Who are the best two-way centers in NHL over past 5 seasons?

Maybe the most important position in hockey is center, certainly more important than wing and probably more important than defense or goaltending. Even more important is having a real good two-way center capable of playing big minutes at both ends of the rink. Think about the recent Stanley Cup winners. In Chicago you have Jonathan Toews. In Los Angeles you have Anze Kopitar. In Boston you have Patrice Bergeron. Going further back you have Datsyuk in Detroit. These four guys are maybe the best two-way centermen in the league but I wanted to take a more analytical approach to answering that question.

Let me first come out and say that evaluating defensive play is something hockey analytics is still pretty poor at doing and I have said several times that the toughest question to answer in hockey analytics right now is how to separate the defender from the goalie. It is just a really difficult problem with the data we have access to today. It may not just be hockey analytics that struggles with answering this question. I have even questioned whether NHL coaches are capable of identifying their best defensive players. With that said, I am going to try and answer the question anyway.

In order to not miss any important players earlier in the week I asked my twitter followers to nominate two-way centers for which I should include in my analysis. I had lots of good responses, and even a few that I hadn’t thought of. A few people even suggested Sidney Crosby and Henrik Sedin both of whom I consider more offensive oriented players but I figured I’d throw them in the study as well. A few people even suggested Tyler Bozak which I can only assume was in jest but I am going to include him for fun, and as a test for my metrics. By testing metrics when evaluating individual players any metric must rate Crosby as the best offensive player in the NHL (no exceptions, he is by far the best) and must not rate Bozak very highly overall (because he just isn’t very good). Failure to meet these two standards indicated the metric in question is flawed. In selecting centers for this study I also wanted to select players with a long track record to make for easier, more reliable evaluation. For this reason I restricted myself to players with at least a 5-year track record and mostly looked at 5-year stats in my study.

With all that said, here are the list of 25 centers I have chosen to look at. I apologize if I missed anyone but I think I got the most important two-way centers as well as a few extras.

ANZE KOPITAR JORDAN STAAL PAVEL DATSYUK
BRYAN LITTLE KYLE TURRIS RYAN GETZLAF
DAVID BACKES LOGAN COUTURE RYAN KESLER
DEREK STEPAN MIKAEL BACKLUND RYAN O_REILLY
ERIC STAAL MIKHAIL GRABOVSKI SIDNEY CROSBY
HENRIK SEDIN MIKKO KOIVU TOMAS PLEKANEC
JOE PAVELSKI NICKLAS BACKSTROM TRAVIS ZAJAC
JOE THORNTON PATRICE BERGERON TYLER BOZAK
JONATHAN TOEWS

RelTM Statistics

The primary metric(s) I am using for this study are my RelTM statistics which can be found at puckalytics.com or stats.hockeyanalysis.com. RelTM statistics are like a combined WOWY and is an indication of the effect the player has on his teammates – i.e. whether they play better or worse with him than apart from him. I won’t go into the details but I feel these are a better reflection of the player than the typical ‘Rel’ statistics which is an On-ice minus Off-ice comparision. Also, with 5 years of data we should be getting fairly stable statistics so I think we should find some value here.

Let me first look at Corsi RelTM statistics in 5v5close (last 5 seasons) situations to minimize score effects.

Two-Way_Centers_CF60RelTM

The good news is Bozak rates poorly but the bad news is Crosby isn’t the top offensive center when it comes to CF60 RelTM. That is a knock against this statistic and one of the reasons why I am not a fan of Corsi, particularly when evaluating individual players offensive performance.

Two-Way_Centers_CA60RelTM

Here negative is better as it means fewer shots against. There is probably more basis for using Corsi to evaluate defensive play but I am still not sold on it, especially when dealing with elite level players. There are some guys, such as Logan Couture, David Backes and Derek Stepan, that I would certainly rate higher in terms of defensive play. CA60 RelTM did pass the Tyler Bozak test though.

