Who deserves the Selke strictly based on numbers?

[tarheelhockey]

Better metrics than "goals" ? Isn't a mark of a good defensive player "preventing goals"?

"Preventing goals" is not a metric, and I assume you know that. The formula you're using as a stand-in for "preventing goals" is clearly not working if you seriously think Aho is worse defensively than some of the players that have been named here.

I'm not even arguing that he's a Selke-level defender (he's not, which was my original point) but arguing that he's a bad defender, let alone one of the worst on his team, let alone one of the worst in the league, is just ridiculous. Like, point-and-laugh level ridiculous. Nobody who's familar with Aho's game would accept that.

 

[mynamejeff420]

"Preventing goals" is not a metric, and I assume you know that. The formula you're using as a stand-in for "preventing goals" is clearly not working if you seriously think Aho is worse defensively than some of the players that have been named here.

I'm not even arguing that he's a Selke-level defender (he's not, which was my original point) but arguing that he's a bad defender, let alone one of the worst on his team, let alone one of the worst in the league, is just ridiculous. Like, point-and-laugh level ridiculous. Nobody who's familar with Aho's game would accept that.

"Preventing goals" (or maybe preventing offense is a better term to use) is really just a stand in for a low GA/xGA/CA/etc. Basically, a player who is good at preventing offense would have a low GA (or xGA, or CA, etc), and a player that's bad at preventing offense would have a high GA/60 (or xGA/60, or CA/60, etc).

Turns out Aho is really bad at stopping the other team from scoring on him, because when he's on the ice the other team scores a lot, and gets a ton of offense in the way of shot quality (ie xGA/60). That's what makes him a bad defensive player.

 

[tarheelhockey]

"Preventing goals" (or maybe preventing offense is a better term to use) is really just a stand in for a low GA/xGA/CA/etc. Basically, a player who is good at preventing offense would have a low GA (or xGA, or CA, etc), and a player that's bad at preventing offense would have a high GA/60 (or xGA/60, or CA/60, etc).

Turns out Aho is really bad at stopping the other team from scoring on him, because when he's on the ice the other team scores a lot, and gets a ton of offense in the way of shot quality (ie xGA/60). That's what makes him a bad defensive player.

I'm really not sure what to tell you at this point. Aho is NOT a really bad defensive player. Nobody in the hockey world thinks of him as a really bad defensive player.

If theoretical data leads to a conclusion that's jarringly out of step with observational results, that's where you take a step back and ask whether the model isn't throwing a false positive.

 

[TomasHertlsRooster]

I'm really not sure what to tell you at this point. Aho is NOT a really bad defensive player. Nobody in the hockey world thinks of him as a really bad defensive player.

If theoretical data leads to a conclusion that's jarringly out of step with observational results, that's where you take a step back and ask whether the model isn't throwing a false positive.

The Hurricanes allow 5-on-5 goals at a significantly higher rate with Aho on the ice than they do with any other full-time forward on the ice. How are those observational results jarringly out of step with the data with which leads to the conclusion that he has a very poor impact on his team's rate of goals against?

 

[tarheelhockey]

The Hurricanes allow 5-on-5 goals at a significantly higher rate with Aho on the ice than they do with any other full-time forward on the ice. How are those observational results jarringly out of step with the data with which leads to the conclusion that he has a very poor impact on his team's rate of goals against?

Oh my god.

This is the undead zombie of the Plus-Minus Argument.

 

[mynamejeff420]

I'm really not sure what to tell you at this point. Aho is NOT a really bad defensive player. Nobody in the hockey world thinks of him as a really bad defensive player.

If theoretical data leads to a conclusion that's jarringly out of step with observational results, that's where you take a step back and ask whether the model isn't throwing a false positive.

It's not theoretical data though, it's goals against. It's the most concrete data you can find. There's nothing theoretical about goals against.

Oh my god.

This is the undead zombie of the Plus-Minus Argument.

As for this, if Plus-Minus isn't your thing, thankfully there are other better options for evaluating defensive play aside from Goals Against, like Expected Goals Against. Aho grades out poorly in that metrics as well, as the Canes allow a lot of shot quality against when he's on the ice.

