clock menu more-arrow no yes mobile

Filed under:

Winnipeg Jets three-quarter poll player evaluations

A deeper look into how our Jets players have fared thus far.

Marianne Helm

The trade deadline is looming and people everywhere are talking rosters: who should stay, who should go, who will stay, who will go, etc. etc. Instead of attempting to answer any of these questions, this article will give you additional information on the Jets players and help you decide.

Corsi, shot attempts which are a close proxy to puck possession, is one of the better statistical player evaluators out there and a strong predictor of future success. It basically shows what plus/minus aims to tell you about a players overall contribution.

Stephen Burtch of Pension Plan Puppets has constructed a few algorithms based off of some regressions of Corsi events for and against. He has kindly given me the algorithm to run on Winnipeg Jets players. The regression looks at the main factors widely available that are known to affect a player's Corsi: how much they play, who is on the ice with them, who is on the ice against them, and how much they are deployed in either the offensive or defensive zone. The algorithm essentially looks at what is the norm level of Corsi events for and against for a player to see given the same situations. Then, one can see how a player is able to affect the game in repressing attempts against or promoting attempts for.

Keep in mind that some of these players are still quite short on sample size; Burtch estimated that a 2-3 season sample is best for much of this. Also note that being the best relative to usage does not always suggest best player overall.


Shot attempt facilitation

Name CF20 exCF20 dCF20
Eric O`Dell 19.46 14.66 +4.80
Evander Kane 20.75 18.22 +2.53
Michael Frolik 20.13 17.95 +2.17
Andrew Ladd 20.57 19.32 +1.25
Bryan Little 20.38 19.52 +0.86
Eric Tangradi 17.87 17.18 +0.69
Devin Setoguchi 18.57 18.51 +0.06
Jim Slater 17.00 17.13 -0.13
Olli Jokinen 18.00 18.98 -0.98
Mark Scheifele 18.24 19.65 -1.41
Anthony Peluso 15.25 17.28 -2.03
Blake Wheeler 18.51 20.76 -2.25
James Wright 15.31 17.78 -2.47
Chris Thorburn 14.75 18.51 -3.76
Matt Halischuk 15.81 21.66 -5.85


* You can see here that sample size is likely effecting O`Dell's numbers. His Corsi For numbers aren't that high but his expected numbers are incredibly low. We'll need more sample in order to judge him.

* There's not that much surprise that Kane, Frolik, Ladd and Little have good numbers here. They are pretty good at the puck possession game and creating offense out of it.

* There is some surprise that Scheifele and Wheeler's are so low. Scheifele has spent ~300 of his 850 mins with one of the two bottom dwellers Halischuk and Thorburn, but on the other hand all of the Jets forwards that have spent 100+ 5v5 mins with and without Scheifele have fared better without Schiefele, except Thorburn. It is true though that most rookies do not tend to do well at the start of their pro-careers in possession. What's your excuse Wheeler?

* Not really surprised that James Wright, Chris Thorburn or Matt Halischuk are near the bottom here.

Shot attempt repression

Name CA20 exCA20 dCA
Eric Tangradi 15.14 18.91 -3.77
Eric O`Dell 15.14 17.95 -2.81
James Wright 16.86 19.35 -2.49
Bryan Little 18.95 20.82 -1.87
Andrew Ladd 18.87 20.71 -1.84
Anthony Peluso 16.94 18.27 -1.33
Jim Slater 18.37 19.51 -1.13
Olli Jokinen 19.22 19.83 -0.61
Michael Frolik 18.97 19.26 -0.29
Evander Kane 19.28 19.32 -0.04
Blake Wheeler 19.27 19.24 +0.03
Chris Thorburn 19.10 18.94 +0.17
Mark Scheifele 19.44 19.05 +0.39
Devin Setoguchi 19.69 19.25 +0.44
Matt Halischuk 19.97 18.68 +1.30


* Tangradi this one impressively. While sample size may be an effect, Tangradi was above average in this category last season as well. So while he may not be the Jets best defensive forward, he does seem to have an ability to prevent chances against.

* O`Dell's numbers are impressive, although they are only at about 80 mins (as are Slater's) which for example is 10% of the sample size of Andrew Ladd.

