Already member? Login first!

Comments / New

A Statistical Look At Kyle Turris

In the last of the Kyle Turris related pieces this week, I want to present some statistical evidence both for and against Turris’ inclusion in the lineup.  It’s obvious how I personally feel, I believe, but I’m going to present this in the most unbiased way I can in order to lay the evidence out in front of you, the reader, and allow you to make your own judgement.

Predators Team Statistics

Let’s start by taking a look at some good old fashioned counting statistics at 5 on 5 and all situations (all data courtesy Evolving-Hockey.com):

I’ve sorted by the per 60 minutes rates for each category, but here you can see that Turris is in the upper half of the team in each of these offensive categories.

  • His three goals ties him with Austin Watson, Calle Järnkrok, Matt Duchene, Ryan Johansen, Mikael Granlund, and Mattias Ekholm; and puts him ahead of Craig Smith, Ryan Ellis, Rocco Grimaldi, and others. He is sixth on the team when adjusting for TOI.
  • His seven points puts him ahead of Smith, Watson, Fabbro, Hamhuis, Miikka Salomaki, and Matt Irwin.  He is ninth on the team per 60 minutes.
  • Three primary assists (the final pass before a goal) sees him again ahead of about half of the team, including Forsberg, Smith, Johansen, Josi, and Granlund.  He is eighth per 60 minutes.
  • His 2.06 expected goals lands him 12th on the team in total, and tenth per 60 minutes./
  • At all situations Turris grades similarly to his 5 on 5 play. However, his ranking via rates per 60 minutes sees him a little higher, especially in primary assists, where he’s fourth on the team in primary assists/60 if you don’t count Olivier’s small sample size.  This is likely due to his work on the power play./

These are not especially great numbers for someone who was supposed to be the second-line center of the future, but considering how many different linemates he’s had in this time (and the fact he’s now been out four games), and some of the people he is either tied with or ahead of, he’s producing well above deserving to be buried on the fourth line.

Isolated Impact

This chart—which I use quite a bit—attempts to measure the impact of JUST the player when they are on the ice, and excludes effects from teammates, competition, score, etc. at 5 on 5.

On top, Turris is doing well on offense—he’s producing a 10% offensive threat over league average, especially in the slot in front of the goaltender.  Defensively, the threat works the opposite way: he’s 1% worse on defense than league average.  This isn’t awful, but we’ll also return to the topic of defense later.

His shooting threat (basically how likely he is to score a goal when he shoots compared to average) is a respectable +3%.

To compare, we can look at Calle Järnkrok:

They’re essentially mirror-images of each other—Järnkrok is right around average offensively, but pretty good on defense.  Turris was getting PP time where Järnkrok is being used primarily on the PK.

On-Ice Summary and Usage

The top three bars show a player’s production compared to league average deployment.  For ice time, Turris’s average of 14 minutes of total ice time puts him at the level of an average fourth-line player (where the colored bars end lines up with the red/blue bars determines the level).

For even-strength scoring (primary points, which are goals and primary assists), his 1.7 point average puts him at the low-end of a first-line player in the NHL.  Quite a difference between his usage and outcome.

The power play isn’t surprising, as his 3.0 primary points put him on an average second power-play unit—which is where he was.

However, the two charts below shot shots and goals against at 5 on 5, when he is on the ice (blue box) compared to when he is off (red box).  When he is on the ice, the team gets around 7 shots more than when he is off, but they give up about ten more.  The shot differential gets lower when he is on the ice (the team goes from 1.5 shots more than the opposition to 0.5), but this shows that he helps to produce shot volume at the expense of shots against.

Goals don’t look as good—they score less and allow more when he is on the ice (this speaks a lot to the team’s goaltending as well). The differential goes from 6.5 goals more than the opponent, to 4.5 less.

Shot Quality

This graph quantifies the average quality of an unblocked shot attempt—something the Predators as a team have struggled with as of late (though they improved significantly in the last two wins!).  The orange bars represent the individual game total xG/Unblocked shot attempt and are labeled with Turris’ top linemates for that game.  The blue line represents his three game rolling average – it is the average of the current game and the two before it.  This allows us to see trends over time.

Turris’ rolling average stays above the team overall average through much of October until it finally dips below on Oct. 24th.  Over the next eight games, however, Kyle Turris would have five games where he had different linemates than the game before.  That’s no way to build consistency.  When he takes his shots, he tends to make them worthwhile, and still does a respectable job when he has been put in the line grinder.

Wins Above Replacement and Regularized Adjusted Plus-Minus (RAPM)

Without getting too deep into the theory, surely you have seen these charts that represent the components that go into Wins Above Replacement (WAR)—a metric to gauge the overall value of a player statistically—floating around.  This model is from Evolving-Hockey and you can see some issues with Turris’ game from this angle:

Offensively, he’s doing just fine—he’s slightly above the mean for goals for per 60 and expected goals per 60, and pretty good at creating volume.  Unfortunately, his expected goals against and shot volume against is pretty lousy—again speaking to his issues defensively above.  His power-play numbers surprise me a bit, but since all three metrics are shot-based (you have to shoot to score, create quality or get an attempt), this might be a bit misleading, as his value has been in setting up shots for the other forwards on the second unit.