It's not a false positive if every metric available determines that Aho is a bad defensive player, theoretical and concrete. Maybe your eye test is the one throwing out the false positive.

 

[TomasHertlsRooster]

Oh my god.

This is the undead zombie of the Plus-Minus Argument.

Looking at the rate at which a player's team allows 5-on-5 goals when they are on the ice, and comparing that to the rate at which their teammates allow 5-on-5 goals when they are on the ice, is extremely far removed from plus/minus, and takes away most of the issues with plus/minus.

It still suffers from the issue that it can be skewed goaltending, but that can generally be solved by using expected goals against instead of goals against. When we do, Aho's defensive metrics don't look much better than it did when we looked at actual goals against; he is still 3rd worst among Hurricanes forwards, behind rookie Martin Necas, and noted defensive drag Ryan Dzingel.

 

[Michael Farkas]

When do the folks who don't give "analytics" such a bad name and reputation step in and save everyone from this miscarriage...?

 

[Doctor No]

It still suffers from the issue that it can be skewed goaltending,

And opponent matchups (which is significantly more important here than goaltending variation).

 

[mynamejeff420]

And opponent matchups (which is significantly more important here than goaltending variation).

Actually RAPM takes in to account things like matchups and usage and teammates and things like that. The raw xGA total would definitely have that issue, but RAPM xGA does not. In that metric Aho actually ranks 2nd worst on the Canes, behind only Erik Haula. Even Necas and Dzingel right behind him. I imagine the reasons Necas and Dzingel end up looking better than Aho defensively once usage is taken into account is due to the model blaming Haula for the shitty defense, but regardless, even with usage taken in to account, Aho does not look remotely good defensively.

 

[Horse McHindu]

I would have gone with Alex Ovechkin. The guy was having such a good year and it would be tough to overlook him.

 

[Doctor No]

Actually RAPM takes in to account things like matchups and usage and teammates and things like that. The raw xGA total would definitely have that issue, but RAPM xGA does not. In that metric Aho actually ranks 2nd worst on the Canes, behind only Erik Haula. Even Necas and Dzingel right behind him. I imagine the reasons Necas and Dzingel end up looking better than Aho defensively once usage is taken into account is due to the model blaming Haula for the shitty defense, but regardless, even with usage taken in to account, Aho does not look remotely good defensively.

No, I agree that opponent matchups can be adjusted for - I just don't see it being discussed in the most recent posts in this thread, in statements such as "The Hurricanes allow 5-on-5 goals at a significantly higher rate with Aho on the ice than they do with any other full-time forward on the ice." That statement certainly sounds unadjusted for matchups.

 

[mynamejeff420]

No, I agree that opponent matchups can be adjusted for - I just don't see it being discussed in the most recent posts in this thread, in statements such as "The Hurricanes allow 5-on-5 goals at a significantly higher rate with Aho on the ice than they do with any other full-time forward on the ice." That statement certainly sounds unadjusted for matchups.

Yeah that does sound pretty unadjusted to me too. I think with a statement like that you need to look at it and say "why is that happening?" Is Aho allowing many more goals against because he's getting shitty goaltending (his xGA is much lower than his GA)? Is he playing really tough minutes, or are his teammates dragging him down (his unadjusted xGA is much higher than his RAPM xGA)?

I think the unadjusted GA is a good starting point, but maybe not the best way to finish. It shouldn't be the conclusion.

The thing is though, it's not like his xGA is any better. So the goalie can't be to blame. As for his RAPM xGA, it's also terrible, so it must not be his teammates/usage either. Given that information, it's tough to say his high GA is anything more than his doing. If it wasn't, there would be some sort of disparity between the GA rates and the xGA rates and/or the RAPM xGA rates.

 

[TomasHertlsRooster]

And opponent matchups (which is significantly more important here than goaltending variation).

I'm fully in favor of adjusting for matchups, which is done by metrics such as RAPM and GAR. Here is Sebastian Aho's RAPM chart from this season:

As you can see, he is an excellent offensive player, but he really struggles to suppress expected goals against. This is pretty consistent with his results in the 3-year sample prior to this season.