Two-Way_Centers_CFPctRelTM

CF% RelTM ranks Patrice Bergeron as the best two-way center in the league which on the surface clearly passes the smell test. It also has Tyler Bozak easily the worst in this group so it passes the Bozak test as well. Crosby being middle of the pack doesn’t though, simply because offensively he is so far superior to everyone else that should overcome any defensive liabilities he may have. There is no way Backlund or Grabovski is a better two-way player than Crosby.

Two-Way_Centers_GF60RelTM

This is why I prefer using long-term goal based statistics – Crosby is far and away better than anyone else which is the way it should be. Bozak isn’t last but is well down the list as well which is good though I am surprised Plekanec has such a negative impact on his teammates offensively.

Two-Way_Centers_GA60RelTM

Tyler Bozak is terrible defensively. Check. David Backes is the best defensive foward in this group. Maybe a bit surprising but I can accept it (the Blues have been a solid defensive team for a number of years now). Sidney Crosby and Nicklas Backstrom being middle of the pack as more offensive oriented players, makes sense. Maybe a bit surprised Bergeron isn’t higher but for the most part I think this all passes the smell test.

Two-Way_Centers_GFPctRelTM

Who is the best two-way forward? None other than David Backes according to GF% RelTM. The good news is Crosby ranks much higher here behind only Backes, Kopitar and Sedin and just ahead of Toews and Bozak ranks last so we pass the Bozak test.

The last statistic I want to look at is a combination of GF60 RelTM and CA60RelTM. When I combined these two I came up with this:

Two-Way_Centers_GF60CA60RelTM

Now this is getting interesting. Would anyone argue with the best centers in the league over the past 5 seasons being Crosby-Datsyuk-Toews-Kopitar-Sedin-Bergeron-Thornton? I haven’t looked at GF and CA statistics combined very often but I think it may be a worth while statistic to pay more attention to when evaluating a players overall performance.

The final thing I want to look at with these 25 players is defensive usage. Here I am going to look at two statistics: 1. The percentage of a teams 4v5 PK ice time that the player gets assigned (when healthy and in the line up) and 2. The percentage of a teams 5v5 when leading ice time (when healthy and in the line up).

Two-Way_Centers_4v5TOIPct

This is why I don’t consider Crosby or Sedin defensive players – they just aren’t given any ice time killing penalties. I am surprised Datsyuk wasn’t relied on more on the PK.

Two-Way_Centers_5v5LeadingTOIPct

Ok, I am shocked that Bergeron has the smallest percentage of his teams 5v5 ice time when they are protecting a lead and kind of surprised Getzlaf and Crosby are getting the most. I guess the Penguins and Ducks have been a ‘go for the kill’ team rather than a ‘protect a lead’ team.

That is a lot of data for you to digest and I am not sure I really answered the question of who is the best two-way center in the NHL. If you forced me to give you an answer I would probably eliminate guys like Crosby and Sedin who are primarily offensive players and say the best true two-way center in the NHL the past five seasons has been Toews (over Datsyuk because he gets so much more PK ice time). Datsyuk, Kopitar and Bergeron are pretty close behind though. So, I guess in the end we didn’t learn anything we didn’t already know.

 

This article has 22 Comments

  1. Two reasons why Datsyuk hasn’t seen more time on the PK: his age, and the fact that many if not most of the Wings’ forwards can kill penalties with aplomb (it’s sort of a job requirement, I guess). Age is the big thing, I think. Babcock wants to limit Datsyuk’s wear and tear. To that end, he essentially didn’t use Datsyuk (or Zetterberg) on the PK this past season.

  2. Speaking of Zetterberg, I’d have to think he rates as one of the game’s better two-way centers. Though, he’s definitely spent some time on the wing, and when he does move to the wing, his centerman is always Datsyuk…so I guess there wouldn’t be enough data compared to someone who always lines up in the middle.

  3. “By testing metrics when evaluating individual players any metric must rate Crosby as the best offensive player in the NHL (no exceptions, he is by far the best) and must not rate Bozak very highly overall (because he just isn’t very good). Failure to meet these two standards indicated the metric in question is flawed.”

    So essentially this entire analysis was just an exercise in confirming your pre-conceived opinions?

