This posting started as an analysis of Tyler Bozak - and what he might be costing the Leafs at both ends of the ice in comparison to other NHL Centres of comparable value. Eventually as I looked further into it - it has morphed into a more full blown dissection of just how awful the Leafs selection of their top 6 C`s might actually be. I'm going to look at things here in 3 parts - first I'll explain my methodology in selecting my set of NHL C's for the sake of comparison. Secondly, I'll explore how the Leafs C's stack up against the competition offensively and defensively. To conclude I'm going to explore alternative options that make sense to me - and illustrate just how concerning this should be for Leafs fans.
Selection of Comparable Cs
Initially I started off looking at all forwards with over 1000 minutes of TOI in the prior 2 seasons, but then I thought that if I'm looking at Bozak specifically, it makes more sense to compare him to his peers. Then I had to decide how I narrow down on C's and specifically C's with an offensively slanted top 6 role in the NHL. To that end, I started off by totaling up the face-off numbers for all the players in the NHL over the past 2 NHL seasons and this year so far, and comparing the ratio of Power Play to Short Handed draws to weed out C's that largely take on a defensive role with limited offensive usage.
I set the ratio for the cut off at 0.75 initially filtering out any C that had 3 PP draws for every 4 SH draws. Unfortunately for some NHL clubs this limited me to one top 6 C, so in certain instances I included the next highest C for a given team. Some teams also ended up including 3 C's as they either had recently added a top 6 C, or the player had previously had that role for a prior franchise. Thus the total group of players to have played the role in the past 2+ years ended up at around 75 skaters. Unfortunately a number of rookie or second year players with limited minutes ended up on the list and their 2 year data wasn't worth including in the sample. Those players specifically include Barkov in FLA, Monahan in CGY, and Arcobello in EDM. This left me with a group of 72 NHL Centres that have played significant 1st or 2nd line minutes over the past 2 years.
The specific players I would like Leafs fans to focus on here are Bozak, Bolland, and Kadri - the 3 current Leafs that made the list. I'll be including all the data for every player I looked at for the posting though, so if you're reading this and you're NOT a Leafs fan - feel free to follow along through the methodology to see where your players end up.
Offensive and Defensive Analysis
Ok - so in looking at Bozak as a top six C, there are two key areas of concern: his offensive contribution and his defensive role. He doesn't play goal so I'll be ignoring his impacts on team SV%, but everything else is basically fair game. First his offensive contributions are up for discussion and debate.
Offensively, a player can contribute in a number of ways, but the outcomes are what most fans pay closest attention to. Goal production tends to be what resonates for most fans, but goals are a result of shots, and passes that lead to shots.
Last season Rob Vollman at Hockey Prospectus introduced the concept of "passes". Unfortunately the idea would ideally be to track passes that lead to shot attempts on net, but that data isn't logged by the NHL (and if teams are tracking it they don't share with the public). So rather than sit back and do nothing - as so many anti-stat types would have us do! - a proxy solution appears to arise out of some logic and simple math. I've modified Vollman's calculations here by using Fenwick Events rather than shots on goal.
Basically if we assume that a player's 1st Assists (passes that do lead directly to goals) are distributed fairly evenly over the long term (i.e. he'll be credited for a fairly equivalent proportion of passes that do score goals as he is for passes that don't) we can generally assume this ratio holds for most players.
Obviously in some instances this could be an underestimate (i.e. Crosby likely has a higher proportion of his passes end up as goals than a player like oh - Trevor Smith?), but this likelihood is compensated for in the other aspect of the calculation - which is the players' On-Ice SH%. That is to say the percentage of shots that actually end up in the net.
For this calculation I'm actually going to use Teammate Fenwick SH% - which is basically the On-Ice SH% of all Fenwick Events (shots on goal and missed shots - but no blocked shots) by a player's team-mates when he is on the ice (we do this so the player isn't penalized because of the shots they're taking that go in for goals). If we take the players' 1st Assist rate per 60 and divide it by the players' Teammate Fenwick SH% we should get a decent approximation of the Fenwick Attempts that resulted from passes by the player. That is to say, the number of unblocked shot attempts the player set up or produced the final pass for. Obviously there is SOME randomness in this calculation's results, but the idea is that over the long term that randomness would wash out.