The reason that I don't blaze into the threads with RAPM or GAR or whatever, is because when I'm talking with somebody who isn't a big fan of analytics, or fully understanding of them, I try to work backwards from the raw metrics in order to show how the results of RAPM or GAR type metrics come to a certain conclusion, and make it easier for some to reconcile a conclusion that might initially sound outlandish to them based on their pre-conceived notions. When I personally analyze players, I tend to look into stuff like RAPM/GAR to get a broad idea of their impacts, and then I look deeper into the metrics (such as off-ice/on-ice expected goals against or whatever) in order to get an idea of whether or not their regressed results may be somewhat inaccurate. If you just walk into a thread and post a chart, people aren't going to like it - even if that chart actually provides a more nuanced, detailed, and in-depth analysis than any of the stuff you post.

Also, while opposing matchups are certainly important, and should be taken into consideration, I think that goaltending variance plays a bigger role in goals against, and I think that the impact of opposing matchups gets overstated. The reason it gets overstated is not because it isn't very important, but because the variance in opposing matchups generally isn't as high as people make it out to be. Generally, top players play most of their minutes against top players, and the variance of TOI against top players isn't that high among top players.

Granted, it's more important when you're comparing the minutes of a #1C to his team when he's off the ice, although Jordan Staal is actually Carolina's matchup center. At 5v5, Aho's most common forward opponents this year were Kuznetsov, Kreider, and Buchnevich, while Staal's were Panarin, Couturier, and Strome. Staal is facing, in my opinion, two of the very best players in the game. Also at 5v5, using PuckIQ's quality of competition metric, Aho spends 33.41% of his minutes against "elite" competition, 41.82% of his minutes against "middle" competition, and 24.77% of his minutes against "gritensity" (~4th line/replacement level) competition. By contrast, Staal spends 46.6% of his ice time against "elite" competition, 27.73% of his ice time against "middle" competition, and 23.63% of his ice time against "gritensity" competition. As you can see, Staal is clearly Carolina's matchup center, but the variance in competition isn't that high; the two of them play a more similar percentage of their ice time against middling and elite competition than one might expect. And, despite playing Staal playing the tougher matchups, the Hurricanes allow goals at a 27% higher rate and expected goals at a 18% higher rate with Aho on the ice than they do with Staal. And lastly, by RAPM, we can see that Jordan Staal's defensive impact is basically the opposite of Aho's: nearly two standard deviations above average.

If there's one Hurricanes forward who should be in the Selke conversation, it's probably Staal. However, the GF/60 impact depicted in these charts is probably a better proxy for what the PHWA primarily looks at - goals and points - than any of the other metrics depicted here, Aho, whose GF/60 impact is far superior to that of Staal, will likely wind up with more Selke votes than Staal, just like last year. Doesn't mean he deserves those votes.

 

[Michael Farkas]

Do these metrics have Aho being a bad defensive player over the last three seasons...?

 

[tarheelhockey]

Actually RAPM takes in to account things like matchups and usage and teammates and things like that. The raw xGA total would definitely have that issue, but RAPM xGA does not. In that metric Aho actually ranks 2nd worst on the Canes, behind only Erik Haula. Even Necas and Dzingel right behind him. I imagine the reasons Necas and Dzingel end up looking better than Aho defensively once usage is taken into account is due to the model blaming Haula for the shitty defense, but regardless, even with usage taken in to account, Aho does not look remotely good defensively.

I just need to emphasize what a completely ludicrous statement it is to suggest that Aho is on the Haula/Necas/Dzingel tier defensively.

It's pretty much impossible for me to objectively shoot this stuff down without having access to the full data set. But I can tell you just from basic everyday knowledge of these players, something isn't right if the model is generating these conclusions.

If Aho defended as poorly as is being asserted here, he would be absolutely skewered for it and regarded as a completely different type of player.

It's not theoretical data though, it's goals against. It's the most concrete data you can find. There's nothing theoretical about goals against.

FWIW, this is what I meant by the Plus-Minus Argument. The defense of that stat for, oh, 50 years or so was that the bottom line is goals-in and goals-out, and nothing is more concrete than those results. In reality, plus-minus very much IS a theoretical model for measuring performance, as demonstrated by its routinely mis-assigning blame for bottom-line results (as appears to be the case here).

