the meaty part of the curve.
Keep in mind that these aren't the players who I think should be on the team. But I think this model does a good job at identifying strong candidates to make any international team. As always, remember to keep the caveats in mind.