|
|
Post by Boston_Habs on Oct 2, 2017 20:00:01 GMT -5
Andrew Berkshire does great work bringing the intricacies of the data analytics movement in hockey to the masses. Today he ranked the top 23 dmen in the league based on the following criteria: Offense: even strength and PP goals, primary assists, secondary assists, scoring chances, high-danger scoring chances, scoring chances generated per teammate, shot attempts, slot passes completed, penalties drawn, relative on-ice goals per 60 minutes, and offensive zone pass completion rate. Transition play: 5v5 outlet passes, stretch passes, controlled carries out of the defensive zone, neutral zone passes, controlled entries into the offensive zone, Corsi and Corsi relative to teammates, pass completion rates. Defense: 5v5 and shorthanded loose-puck recoveries, pass blocks, body checks, penalties taken, successful dump-outs per 60 min, on-ice goals against relative to teammates, on-ice shot attempts against relative to teammates, turnover rates relative to teammates by zone. All of these incidents or data points can be observed, collected, and analyzed and then squared with the eye test and actual production to see who makes the biggest impact at the position. It's probably the biggest change in how the game is viewed in the modern era, providing coaches and GM's with new insight into what makes a player special beyond just points and +/-. It's this kind of analysis that got the Habs data analytics guy fired after begging Marc Bergevin not to trade Subban for Weber. So FWIW: 1. Erik Karlsson 2. PK Subban 3. Drew Doughty 4. Victor Hedman 5. Mark Giordano 6. Duncan Keith 7. Brent Burns 8. TJ Brodie 9. Cam Fowler 10. Dustin Byfuglien 11. Alex Pietrangelo 12. Roman Josi 13. Torey Krug 14. Seth Jones 15. Justin Faulk 16. John Klingberg 17. Kris Letang 18. Jared Spurgeon 19. Jacob Slavin 20. Colton Parayko 21. Oliver Ekman-Larsson 22. Ryan Suter 23. Jeff Petry Notice that Shea Weber is nowhere to be found on that list - he just doesn't drive enough offense or transitional play to be a complete player. And while I wouldn't trade Shea Weber for Torey Krug, I certainly wouldn't have traded (arguably) the 2nd best all around dman in the league for Weber. www.sportsnet.ca/hockey/nhl/definitive-ranking-nhls-top-23-defencemen-three-seasons/ |
|
|
|
Post by frozone on Oct 2, 2017 22:41:30 GMT -5
The even bigger takeaway for me is Jeff Petry's ranking. Have we been heavily underrating him?
|
|
|
|
Post by blny on Oct 3, 2017 7:33:12 GMT -5
I think it was TSN (?) that released a top defenders list recently. Much more of a popularity contest. Had Subban way off the radar.
|
|
|
|
Post by Boston_Habs on Oct 3, 2017 8:06:12 GMT -5
The even bigger takeaway for me is Jeff Petry's ranking. Have we been heavily underrating him? Possibly, although he has benefited from more sheltered minutes compared to bona fide #1 guys. It will be interesting to see how Petry does this year with Markov gone. He's now our best puck-mover at the position and will be counted on more heavily this year. |
|
|
|
Post by BadCompany on Oct 3, 2017 8:47:23 GMT -5
I like Jeff Petry, but any ranking that has him ahead of Shea Weber is missing something in its methodology. For that matter, any ranking that has Jeff Petry in the top 25 is probably missing something in it's methodology. By this ranking Petry is a #1 defenseman. Does anybody really believe that? And if it's true, what are we worried about? Petry is our #1, and the "average" Shea Weber can be our #2. Sounds like a pretty good blueline to me. The NHL Network had Weber 9th overall (Subban, because we have to go there, is 6th): link |
|
|
|
Post by Boston_Habs on Oct 3, 2017 9:32:24 GMT -5
I like Jeff Petry, but any ranking that has him ahead of Shea Weber is missing something in its methodology. I don't think it's the methodology. The issue is the weighting of the factors (offense, defense, transition) and your mileage will vary depending on how much weight you assign to them. Berkshire arrives at a "score" for each metric and adds them up, which is another way of saying it isn't weighted to any particular factor. I think that becomes a personal issue or something for a a GM to decide. What matters most? But this is part of what the analytics crowd is trying to do. Conventional wisdom says of course Shea Weber is better than Jeff Petry, with that booming shot, big size, and shut down ability in the D zone. And in the dying seconds of a 2-1 game, of course I want Shea Weber on the ice over Jeff Petry, Torey Krug, and some other guys on that list. That's because, at least in those situations, we value "defense first" and goal prevention in a dman as the most important quality. But what about the rest of the game? To me it's about the net contribution. For example, the stats would say that Andrei Markov was a better dman than Shea Weber last year for the simple reason that although Weber's goal prevention/shot suppression numbers were better than Markov's (but not that much), they were more than offset by Markov's superior ability as a puck mover, passer, offense and shot facilitator. Markov did more, on a NET basis, to facilitate plays than Weber, so did Petry, but for some people it just comes down to having that nasty guy at the back end in crunch time. Or the leader in the room. Weber has those intangibles, but again, how do you value them? The NHL list seems like the conventional media/fan wisdom, which isn't to say it's totally wrong. The top of that list is close to the analytics version, but after that it gets iffy. Ryan Suter isn't a top 10 dman anymore, Zach Werenski at #13 seems like a stretch, as is Dougie Hamilton at #16. |
|
|
|
Post by BadCompany on Oct 3, 2017 10:39:52 GMT -5