 

[TomasHertlsRooster]

I just need to emphasize what a completely ludicrous statement it is to suggest that Aho is on the Haula/Necas/Dzingel tier defensively.

It's pretty much impossible for me to objectively shoot this stuff down without having access to the full data set. But I can tell you just from basic everyday knowledge of these players, something isn't right if the model is generating these conclusions.

If Aho defended as poorly as is being asserted here, he would be absolutely skewered for it and regarded as a completely different type of player.



FWIW, this is what I meant by the Plus-Minus Argument. The defense of that stat for, oh, 50 years or so was that the bottom line is goals-in and goals-out, and nothing is more concrete than those results. In reality, plus-minus very much IS a theoretical model for measuring performance, as demonstrated by its routinely mis-assigning blame for bottom-line results (as appears to be the case here).

Feels like your entire argument here is mostly argument ad populum. "Most people think he's a good defensive player, so he must be a good defensive player."

Here's the entire data set for RAPM this year.

Here is the explanation for how RAPM is calculated:

Reviving Regularized Adjusted Plus-Minus for Hockey

If there's any more information that you'd like me to provide, please let me know and I'll try to help. I'm excited to see you objectively shoot this stuff down, because so far, you haven't really brought anything besides saying that something is definitely true because you and other people know it is.

Do these metrics have Aho being a bad defensive player over the last three seasons...?

Had you read my post, you would've seen the answer to this:

As you can see, he is an excellent offensive player, but he really struggles to suppress expected goals against. This is pretty consistent with his results in the 3-year sample prior to this season.

View attachment 342220

When do the folks who don't give "analytics" such a bad name and reputation step in and save everyone from this miscarriage...?

Maybe you should actually read what people have to say instead of saying stuff like this?

 

[tarheelhockey]

Feels like your entire argument here is mostly argument ad populum. "Most people think he's a good defensive player, so he must be a good defensive player."
...
I'm excited to see you objectively shoot this stuff down, because so far, you haven't really brought anything besides saying that something is definitely true because you and other people know it is.

I'm not appealing to a mob here. I'm talking about broad consensus in the community which includes NHL general managers, coaches, players, analysts, and extremely critical fans like you find on HFBoards.

I've never heard anyone call Aho one of the worst defensive forwards in the league, not one time, until this thread. That assertion conflicts not only with my personal direct observation of the player, it conflicts with the consensus assessment of his abilities. To say the least, the burden of proof for something this counter-observational goes well beyond simply pointing to a proprietary metric and saying "case closed".

You may personally put 100% confidence in the numbers and 0% confidence in human observation and expertise, but not a whole lot of people think that way, even in this forum.

Here's the entire data set for RAPM this year.

View attachment 342280

Just to clarify, to be sure I'm reading the data correctly.

This metric shows Trevor van Riemsdyk as better defensively than Jaccob Slavin?

 

[The Macho King]

I'm not appealing to a mob here. I'm talking about broad consensus in the community which includes NHL general managers, coaches, players, analysts, and extremely critical fans like you find on HFBoards.

I've never heard anyone call Aho one of the worst defensive forwards in the league, not one time, until this thread. That assertion conflicts not only with my personal direct observation of the player, it conflicts with the consensus assessment of his abilities. To say the least, the burden of proof for something this counter-observational goes well beyond simply pointing to a proprietary metric and saying "case closed".

You may personally put 100% confidence in the numbers and 0% confidence in human observation and expertise, but not a whole lot of people think that way, even in this forum.



Just to clarify, to be sure I'm reading the data correctly.

This metric shows Trevor van Riemsdyk as better defensively than Jaccob Slavin?

I fail to see the issue.

 

[Bear of Bad News]

I fail to see the issue.

Care to elaborate? Otherwise, thanks for the response I guess.

 

[mynamejeff420]

I just need to emphasize what a completely ludicrous statement it is to suggest that Aho is on the Haula/Necas/Dzingel tier defensively.

It's pretty much impossible for me to objectively shoot this stuff down without having access to the full data set. But I can tell you just from basic everyday knowledge of these players, something isn't right if the model is generating these conclusions.