    The entire purpose of these “advanced” statistics is to challenge subjective analysis like “Crosby is the best, duh” and “Bozak is the worst, duh”. If the statistics fail to reinforce that, the possibility exists that your subjective assumptions are wrong as opposed to the statistics being “flawed” (although, to be fair, no statistic can ever be perfect). A more thoughtful review would have explored why statistical analysis conflicted with “the eye test” (if or when it did).

    1. When conducting research/analysis you have to constantly come back and ask yourself, does this make sense? This becomes more and more important the more abstract the analysis becomes. Otherwise you end up with conclusions like Tyler Kennedy is the 3rd best player in the NHL. Now maybe Kennedy was the third best player in hockey but the burden of proof to overturn everything we know about hockey and the opinion of every player, coach, GM, scout and fan in the game is extremely high. The proper response is not to justify that Kennedy is the third best (or even among the best) players in the game but to ask where and why did my analysis go wrong.

      There is a ton of evidence, objective and subjective, that points to Crosby being easily the best offensive player in the game. I actually linked to one early in the article. There is also a significant amount of evidence that Bozak is a poor hockey player, especially relative to the other players in this article. For the purposes of this article using Crosby’s offense and Bozak’s overall play are two useful players to run a reality check on the metrics being used allowing us to gain more confidence in the metrics being used. If metrics can’t properly identify Crosby as being good offensively or Bozak being not in the same league as the majority of the other 24 players we can have zero confidence in a metric being able to determine whether Kopitar is better than Bergeron or Datsyuk.

  4. Why is Malkin not in this article? Malkin is one of the best two way centers in the league. Might even be the best player based on these stats because he is better than Crosby as a complete player.

    1. Do you have a reasonable source for “Malkin is better than Crosby as a complete player”, or is that just a personal hunch?

      1. Well, he’s got a positive CF% & FF% all while having a higher percentage of D-Zone starts over the last 5 years. Or the simple fact that he’d be a first line center on 25+ NHL teams should probably get him inclusion in the field, right?

        I’m not ready to call him a more complete player than Crosby, but that’s what you’re here to tell us.

  5. “This is why I don’t consider Crosby or Sedin defensive players – they just aren’t given any ice time killing penalties.”

    How ice time is distributed depends entirely on the coach; if Hitchcock thought that David Backes was best served by playing mostly offensive minutes, and he got very little PK time, would you similarly consider Backes “not a defensive player”?

  6. This is interesting; thanks for sharing the data and giving some background explanation. For what it’s worth, I think PK time is much less relevant than you do for evaluating “2-Way” Centers. Maybe a more concrete definition could be developed for 2waycenter, because I always found it an odd term (odd in that there is no consensus on what it means). Similar to Selke trophy winner. What does that even measure, exactly? As a Penguin fan I obviously have particular biases, but I would be interested in arguments as to why Crosby probably should or should not win the Selke most years too.
    The logic behind giving your best forwards PK minutes is interesting to me. On the one hand, if the score is close or you have a lead, it makes sense to put your best men out on the PK. If you’re trailing, it makes less sense to me. Or maybe more broadly, it comes down to coaching decisions and team style, I think. “Defensive-minded” teams might generally invest resources in preventing their opponent from scoring, rather than invest resources in generating their own goals (STL, LAK, CHI, BOS). A preference for goal prevention over offensive production on a systems level is going to accentuate skilled forwards who excel at the “defensive component” of the game. And clearly there are coaches who do this and coaches who don’t.

    I’m kind of saying give Crosby all the Selkes, but I’m also being sincere.