* Speaking of Ladd, it is not much surprise that Ladd and Little are near the top here. Ladd was a top 3 in this category last season as well. Little did not post as solid numbers here but much of it was likely Little playing d-zone specialist for much of the season with Slater injured.

* This does not paint a pretty picture for Setoguchi, who has been known for a while for not being the best defensively.

* Oof! While sampling bias may be making it more severe, Halischuk has always been known as a bit of a possession sink and this is real bad. He is about 3x worse than the next Jet.

Overall effect on puck possession and predictive modeling

Name Corsi% exCorsi% dCorsi% TOI/60 Line Percentile exGD/60
Andrew Ladd 52.2% 48.3% +3.9% 13.97 1 65 +0.35
Evander Kane 51.8% 48.5% +3.3% 15.65 1 60 +0.31
Bryan Little 51.8% 48.4% +3.4% 14.25 1 60 +0.31
Michael Frolik 51.5% 48.2% +3.3% 13.90 1 60 +0.27
Blake Wheeler 49.0% 51.9% -2.9% 14.28 1 30 +0.01
Olli Jokinen 48.4% 48.9% -0.5% 13.96 1 25 -0.06
Mark Scheifele 48.4% 50.8% -2.4% 13.92 1 25 -0.06
Devin Setoguchi 48.5% 49.0% -0.5% 13.32 2 25 -0.09
Eric O`Dell 56.2% 45.0% +11.2% 5.96 4 95 +0.06
Eric Tangradi 54.1% 47.6% +6.5% 8.33 4 90 -0.07
Jim Slater 48.1% 46.8% +1.3% 7.44 4 55 -0.44
James Wright 47.6% 47.9% -0.3% 7.28 4 50 -0.47
Anthony Peluso 47.4% 48.6% -1.2% 5.58 4 50 -0.48
Matt Halischuk 44.2% 53.7% -9.5% 10.29 4 25 -0.68
Chris Thorburn 43.6% 49.4% -5.8% 9.16 4 20 -0.72

Note: I have decided to look at overall contribution in two different types of context. First one is just looking at the player's Corsi% difference from expected Corsi% given the regressions used above. The second is based off of this article; it uses a players 5v5 TOI to determine what "line" they are being used as, and then compares what percentile their Corsi% is relative to their peers (higher being better), and then gives the expected goal differential one would normally get for that Corsi%. I highly recommend reading the article and looking at the patterns.


* Right off the bat you may notice something weird about the Jets 5v5 TOI: there is a huge gap in it. The Jets have seven forwards in the top 100 5v5 TOI/GP. The norm for a fourth line is between 9-10.5 mins, which only two Jets forwards fall into, and mostly because they've spent large chunks in the top 9. Part of this is problems with depth but part of this is from mismanagement in player usage, predominately from the Noel-era

* Ladd, Kane, Little, and Frolik have all done well. All of them have taken big minutes and have performed above average relative to their peers in both measurements. They have pushed the play well and haven't fared too bad in the scoring department.

* Wheeler has never been a player that can push the play forward as well as the previously mentioned players; he does however score very well, especially when placed with possession players. Unfortunately, Wheeler spent a chunk of time with Thorburn in a few occasions (which I had advised against) which pulled down his numbers.

* Jets centre depth is still a problem. Jokinen has not been a top-six 5v5 centre for many years now and Scheifele - while promising - still has a lot of growth before he is ready for top line minutes. If the Jets had depth this wouldn't be as large of a concern, as these players could be given second line or third line minutes.

* The Jets evaluation of whom their best players are have been accurate, but their evaluation of their best fourth line players have been off the mark. Halischuk and Thorburn have performed the worst out of the Jets defensive players, yet these are the two who have been given top-nine duty whenever injuries have arisen.


Shot attempt facilitation

Name CF20 exCF20 dCF20
Grant Clitsome 20.40 19.10 1.30
Dustin Byfuglien 19.58 18.70 0.88
Mark Stuart 18.08 17.64 0.45
Adam Pardy 18.11 17.72 0.39
Zach Bogosian 18.82 18.89 -0.06
Jacob Trouba 18.27 18.58 -0.31
Toby Enstrom 18.32 19.57 -1.26
Keaton Ellerby 16.49 18.39 -1.90
Paul Postma 15.92 18.41 -2.49


* Clitsome was injured early and has likely been influenced by sampling bias since his career norm is about 17 for CF per 20 mins, and the few minutes he has is almost all with offensive-defenseman stalwart Byfuglien.