If Aho defended as poorly as is being asserted here, he would be absolutely skewered for it and regarded as a completely different type of player.



FWIW, this is what I meant by the Plus-Minus Argument. The defense of that stat for, oh, 50 years or so was that the bottom line is goals-in and goals-out, and nothing is more concrete than those results. In reality, plus-minus very much IS a theoretical model for measuring performance, as demonstrated by its routinely mis-assigning blame for bottom-line results (as appears to be the case here).

The full data set was already posted so I don't need to rehash that but I'll happily address the second point.

As for the Plus-Minus argument, I wasn't trying to argue that because of the Goals Against alone Aho is a bad defensive player, it's just a concrete result that should lead to more analysis. Mis-assigning blame for Goals Against is a definitely a huge issue with using a metric as simple as GA/60. Is the issue just a bad goalie, shitty luck, the competition, the players around him? It's tough to say using raw GA/60.

Now JTR already mentioned the competition aspect so I'd just refer back to that: Aho plays easier minutes than Staal, who has terrific defensive numbers, and QoC isn't really that big of a deal anyways.

The "bad goalie/shitty luck" argument would be easily explained with metrics such as xGA/60 or CA/60, if it was simply bad luck or a shitty goalie, those numbers would be better even if the GA wasn't. In that scenario it would be unfair to criticize a player for being bad at defense. If he was truly good at defense and the GA were unfairly painting him in a negative light, then his xGA or CA would refute the idea that he's poor defensively. They don't.

Metrics like RAPM and GAR do adjust for things like teammates, usage, and competition though. So if it was his usage that was to blame for his poor GA/xGA/CA numbers, it would be reflected in his RAPM or his GAR. But it's not, he still grades out as poor defensively once you adjust for those things as well.

So if Aho isn't doing well defensively by any of the raw metrics, and his adjusted metrics just as bad, what other conclusion can you make aside from him being bad at defense?

 

[TomasHertlsRooster]

I'm not appealing to a mob here. I'm talking about broad consensus in the community which includes NHL general managers, coaches, players, analysts, and extremely critical fans like you find on HFBoards.

I've never heard anyone call Aho one of the worst defensive forwards in the league, not one time, until this thread. That assertion conflicts not only with my personal direct observation of the player, it conflicts with the consensus assessment of his abilities. To say the least, the burden of proof for something this counter-observational goes well beyond simply pointing to a proprietary metric and saying "case closed".

You may personally put 100% confidence in the numbers and 0% confidence in human observation and expertise, but not a whole lot of people think that way, even in this forum.



Just to clarify, to be sure I'm reading the data correctly.

This metric shows Trevor van Riemsdyk as better defensively than Jaccob Slavin?

Well, the good news here is that what has been shown has already well exceeded somebody pointing to a proprietary metric and saying "case closed". The raw inputs for these numbers (such as the on-ice GA and xGA) have been looked into, quality of competition has been looked into, his teammates' numbers have been looked into, etc. I would happily go a step further into this and look at his teammates' numbers with and without him, and

I don't put 100% confidence in the numbers and 0% confidence in human observation. However, when the human observation consists of entirely "this is definitely not true because I know it's not", which is all that you've provided so far, then I don't know why I should have much confidence in that human observation either. If somebody were to give me a detailed breakdown of what Aho does well defensively - ideally, with some video analysis to support what they're saying - then I'd be willing to re-consider my position.

Yes, in terms of suppressing expected goals against (the best measure of defensive impact, in my opinion), this metric does indeed show Trevor van Riemsdyk posting a superior impact to Jaccob Slavin this season. Slavin does, however, have the superior impact on suppressing shots and actual goals.

 

[tarheelhockey]

So if Aho isn't doing well defensively by any of the raw metrics, and his adjusted metrics just as bad, what other conclusion can you make aside from him being bad at defense?

The other conclusion you could make is that the adjusted metrics do a poor job quantifying defense, throwing false positives like “Trevor van Riemsduk is better than Jaccob Slavin” or “Aho is one of the worst players in the league”. Those conclusions are completely unsupportable in a real-life context, and both have a very strong smell of not properly weighting TOI or matchups.