    Last thing, I have sometimes seen people talk about zone starts and zone finishes to evaluate defensive forwards (i.e., what % of player A’s shifts began in the defensive zone compared to the offensive zone, and what percentage of the same player’s shifts ended in the offensive zone as compared to the defensive zone? A player like Jordan Staal (at least I know for sure when he was a Penguin) starts most of his shifts in the defensive zone and ends most of his shifts in the offensive zone. He was literally going on the ice when the puck was in Penguins territory, retrieving it, forcing it up the ice, and either generating offense directly or at least drawing an O zone faceoff for Crosby or Malkin. For me, that is more important than killing penalties. Bergeron is ridiculously good at this too. Others, etc. Compare with Sedins, who start an overwhelming % of their shifts in the offensive zone (at least under Vigneault, who decided it made the most sense to put his best players in position to score and to spend most of their minutes in the offensive zone). But forwards who are relatively terrible at defense or relatively terrible at hockey in general start a lot of shifts in the offensive zone (i.e. get “sheltered” minutes) and end a lot of shifts in the defensive zone, lol. If you put Johnny Sandpaper out for an offensive zone draw and 40 seconds later he’s leaving the ice from his own end, and your best player is sent out for the defensive zone draw, you just wasted half or more of your best player’s shift / chance at generating offense because he has to go all the way to his own end, win the puck, and carry it out of there to attack. That best player is not going to generate as many goals / assists because of the rules of space and time. But that’s stereotypically what 3rd line centers are for: they go get it and bring it to the attacking zone, so the 1st and 2nd centers can bring their units on in the O zone as often as possible.

  7. How do you weight the TM stat? For example TMGF60. I’m wondering how the stat behaves as lines become more or less stable.

    1. The TM stats are a weighted average all teammates ‘apart’ stats (from the WOWY pages) weighted by TOI with the player.

      For goals against all players are included in the weighted average (including goalies) but for all other stats (goals for and fenwick/corsi) only forwards and defensemen are included.

      The exception is for players that have fewer than 100 ‘apart’ minutes. These situations would cause a small sample size to be given the largest weight in the average so for these players I use the players overall (with and apart) stats. Only a few pairs are affected by this once you get to a seasons worth of data (guys like Daniel and Henrik Sedin and some defense pairings).

      For teams that shuffle lines a lot you will get the best results as the more mixing of players means you are better able to isolate individuals from each other. Unfortunately it will always be difficult to isolate Daniel from Henrik Sedin.

      1. Hmm. I don’t think you can use this stat to compare players across teams.

        To begin, some notation: TOI(A,B) is the time on ice player A spends with player B, and TOI(A,#B) is the time on ice player A spends without B. With this convention, TOI(A,A) is player A’s time on ice. Similarly, GF(A,B) is the goals for stat for player A when player B is not on the ice. Finally, SUM(1,n)[f(k)]=f(1)+…+f(k).

        We will be looking at the stats of player 0, whose team is made of players 1,2,…,n.

        Let’s examine the GFRelTM. I’m not bothering about the ‘per 60’ aspect, as that is just a coefficient that factors out.

        We have that the GFRelTM of player x is:

        [GF(x,x)/TOI(x,x)] – SUM(1,n)[W(k)*GF(k,#x)/TOI(k,#x)]

        Where W(k) is the weighting factor of player k with respect to player x. Now, I suspect that:

        W(k) = TOI(k,x)/TOI(x,x)

        Let’s contrast two scenarios. Each scenario will consist of Crosby, a second line centre, and four identical plugs. In these examples, hockey is a 3v3 game.

        Crosby and the other player never play with each other (so they always play with two plugs). Every player is on the ice for 10 minutes. Crosby always scores 1 goal per minute. The plugs never score.

        First example: The second line centre is Malkin, who scores at a rate of .8 goals/minute. In this case the GFRelTM of Crosby and Malkin respectively are -0.6 and -1.2.

        Second example: The second line centre is Bozak who scores at a rate of .2 goals per minute. In this case the GFRelTM of Crosby and Bozak are 0.6 and -1.8.

        To contrast, Scott and Glass never play together, have 4 identical plugs who never score, and both of those guys score at a rate of 0.01 goals/minute. Their GFRelTM are both -0.01.

        Either I’m misunderstanding these RelTM stats, or they are very inherently flawed.