I like Jeff Petry, but any ranking that has him ahead of Shea Weber is missing something in its methodology. I don't think it's the methodology. The issue is the weighting of the factors (offense, defense, transition) and your mileage will vary depending on how much weight you assign to them. Berkshire arrives at a "score" for each metric and adds them up, which is another way of saying it isn't weighted to any particular factor. I think that becomes a personal issue or something for a a GM to decide. What matters most? But this is part of what the analytics crowd is trying to do. Conventional wisdom says of course Shea Weber is better than Jeff Petry, with that booming shot, big size, and shut down ability in the D zone. And in the dying seconds of a 2-1 game, of course I want Shea Weber on the ice over Jeff Petry, Torey Krug, and some other guys on that list. That's because, at least in those situations, we value "defense first" and goal prevention in a dman as the most important quality. But what about the rest of the game? To me it's about the net contribution. For example, the stats would say that Andrei Markov was a better dman than Shea Weber last year for the simple reason that although Weber's goal prevention/shot suppression numbers were better than Markov's (but not that much), they were more than offset by Markov's superior ability as a puck mover, passer, offense and shot facilitator. Markov did more, on a NET basis, to facilitate plays than Weber, so did Petry, but for some people it just comes down to having that nasty guy at the back end in crunch time. Or the leader in the room. Weber has those intangibles, but again, how do you value them? The NHL list seems like the conventional media/fan wisdom, which isn't to say it's totally wrong. The top of that list is close to the analytics version, but after that it gets iffy. Ryan Suter isn't a top 10 dman anymore, Zach Werenski at #13 seems like a stretch, as is Dougie Hamilton at #16. Sure, I get that, and I appreciate what the analytics crowd is trying to do. It can be very useful. But I still think that they’re not quite there yet. By according equal weight to all numbers, without taking into consideration the intangibles, they come up with a methodology that, in my humble opinion, has a problem. Yes, Markov’s numbers look better than Weber’s in many areas… but Markov played over three minutes less per game than Weber did last year. We’ve seen what happens to Markov when he plays too much; his play deteriorates. While Markov - or Petry - may have the better transition numbers, at the end of the day I don’t want either of those guys on the ice for 25+ minutes per game. But I'm okay with Weber doing that, as there doesn't appear to be any noticeable change in his play from 21 to 25 minutes per game. Heck, Ales Hemsky has a very good points-per-60 minutes ratio, which taken to its logical extreme means he should play all 60 minutes of a game. Of course that’s silly, and I don’t mean to denigrate your point. It’s just that by not putting any weight on any one factor they are in a sense applying their own subjectivity to the matter, and by ignoring the intangibles they ignore a major part of player analysis. Of course you can’t measure the intangibles, but I still believe they play a part in the overall assessment. At least when it comes to making rankings anyways. Any ranking system is going to be subjective, no matter how much it claims not to be. That’s what makes them fun!! |
|
|
|
Post by frozone on Oct 3, 2017 15:57:21 GMT -5
I agree with BC.
Also, I would change the title of the article. This is a list of the most well-rounded defensemen in the league... the defensemen who can literally do it all, whether its skate it out, defend or play the PP. In that context, the rankings make more sense to me.
Guys like Weber, Seabrook, etc will get a big fat zero in a couple of those categories. I don't think I've ever seen them carry the puck into the offensive zone. That will cripple their rankings pretty quickly.
My issue with the advanced stats community is that they are usually very selective with their criteria. There seems to be a love-fest with the passers/possession wizards and surprisingly little love given to the shooters/finishers. If one wanted to, they could make a "strong" case that Scott Gomez is an excellent possession player. And they wouldn't be wrong if we're talking about possession only. But in a league where goals are so scarce and valuable, shouldn't a player like Weber, who excels at scoring goals and preventing them, at least be ranked above Petry?
|
|
|
|
Post by PTH on Oct 3, 2017 18:18:50 GMT -5
I agree with BC. Also, I would change the title of the article. This is a list of the most well-rounded defensemen in the league... the defensemen who can literally do it all, whether its skate it out, defend or play the PP. In that context, the rankings make more sense to me. Guys like Weber, Seabrook, etc will get a big fat zero in a couple of those categories. I don't think I've ever seen them carry the puck into the offensive zone. That will cripple their rankings pretty quickly. My issue with the advanced stats community is that they are usually very selective with their criteria. There seems to be a love-fest with the passers/possession wizards and surprisingly little love given to the shooters/finishers. If one wanted to, they could make a "strong" case that Scott Gomez is an excellent possession player. And they wouldn't be wrong if we're talking about possession only. But in a league where goals are so scarce and valuable, shouldn't a player like Weber, who excels at scoring goals and preventing them, at least be ranked above Petry? As someone else alluded to, there's also total time on ice. A guy who can play 30 minutes is better than a guy who tops out at 20. In fact, a 30 minute guy means you can have a 20 minute guy on the 2nd pairing, and a 10-minute, shielded-from-all-kinds-of-situations guy on the 3d pairing (thinking of guys like M-A Bergeron here, who could be a wiz on the PP but didn't have much of a defensive game at all). Another factor might be how hard it is for a guy to gel - Weber seems to be ok regardless of his partner, Markov would make any partner better but wasn't always a fit with some guys, Subban seemed to sometimes need time to gel and wasn't a good fit with some guys, etc. Again, another one: who can play both sides ? A 30 minute D who can switch sides is far more valuable than one who can't, since it means the coach has far more options for spreading out the icetime among the other 5 (instead of the other 2 on his side). |
|