I don't put 100% confidence in the numbers and 0% confidence in human observation. However, when the human observation consists of entirely "this is definitely not true because I know it's not", which is all that you've provided so far,

I think there’s a large difference between me personally holding an opinion, and the entire world holding a consensus. This isn’t just one guy’s eye test we’re talking about, it’s the collective analysis of everyone who has an opinion on the player, and none of them have this dramatic view of Aho as a purely one-way player.

If somebody were to give me a detailed breakdown of what Aho does well defensively - ideally, with some video analysis to support what they're saying - then I'd be willing to re-consider my position.

I can sit here and put the kind of time into a TSN-quality presentation that should command a paycheck, but why? The idea that you should re-consider your position should flow naturally from the fact that your position only exists as the result of one website’s proprietary formula, with zero supporting evidence elsewhere.

Yes, in terms of suppressing expected goals against (the best measure of defensive impact, in my opinion), this metric does indeed show Trevor van Riemsdyk posting a superior impact to Jaccob Slavin this season. Slavin does, however, have the superior impact on suppressing shots and actual goals.

Then this metric is unreliable. I don’t know how the heck else to demonstrate it than simply pointing out the absurdity of this false positive. Nobody, literally nobody, thinks TVR is in any way better defensively than Slavin.

I just noticed two more that are unsupportable and bordering on completely insane: this metric has Nino Niederreiter better defensively than Teuvo Teravainen, and Jake Gardiner better defensively than Brett Pesce. If that last one doesn’t send a massive red flag, what ever could?

 

[mynamejeff420]

The other conclusion you could make is that the adjusted metrics do a poor job quantifying defense, throwing false positives like “Trevor van Riemsduk is better than Jaccob Slavin” or “Aho is one of the worst players in the league”. Those conclusions are completely unsupportable in a real-life context, and both have a very strong smell of not properly weighting TOI or matchups.



I think there’s a large difference between me personally holding an opinion, and the entire world holding a consensus. This isn’t just one guy’s eye test we’re talking about, it’s the collective analysis of everyone who has an opinion on the player, and none of them have this dramatic view of Aho as a purely one-way player.



I can sit here and put the kind of time into a TSN-quality presentation that should command a paycheck, but why? The idea that you should re-consider your position should flow naturally from the fact that your position only exists as the result of one website’s proprietary formula, with zero supporting evidence elsewhere.



Then this metric is unreliable. I don’t know how the heck else to demonstrate it than simply pointing out the absurdity of this false positive. Nobody, literally nobody, thinks TVR is in any way better defensively than Slavin.

I just noticed two more that are unsupportable and bordering on completely insane: this metric has Nino Niederreiter better defensively than Teuvo Teravainen, and Jake Gardiner better defensively than Brett Pesce. If that last one doesn’t send a massive red flag, what ever could?

When Aho is on the ice, the Hurricanes allow a lot of goals, and do so at a much higher rate compared to when he's on the bench. That's not up for debate, it's just fact.

Now that we have that information, we can then move on to the other argument, does it mean Aho is bad defensively? In my opinion, yes. Considering all the expected and adjusted metrics suggest that Aho's defense is bad, and the Hurricanes are getting poor defensive results from an on ice perspective with Aho on the ice, it makes it very unlikely that it's throwing out a false positive. The false positive may just be your assessment of Aho being good defensively.

As for the collective thinking Aho is good defensively, I'm not so sure that's accurate. I'm happy to be proven wrong but truth be told I haven't seen a lot of discussion about Aho's defensive game, good or bad. If the collective truly does believe that Aho is an elite defensive player, they sure have been quiet about it.

"The idea that you should re-consider your position should flow naturally from the fact that your position only exists as the result of one website’s proprietary formula, with zero supporting evidence elsewhere."

I don't think you're fairly characterizing JTR's point here. It's not just the proprietary formula that's spitting out the conclusion that Aho is poor on defense, it's the "proprietary formula", plus all his on ice metrics, unadjusted. That's not some formula, it's just the results when he's on the ice. As I mentioned, if it was just the formula and nothing else, there might be an issue. But it's not. The supporting evidence is the high amount of goals against when he's on the ice. That isn't "zero supporting evidence". Both JTR and myself have gone into detail about how that formula arrived at that conclusion. You ignoring that does not mean there is zero supporting evidence.