        1. Yes, you can construct scenarios where the statistic is going to fail. It is not a perfect stat.

          Just a bit of a background here. Traditional ‘Rel’ statistics perform a similar relative comparison but instead of at the player level as I do it is done at the team level. So the Rel statistic is On Ice Stats – Off Ice Stats. The problem I had from these Rel statistics arose many years ago when comparing Red Wing and Islander players. The Red Wings were one of the best teams in the league and the Islanders one of the worst. In these Rel statistics Kris Draper looked terrible (worst in the league) and yet he was selected for the Olympic team. The reason is he was playing with Maltby while he never got to play on the Red Wings top two lines which were stacked with Yzerman, Datsyuk, Zetterberg, Shanahan, Holmstrom, etc. Even good players would look terrible against those guys. Meanwhile a lesser 3rd line player on the weak Islander team would look far better because he didn’t have to compare with a stacked top two lines. For that matter weak teams often shuffle their line up a lot and that 3rd line player probably got ice time in the top six. Draper never did. I wanted to come up with a better system where a players on-ice statistics are not being compared to the teams performance when they aren’t on the ice and using players he rarely gets to play with.

          There is where I came up with the idea of looking at whether the players he does play with perform better or worse with him than apart from him. As you point out this is not a perfect system either but I think it is better and I know some others have used it in models and found it provides better results. There is still a bit of a problem with players on strong teams getting penalized because they can’t make a superstar look better while if they were on a weak team they may be able to make a mediocre player look better but I think it is less of a problem using RelTM than traditional Rel statistics. Furthermore defensemen are inserted into the equation and defensemen play with everyone on the team, from first liners to fourth liners. The defensemen on the team will come closer to representing team average statistics so they act as a bit of a neutralizing effect to minimize some of the problems you bring up. So, in reality your specialized scenario never really exists to full effect. Also, increasing to longer time periods provides for more mixing and matching of players, including movement between teams, which will again normalize things.

          How to best isolate an individual players talent and how to normalize that across the league so we can best compare players is a really tough question that we probably can’t solve perfectly with the data we have. We have to do the best we an and I believe (though I may be biased) RelTM is currently the best method to do so.

          To finish up, yes, players on extremely good teams or playing primarily with extremely good players may be getting penalized a bit for doing so. Similarly players playing on an extremely poor team playing primarily with extremely poor players probably are getting over valued. There is a lot of parity in the league right now though so this minimizes the necessary corrections some what.

          1. The effect I was pointing out was how player A can never see ice time with player B, and still have player B significantly impact player A’s RelTM statistic. Defenseman wont average this out; my example uses ‘averaged out’ players (which makes the proportion of ice time they see with A and B irrelevant) and adding more players simply exacerbates the effect.

            The point being that this stat can’t be looked at in isolation. RelTM stats are dependent on how players on the same team who never see ice time together relatively perform.

            I would bet a fair bit that this is why Crosby’s numbers are dragged down in some of these stats.

          2. Yes, Malkin will impact Crosby’s Rel statistic. But so will Sutter, and so will the fourth line center. Some will make Crosby look better. Some will make him look worse. Over the long haul and across all teammates a story starts to get told. Good players rise to the top.

            It isn’t perfect (we will never have the perfect stat) and there may need to be more normalization across teams (this is an extremely difficult thing to do, if even possible) but I think the stat is telling us something useful. More useful than the alternatives.

          3. Couldn’t reply to your last post. Hence this. Apologies.

            Yes Malkin and Sutter both have an impact on Crosby’s numbers, as will the fourth line centre. However, the fourth line centre will only have a significant effect on Crosby’s numbers if the rest of the fourth line sees decent ice time with Crosby. Which doesn’t happen.

            I don’t think it is a coincidence that Toews has great RelTM stats with a coach who likes to juggle lines. I don’t think it is a coincidence that Crosby has surprisingly low RelTM stats when he is consistently partnered with the same wingers (who will realistically see the second most ice time with Malkin).

            These conclusions arise almost immediately from examining the formulas used. The RelTM stats need some sort of accompanied perturbation analysis to be taken seriously.

  8. Nice article. Seems like Claude Giroux should have been included. He spends a decent amount of time on the PK, and has good offensive stats.

Comments are closed.