Quick Numbers: Predicting Jets offense
By Garret Hohl
7 years ago
© Bruce Fedyck-USA TODAY Sports
The Winnipeg Jets scored goals at an above league average pace for 5v5 thus far this season.
How much of this is sustainable? Which players should we expect to be on the upswing for the remainder of the season? Which players should we expect to be on the down swing?
My method with looking into this is extremely simple, and I have no doubt I could improve upon this. I simply looked at each player’s individual expected goals and then weighted their primary and secondary assists.
Expected goals are simply a player’s shots weighted by variables shown to have impact on shot quality (distance, angle, handedness, rebound, rush shot, etc), as well as a player’s personal regressed shooting percentage relative to league average.
The weighting for assists are based on DTM’s Box Plus-Minus model, where primary and secondary assists play a relatively different level of importance for forwards than defenders.
If I were to take things further in the future, I would use a player’s assists production by looking at on-ice expected goals and their regressed historical share of assists per on-ice goal.
- The Jets top four scorers in Nik Ehlers, Bryan Little, Mark Scheifele, and Patrik Laine all have been a bit over performing, but by very little.
- In fact, there is not much over performance by the Jets, aside from Andrew Copp who has been on a shooting percentage hot streak, Tyler Myers who has very little sample, and Shawn Matthias who has scored quite the large share of goals.
- Mark Stuart seems like an over performer, but that’s only due to not regressing assists.
- There are some substantial under performers, mostly from players that the Jets are looking for secondary offense from.
- Matthieu Perreault has been struggling with an extremely abnormal shooting percentage both this and last season. The Jets are hoping for more goals from the forward so hopefully things will come back for the feisty forward.
- Nic Petan is another obvious under performer. It’s amazing that his assists rates are what they are given the low level of Quality of Teammates he has suffered through the season. Unlike Perreault, xGoals is assuming Petan is around a league average finisher due to not having much recent history to go off of.
- The correlation coefficient between predicted and actual is about 0.8.
For context the median first liner puts up 2.0 points per sixty minutes, while second, third, and fourth liners come up at 1.73, 1.42, and 1.04 points per sixty minutes. With defenders, the median scoring for each pairing goes 0.74, 0.66, and 0.56 points per sixty minutes.
Recent articles from Garret Hohl
