Certainly at times the player actually needs to produce as well, esp. a winger. I can buy that argument much more with centers and defensemen that are carrying a much heavier work load defensively. For a winger they've got to also have the goods offensively to a greater extent. It's not like he was getting prime match-up duty or something to where the offensive production can just be hand-waved away.
He's a complementary top sixer at best and not necessarily anything more than that. A decent regular season sort of player but when the playoffs came around he was rather quickly demoted. He was a ten minute a night guy come playoff time. An even player. He had one goal in EDM's final 13 playoff games. So, again, not sure I'd bet on an immanent break-out or that he's altogether an instrumental winner.
With Tkachuk he's 50% likelier to begin a shift in his own zone with Brown on his line, I don't think you realize how glaring it is that the stats are the same.You're reading these stats wrong. Tkachuk's xGF is 51.66% with Brown, 51.40% without him. A negligible difference. Tim Stutzle similarly did not benefit from playing with Brown, despite being a skilled player himself.
Meanwhile McDavid + Puljujarvi is 60%, McDavid - Puljujarvi is 56.5%, a noticeable difference. And Draisaitl + Puljujarvi is 59%, while Draisaitl - Puljujarvi is at 45%, an enormous difference. McDavid and Draisaitl unquestionably played better with Puljujarvi on their wings.
Capfriendly retweeted something from Brown's agency saying he was signed and framing it as an extension, but it seems like it was a misunderstanding.What was the tweet?
It was a CapFriendly tweet retweeting the player's agency's tweet that said Brown had been "signed" by WSH. I guess it was a mistake on their end. The agency quickly deleted their tweet. Then CapFriendly deleted their retweet of the deleted tweet.What was the tweet?
With Tkachuk he's 50% likelier to begin a shift in his own zone with Brown on his line, I don't think you realize how glaring it is that the stats are the same.
Draisaitl + McDavid - Pool Party 60.19%, with him 54.96%. Just Drais with and without Pulju do look more stark but it's a much smaller sample size of them together without McDavid.
And with all of that it rates him as a slightly better than average forward. Washington has plenty of slightly better than average forwards. They need a true impact forward that can help them score goals when it comes to the postseason. We'll see if they get a center in the coming days, but without getting one the trade for Brown seems superfluous and a bit restricting to making future moves.
And it’s gone. So I think that was a mistake.I saw Connor brown 3 year extension for 3.6 per...
And that is his current contact so I think his agency got some stuff confusedI saw Connor brown 3 year extension for 3.6 per...
GAR attempts to account for those other factors. It's not enough to set up a model to adjust for covariates, you have to feed that model a large enough sample of data for it to learn what the proper adjustments are, and the more factors you're trying to adjust for the bigger the sample size needed. We've had this discussion before, but I think the weird quirks in dGAR suggest that there are simply too many covariates involved for the model to be able to separate out a single player's defensive contributions within a short enough time frame for the results to be meaningful.But GAR and xGAR do take all of these factors into account. It's why I think GAR stats are much more powerful than advanced stats from even a few years ago such as on-ice Corsi, xGF, etc. Even relative stats are kind of archaic at this point compared to GAR.
So the arguments that GAR don't capture context really fall on deaf ears because these models were specifically designed to take context into account to overcome the limitations of the old stats.
WOWY stats are meaningless unless you also look at who each player is being replaced with when they're away. When Tkachuk wasn't with Brown he was primarily with Drake Batherson. When McDavid and Draisaitl weren't with Pulju they were primarily with Zack Kassian. It's not a fair comparison.I don't think this is true. Over the past 3 years Tkachuk has fared better away from Brown than with him in terms of goals and about the same with and without Brown in terms of chances.
Contrast that with Puljujarvi where both McDavid and Draisaitl see a large bump in their goal and chance differential with Puljujarvi instead of without him.
WOWY stats are meaningless unless you also look at who each player is being replaced with when they're away. When Tkachuk wasn't with Brown he was primarily with Drake Batherson. When McDavid and Draisaitl weren't with Pulju they were primarily with Zack Kassian. It's not a fair comparison.