As for your point on Pesce being worse defensively than Gardiner, it is only a 1 year sample, and this season was uncharacteristically bad for Pesce. In the 3 seasons prior to this one no defenseman on the Hurricanes had better defensive numbers than Pesce. Same thing with Teravainen/Niederreiter, over a larger sample Teravainen has better defensive numbers than Niederreiter. If you're looking to evaluate "true talent", looking at only 1 year isn't super helpful, which is why 3 years is generally used. By using a 3 year sample instead of a 1 year sample, the model does find the "right answers", that Pesce is better defensively than Gardiner, and that Teravainen is better defensively than Niederreiter.

As for the van Riemsdyk vs Slavin thing, while van Riemsdyk does have better defensive numbers than Slavin this year, he does have better raw defensive numbers than him (2.25 xGA/60 for TvR, 2.31 xGA/60 for Slavin). While their deployment is different, the numbers are actually a lot more skewed in favor of van Riemsdyk. While Slavin has faced the tougher competition, and van Riemsdyk has been relatively sheltered, the spread isn't huge. The spread in teammates, however, is much bigger, with Slavin getting to play most of his minutes with the best defensive players on the team (Hamilton, Staal, Teravainen, etc), while van Riemsdyk has spent the majority of his time with some of the worst (Aho, Necas, Dzingel, etc). So while the "tougher competition" argument favors Slavin, the "shittier teammates" argument favors van Riemsdyk equally. Given that their deployment is essentially a wash, it's not crazy to say van Riemsdyk has been better defensively this year, especially given his raw numbers are better.

 

[TomasHertlsRooster]

The other conclusion you could make is that the adjusted metrics do a poor job quantifying defense, throwing false positives like “Trevor van Riemsduk is better than Jaccob Slavin” or “Aho is one of the worst players in the league”. Those conclusions are completely unsupportable in a real-life context, and both have a very strong smell of not properly weighting TOI or matchups.



I think there’s a large difference between me personally holding an opinion, and the entire world holding a consensus. This isn’t just one guy’s eye test we’re talking about, it’s the collective analysis of everyone who has an opinion on the player, and none of them have this dramatic view of Aho as a purely one-way player.



I can sit here and put the kind of time into a TSN-quality presentation that should command a paycheck, but why? The idea that you should re-consider your position should flow naturally from the fact that your position only exists as the result of one website’s proprietary formula, with zero supporting evidence elsewhere.



Then this metric is unreliable. I don’t know how the heck else to demonstrate it than simply pointing out the absurdity of this false positive. Nobody, literally nobody, thinks TVR is in any way better defensively than Slavin.

I just noticed two more that are unsupportable and bordering on completely insane: this metric has Nino Niederreiter better defensively than Teuvo Teravainen, and Jake Gardiner better defensively than Brett Pesce. If that last one doesn’t send a massive red flag, what ever could?

Well, first off, the adjusted metric is not saying that Aho is one of the worst players in the league. The fact that you've said that, despite Jeff going out of his way to emphasize that Aho is indeed a good player due to his offensive impact, and me also going out of my way to emphasize that he is an excellent offensive player, is rather telling that you're either being disingenuous or just not taking the time to listen to the perspective of the other side.

Don't get me wrong, I love Aho and he's one of the most talented (offensively) players in the league, and still one of the most valuable players in the league too.

As you can see, he is an excellent offensive player, but he really struggles to suppress expected goals against. This is pretty consistent with his results in the 3-year sample prior to this season.

Keeping the theme consistent, you're further inaccurately characterizing the argument that is being presented to you by suggesting that this position only exists as the result of one website's proprietary formula, with zero supporting evidence anywhere. Again, I don't know if you're taking the time to actually read these posts or not, but here is some of the non-proprietary supporting evidence that has been provided:

How Aho's goals against look relative to his teammates:

the Hurricanes allow 3 goals per hour at 5v5 when Aho is on the ice and 2.36 goals per hour when he is on the bench.