* Stuart being so high and Postma being so low is quite a surprise. My guess is mostly sample size, although there is also some evidence showing that defensmen have much greater control over CA20 than CF20. Postma, like Clitsome, CF20 is not quite what he usually his career -however small- 19 per 20 min.

* Enstrom's CF20 is one of the higher values for the Jets, but his expected values are extremely high as he spent a good amount of time with Byfuglien logging heavy minutes.

* Ellerby has been in a decline for CF20 over the last 4 seasons; hopefully this trend ends if he's to become a regular on the Jets blue line.

Shot attempt repression

Name CA20 exCA20 dCA
Adam Pardy 16.04 19.38 -3.34
Tobias Enstrom 17.84 20.12 -2.27
Paul Postma 16.49 17.96 -1.47
Dustin Byfuglien 18.70 19.72 -1.02
Grant Clitsome 19.02 19.38 -0.36
Keaton Ellerby 19.11 18.43 0.68
Zach Bogosian 19.38 18.60 0.77
Jacob Trouba 20.36 19.13 1.23
Mark Stuart 21.44 18.35 3.09


*Pardy leads the pack here. I have noticed he has been particularly good at preventing opponents breakouts, although he also has had extremely extremely sheltered minutes which could also be a cause.

* Stuart's CA20 numbers are abnormally high even by his standard. While he's not the best at getting out of the zone to prevent his team from being pinned in, he has been a bit better under Paul Maurice and his possession numbers have improved with it.

* I find the order of the list very interesting relative to how "good" these defensemen are viewed defensively. It shows how a player may be able to make up for their defensive inefficiencies by keeping the puck out of their own zone.

* PS: Enstrom is totally only offensively minded.

* PPS: Poor Trouba, always with Stuart. I'm sure Stuart is a good mentor as he plays hard, is committed to fitness, etc. But, he is a bad defensive partner. Not the same thing.

Overall effect on puck possession and predictive modeling

Name Corsi% exCorsi% dCorsi% TOI/60 Pair Percentile exGD/60
Dustin Byfuglien 51.1% 48.7% 2.4% 17.54 1 60 0.120
Tobias Enstrom 50.6% 49.3% 1.3% 17.47 1 55 0.080
Zach Bogosian 49.3% 50.4% -1.1% 18.15 1 40 -0.026
Grant Clitsome 51.7% 49.6% 2.1% 16.12 2 60 0.158
Jacob Trouba 47.3% 49.3% -2.0% 16.41 2 25 -0.162
Adam Pardy 53.0% 47.8% 5.2% 12.83 3 80 0.069
Paul Postma 49.1% 50.6% -1.5% 11.83 3 60 -0.181
Keaton Ellerby 46.3% 50.0% -3.7% 13.53 3 20 -0.360
Mark Stuart 45.8% 49.0% -3.2% 14.48 3 15 -0.392


* The Jets TOI spread seems much better than the forwards.

* It remains that Byfuglien and Enstrom are still the Jets only proven above average top pair defensemen. Bogosian is not far off and has done well when asked, but he remains more of a super-elite #3 than a top pair d-man.

* Trouba is hard to get a handle of where he should be due to being with Stuart so much. Trouba has a near 50 Corsi% when away but around 45% when with Stuart. It is however extremely promising given that he is only recently turned twenty.

* One author at Arctic Ice Hockey isn't the greatest fan of Clitsome, but I still think he is the best of our depth defensemen.

Final Thoughts

There are good reasons to evaluate players with Corsi%:

1) Most people know who can and can't score intuitively quite easily, but it's much hard to tell who is assisting or harming puck possession

2) A player's affect on Corsi% is the input a player gives towards affecting the team's goal differential, and because of this we can extrapolate how much a player is affecting their team overall

The only problems are sample size and some players who may have spent too much time with much stronger or weaker players.