On-Ice Statistics


Player GP TOI GF% CF% xGF% GF/60 GA/60 CF/60 CA/60 xGF/60 xGA/60 Sh% Sv%
Austin Watson 23 241.68 47.35 43.29 43.09 2.69 2.99 48.35 63.34 1.61 2.13 10.98 90.77
Calle Jarnkrok 23 282.4 54.77 52.86 52.25 3.16 2.61 57.44 51.22 2.32 2.12 9.68 90.98
Colton Sissons 23 266.22 49.41 46.07 48.58 2.66 2.72 48.52 56.8 1.77 1.87 9.45 90.44
Craig Smith 23 253.27 66.78 55.94 60.05 3.26 1.62 64.02 50.43 2.83 1.88 9.51 93.26
Dan Hamhuis 21 244.65 44.2 49.9 46.67 2.21 2.79 49.41 49.61 1.89 2.16 8.36 89.67
Daniel Carr 4 38 34.95 43.16 53.7 1.59 2.97 43.71 57.55 1.83 1.58 6.49 84.62
Dante Fabbro 23 417.7 55.77 52.28 51.04 2.41 1.91 56.56 51.63 2.01 1.93 7.85 92.37
Filip Forsberg 17 224.92 50.84 57.21 56.75 3.98 3.85 69.1 51.68 2.94 2.24 11.28 88.68
Kyle Turris 19 212.13 44.89 50.16 48.87 2.85 3.49 62.55 62.15 2.32 2.43 8.73 89.31
Mathieu Olivier 4 29.3 53.23 47.33 51.21 2.19 1.92 52.89 58.85 2.17 2.07 5.52 91.88
Matt Duchene 22 287.75 54.9 57.14 60.44 3.1 2.54 63.81 47.86 2.47 1.62 9.04 90.21
Matt Irwin 10 113.05 19.51 53.88 46.91 1.05 4.34 57.69 49.37 1.77 2.01 3.27 83.81
Mattias Ekholm 23 432.4 53.26 52.82 54.12 2.9 2.54 60.85 54.35 2.46 2.08 9.17 90.87
Miikka Salomaki 5 57.45 100 44.33 51.81 2.02 0 37.84 47.52 1.49 1.39 9.92 100
Mikael Granlund 23 317.93 47.36 54.18 56.7 2.6 2.89 64.37 54.43 2.47 1.89 7.68 90.54
Nick Bonino 23 265.18 70.41 55.15 60.64 3.75 1.57 62.12 50.52 2.75 1.79 10.66 93.73
Rocco Grimaldi 20 235.47 59.6 58.76 60.19 3.77 2.56 70.82 49.7 2.84 1.88 9.56 89.6
Roman Josi 23 408.85 61.26 54.52 57.81 3.9 2.47 68.28 56.95 2.64 1.93 10.53 92.17
Ryan Ellis 23 404.15 62.96 53.64 59.3 4.53 2.66 66.63 57.59 2.83 1.94 12.52 91.9
Ryan Johansen 23 290.5 53.08 51.21 47.11 2.88 2.54 59.58 56.77 2.2 2.47 9.8 92.27
Viktor Arvidsson 22 266.58 62.22 54.31 49.52 3.38 2.05 63.54 53.45 2.2 2.24 10.67 92.88
Yannick Weber 13 150.15 45.77 45.32 44.32 2.07 2.46 43.44 52.42 1.7 2.13 9.5 91.54

Turris grades out lower in most of these statistics, unfortunately, between 11th and 16th on the team in most of the metrics both offensively and defensively.  One caveat, however, is that the team shooting percentage when Turris is on the ice is seventh-lowest, while the save percentage is fourth-lowest—they’ve not had much puck luck when he is on the ice, and he wasn’t getting good goaltending.

However, it could be argued that those two things are a result of his mostly underwhelming numbers in the other categories, while also considering that the majority of players have had 4-5 linemates through the season while Turris has had several more sets. It’s difficult to say for sure.

There are obvious upsides and downsides to Turris as a player, and with a small sample size of games this season, it’s hard to put them into perspective considering opponents, linemates and time on ice.

In my opinion, he is an above-average offensive player who continued to create quality and volume no matter the line he was on, suffered by most metrics defensively, and has value on the second power play unit.  Statistics and analytics aren’t always the best way to evaluate a player, but they also should not be the only way they are evaluated.  You have seen the product on ice, and I have presented the numbers in the best way that I could. It is up to you to decide what you think at this point, because unfortunately, the organization has already made up their minds.