Coincidence vs. Causation, Tyler Bozak, And Defensive Play

David Bolland gets injured after the Leafs hot 14 game string to start the year and everyone is certain his loss is the reason the team flounders. Tyler Bozak misses games which sees Nazem Kadri promoted into the top line centre role. Suddenly Kessel and JvR struggle to score and the conclusion is that it must be because they miss Bozak and that's enough to apparently make Kadri trade bait. The Leafs trade for Tim Gleason (shutdown D extra-ordinaire!) then starts winning again after only managing 4 regulation wins in their previous 32 games played. Conclusion: obviously the team's defense is better because of Gleason's arrival!

What do all of these arguments have in common? Well they indicate a strong tendency on the part of many Leafs fans and observers to assert causation when really all we have is correlation. Player X is present when things go well and absent when things go poorly therefore we assume Player X caused things to go well. This fallacy seems to be the main thrust of this not so great piece of analytical work in the Toronto Star.

Just because Player X is present when wins are occurring doesn't mean Player X is causing the wins. This same logical fallacy arises regularly in almost every field of life and tends to be why we as humans carry around ridiculous superstitions, believe in pseudo-science, and lots of other fairly fanciful things. Unfortunately its ubiquity is what promotes its continuing presence in society. Everyone thinks it's true so it must be true. It's almost too cute in the circular nature of the logic.

Let's look at some basic numbers from prior to this season:

Now many of you will have seen this data before, but here's how things have gone this season:

Now a couple of things should stand out to you here. Firstly, the general nonsense of this statement by Kevin McGran in the aforementioned article:

"He’s not really a point producer, although he’s been fine on that front lately. It’s more his hockey sense that shines. He tends to be in the right place at the right time. He doesn’t mind backchecking. He’s quite good in his own zone."

- Kevin McGran, Tues Feb 4th

Here's why it's absurd: If you look solely at point production and goal scoring (ignoring all the stuff about shot generation, corsi, etc.) - which is basically what McGran does in that piece - Phil Kessel's goals against is HIGHER with Tyler Bozak on the ice. Secondarily, Bozak's goals against is even higher without Phil Kessel on the ice. How the hell the guy who is slated as a "defensively responsible backchecker" comes out looking way worse without the team's top scorer on the ice defensively is beyond me.

By far the best defensive numbers for Kessel happen when he isn't on the ice with Bozak and that's true for the last 7 years. This isn't new information as anyone that followed the Great Grabovski Schism this summer knows. It's always been true and it's a major fault of Tyler Bozak's. When I wrote this piece back near the end of October, I pointed out the following:

"Bozak is actually the 678th worst NHL forward - of the 712 to play 200+ minutes at 5v5 from 2007 through last season - defensively."

- Steve Burtch, Oct 29th 2013

Somehow people have convinced themselves Tyler Bozak's value is as a defensive centre on the Leafs skilled top trio. Nobody in their right mind would call Bozak one of the most talented offensive Cs in the NHL, so he must have value defensively. Unfortunately, there's a lot of evidence that shows just the opposite. Of the 505 Skaters to play 400+ minutes at 5v5 so far this season Tyler Bozak ranks 472nd in Goals Against per 20 minutes of TOI. That's the 7th percentile - in other words only 7% of NHL Skaters are worse than him defensively... or put differently, 93% of guys playing a regular shift in the NHL are seeing the puck go in the net less frequently.

Here's the scary part - he ranks 367th in SV% while he's on the ice at 5v5. That means he's sitting at the 27th percentile in terms of goaltending when he is on the ice. While that's still below average, his GA20 ranking is far worse than his SV% ranking so you can't really blame the goaltending for how crappy his results are. So what's the real problem here? Well - it's shot attempts allowed. You see, Tyler Bozak, Phil Kessel, and James Van Riemsdyk rank 503rd, 504th, and 505th on that same list in terms of Fenwick Events Against per 20 minutes while they're on the ice at 5v5.

They are the worst players in the NHL in terms of allowing shot attempts / scoring chances. They are, by this measure, literally the worst line in hockey defensively. So if Tyler Bozak's main contribution to the trio's play is his defense then it's not showing in these results.

But how do we know he is the part that's weak defensively? Well look at this chart for him and Kessel over the prior 6 seasons regarding Corsi (all shot attempts including those that are blocked):

Obviously prior to this year Kessel's shot attempts for and against were better with Bozak than they were WITH Bozak. Then here are the numbers for just this season:

So this year his defensive and offensive shot attempt numbers both improve marginally with Bozak and get worse marginally without him. But if Bozak skates without Kessel his offensive numbers whither to virtually nothing and his defensive results become abysmal.

The logic that somehow Bozak is the cause of all this escapes me and yet the idea that he's "making" Phil Kessel and James Van Riemsdyk superior this season persists. If you look at his offensive contributions, he's generally seeing much better results this year than in previous years. But the results hinge on a highly variable factor: Tyler Bozak's On-Ice SH% at 5v5 this year is 12.31%. The last time we played a full season of NHL hockey (2011-12) only 3 skaters who played over 500 minutes of 5v5 TOI saw an on-ice SH% that high:

Jordan Eberle - 12.72%, Steven Stamkos - 12.50%, and Martin St. Louis - 12.31%

Nobody posted a number that good in 2010-11, only 4 skaters did in 2010-11, 5 did it in 2008-09, and 2 did it in 2007-08. Nobody has done it two years in a row. That's 14 skaters in 7 years and none of them have managed the feat twice. That's a good indicator that the likelihood of this offensive outburst representing a new normal is quite slim.

So realistically do we think Tyler Bozak is making the Leafs top line better? He is participating more and he's not as behind the play offensively as he used to be. He's still atrocious defensively and doesn't contribute in that area nearly as much as people like to think. He's also riding the crest of an insane wave of shooting luck while he's on the ice. It's a wave that isn't likely to last and nobody should expect it to continue long term. He may end up finishing out the year with such numbers maintaining but he still isn't a top line C in the NHL. A simple look at a list of NHL centres would turn up at least 30 superior to him.

Unfortunately a lot of bad analysis will lead people to the conclusion that he's "proven" to be a solution in the middle on the Leafs top line, when most of the underlying numbers still indicate he's the source of many of the team's problems. I'm still nowhere near convinced Bozak is a top line C, and a career year thanks to crazy shooting luck for the team's top wingers has yet to convince me. On the balance of probabilities, you shouldn't be either.