I do agree with this. But again, GAR does (attempt) to account for this and Puljujarvi rates extremely well in this measure. He's a top 20% player offensively and a top 5% player defensively according to GAR/60. He profiles very similarly to Valeri Nichushkin. I'd love to have someone like Valeri Nichushkin driving play on the first line, and Puljujarvi could be that guy.
Overall, here's the list of the top forwards ranked by GAR/60 that have as many minutes as Puljujarvi:
View attachment 569240
In general this list passes the sniff test. A lot of good players ranked in a reasonable looking order.
Doesn't this indicate that maybe there's a lot more to Puljujarvi's game that the eye-test and Edmonton media mouth-breathers can see? Why is he the exception to this otherwise pretty reasonable list?
Connor Brown, for what it's worth, is #209 on this list. One spot ahead of Lars Eller.
There are no Caps on that list, therefore I unilaterally dismiss its relevance![]()
I think the not-so-secret secret is that Washington has been able to remain in the playoff picture due to their depth throughout the lineup rather than their star power.
But GAR and xGAR do take all of these factors into account. It's why I think GAR stats are much more powerful than advanced stats from even a few years ago such as on-ice Corsi, xGF, etc. Even relative stats are kind of archaic at this point compared to GAR.
I think Wiercioch is a perfect example of a guy where if you just look at Corsi or CorsiRel you get an entirely different picture than if you take into account his context like GAR does. GAR sees Wiercioch as a very ordinary third pairing type defender, even if his on-ice metrics were great:
View attachment 569174
So the arguments that GAR don't capture context really fall on deaf ears because these models were specifically designed to take context into account to overcome the limitations of the old stats.
I really don't like to spend time talking about advanced stats (did already enough in this thread) but the seasons where Wiercioch was a darling was 2012-13 and 2013-14 IIRC
It's also impossible to "take all of these factors into account", the best it can achieve is to take SOME of these factors into account, and partially.
I do appreciate the effort these models make to take context into account more but I just don't think it's possible. Not the game of hockey. It should never be seen as a "black and white" statistic like the number of games, minutes, goals, etc
Thinking that Jesse Puljujärvi is a better hockey player than Connor Brown is why these models will never fully work.
Even if you use 12-13 and 13-14 as the timeframe Wiercioch still ranks as very ordinary: 110th out of 193 NHLers in GAR/60.
While I do agree it's impossible to take every factor into account when developing a model, isn't the same even more true for analyzing players subjectively with the eye test? Can you really watch every single player every single night, compared them to one another in a reasonable way, and develop unbiased opinions on these guys? Do you know your own limitations when it comes to judging the quality of a hockey player? Is the fact that you're an Ottawa fan coloring your opinion on Connor Brown (and Jesse Puljujarvi as well)?
Being a puck hound is pretty much exactly what is needed on the top line with Ovechkin and Kuznetsov. Obviously Ovi is the sniper and Kuznetsov is the creator but they need someone to do the dirty work and to be defensively responsible. Wilson filled the role really well for a while but it’d be nice to be able to put him on the 2nd line with Mantha and put Osh on the third line. The Caps have a pretty gaping hole in their top six and if Brown can be that guy for them then it’ll be well worth the price.Forgot to mention in this thread but Connor Brown is actually better and more efficient in a matchup role playing with a guy like Nick Paul than in an offensive role with a skilled player like Stutzle. You have to let him play his game, which is being a puck hound.
I really don't like to spend time talking about advanced stats (did already enough in this thread) but the seasons where Wiercioch was a darling was 2012-13 and 2013-14 IIRC
It's also impossible to "take all of these factors into account", the best it can achieve is to take SOME of these factors into account, and partially.
I do appreciate the effort these models make to take context into account more but I just don't think it's possible. Not the game of hockey. It should never be seen as a "black and white" statistic like the number of games, minutes, goals, etc
Thinking that Jesse Puljujärvi is a better hockey player than Connor Brown is why these models will never fully work.