To this we can add the individual Fenwick Events the player is responsible for on their own, i.e. the Fenwick Events that they produce per 60 minutes. If we combine these two values - iFenwick Events and Fenwick Passes - we get an idea of how much offense the player is actually responsible for while they're on the ice. This ignores the SH% aspect in the sense that it should just give us a proxy for the events occurring as a direct result of the efforts of the player.
Here is the listing of my set of 72 NHL top 6 C's - ranked by their Total Fenwick Event production (iFenwick + "passes") rate per 60 minutes of 5v5 ice time over the previous two seasons (2011-13). Their percentile score is also shown - this is the percentage of C's in the data set that have fewer passes than the player in question.
|Player||Team||TOI||iFen||FirstA/60||iFen/60||Fen Pass/60||iFen Event|
Ok - so a couple of things I'd like to point out - Bolland and Bozak are 70th and 71st respectively on this list. Kadri is 7th, Grabovski was 33rd. Carlyle - until he was forced to by Bozak's injury - has been playing Bozak as his top line C. He played him ahead of Grabvski, he's playing him ahead of Kadri, he's even played him ahead of Bolland who looks like Chicago's own version of Bozak frankly.
Just an aside on this point - maybe this helps explain WHY Chicago had Bolland as their 3rd C? Just a suggestion that maybe the Stanley Cup champions, who have won multiple times in recent years, and do well according to fancy stats, know more about his value than the Leafs? But I digress.
So - this discussion inevitably leads to a discussion of the fact that Bozak plays with the most productive wingers in the NHL (Kessel/Lupul/JvR) and he deserves credit for that fact. In reality he participates in the lowest percentage of Fenwick Events of ANY of the C's on this list (32% of those that occur with him on the ice). Nazem Kadri - oh so ironically - leads this list of NHL players by participating in a whopping 69% of the Fenwick events that occur while he is on the ice.
It should be noted that much of Kadri's ice time in recent years has occurred alongside Clarke MacArthur, Nikolai Kulemin, Joffrey Lupul, and Matt Frattin - Who range from the low end for top 6 wingers in Fenwick Event production to the very high end (Lupul). His top 2 wingers (MacArthur and Kulemin) combine to rank 46th for the top 2 line mates in iCorsi for this group of C`s (33rd percentile). This hasn't seemed to cause a drastic shift in Kadri's output on the passes front - where he ranks 7th.
Bozak ranks 67th in passes and Bolland ranks 71st - but Bozak's top 2 line mates ranked 5th in combined iCorsi while Bolland`s ranked 29th. Bolland has a bit of an excuse on that front (not much of one) while Bozak just sucks.
All evidence would indicate that offensively speaking Bozak is a total passenger on his line, who happens to benefit occasionally from being present when goals are scored.
So maybe his value lies in the defensive end? Unfortunately all evidence would contradict that as well. Using the regression technique I have used to compare a player's Expected Corsi to their actual Corsi (based on situational usage) dCorsi, Bozak is actually the 678th worst NHL forward - of the 712 to play 200+ minutes at 5v5 from 2007 through last season - defensively. (Bolland is a slightly better 600th in the same ranking)
Bozak's dCorsiA (Delta Corsi Against) is -1.922, which implies he is dragging down his linemates significantly. Bolland posted a -1.122 in Chicago over the last 6 years. For the sake of comparison Phil Kessel is only a -0.682, which is well within 1 standard deviation (1.175) of the mean value of 0. Bozak is 1.636 standard deviations below the mean - he is significantly below average defensively. This is NOT a reason to play him on a top line.
Nazem Kadri ranks 270th defensively with a dCorsiA of +0.301 (he helps his linemates reduce Corsi events against). He is vastly superior to Bozak from a defensive possession standpoint.
I think there's really only one obvious option here - and that's to make Kadri the top line C for the future... and the future is now.