How Aho's expected goals against - the superior measure of defensive play that isn't skewed by goaltending variance - look relative to his teammates:

It still suffers from the issue that it can be skewed goaltending, but that can generally be solved by using expected goals against instead of goals against. When we do, Aho's defensive metrics don't look much better than it did when we looked at actual goals against; he is still 3rd worst among Hurricanes forwards, behind rookie Martin Necas, and noted defensive drag Ryan Dzingel.

The opposing matchups that Aho is facing:

At 5v5, Aho's most common forward opponents this year were Kuznetsov, Kreider, and Buchnevich, while Staal's were Panarin, Couturier, and Strome. Staal is facing, in my opinion, two of the very best players in the game. Also at 5v5, using PuckIQ's quality of competition metric, Aho spends 33.41% of his minutes against "elite" competition, 41.82% of his minutes against "middle" competition, and 24.77% of his minutes against "gritensity" (~4th line/replacement level) competition. By contrast, Staal spends 46.6% of his ice time against "elite" competition, 27.73% of his ice time against "middle" competition, and 23.63% of his ice time against "gritensity" competition. As you can see, Staal is clearly Carolina's matchup center

So, let's review the information here...

1. The Hurricanes allow 5-on-5 goals at a significantly higher rate with Aho on the ice than they do with any other forward. (This is not proprietary.)

2. The hurricanes allow 5-on-5 expected goals at a higher rate with Aho on the ice than they do with any other full-time forward besides Ryan Dzingel, Erik Haula, and Martin Necas. (This is not proprietary.)

3. Aho is not the Hurricanes matchup center. That is Jordan Staal, who plays against the highest quality of competition of any Carolina forward, and he and Aho play practically no minutes together. Despite this, Jordan Staal's defensive metrics are all far stronger than Aho's. (This is not proprietary.)

4. The best regression model we have, which accounts for competition, teammates, usage, and other factors, shows that Aho's per-minute defensive impact on expected goals against at even strength is 15th worst out of the 330 forwards who have played at least 500 minutes. (This is proprietary, but the results to this proprietary model have been shared here, as well as the calculations behind the model.)

Now, let me add just a few more things that are not proprietary.

5. Among 329 forwards who have played at least 500 minutes at 5-on-5, Aho's rate of goals against (3) ranks 51st. (16th percentile) His rate of expected goals against (2.66) ranks 48th. (15th percentile) His relative-to-teammate rate of goals against (+0.5) ranks 31st. (9th percentile) His relative-to-teammate rate of expected goals against (+0.17) ranks 49th. (15th percentile). (This is not proprietary.)

6. The rate of 5-on-5 expected goals which the Hurricanes allow with Aho on the ice (2.66) is 9% higher than the rate of 5-on-5 expected goals which they allow when Aho is on the bench (2.44). (This is not proprietary.)

7. With Aho on the ice, the rate at which the Hurricanes allow 5-on-5 expected goals is higher than that of every other team besides the Chicago Blackhawks, the Winnipeg Jets, and the New York Rangers. With Aho off the ice, the rate at which the Hurricanes allow 5-on-5 expected goals ranks 15th. (This is not proprietary.)

8. While zone starts have a fairly minimal impact on all on-ice metrics, Aho's shift start deployment is very soft. Among full-time Hurricanes forwards, only Jordan Martinook is starting a lesser percentage of his 5-on-5 shifts in the defensive zone. (This is not proprietary.)

I have provided plenty of supporting evidence. Aho's team is allowing goals and dangerous shots at a far higher rate when he is on the ice than they are when he is on the bench. Other Hurricanes forwards are generally allowing goals and expected goals at a lesser rate. The raw and relative-to-teammate rate at which he is allowing goals and expected goals are all no higher than 16th percentile among full-time forwards. He is not his team's matchup center. He is given very soft zone deployment.

There is a huge gap between Aho's raw, non-proprietary/regressed results, and those of a forward who is even half-decent defensively. In order to show us how Aho is good defensively, despite the raw results, you're going to need to provide a ton of contextual factors which show that Aho's deployment is burying him, and you're going to need to explain why the proprietary models - the methodology behind and the results of which have been provided to you - are not properly accounting for these factors.

So far, everything that you have provided has been entirely subjective, based on nothing but your feelings about his defensive play, and the fact that those feelings are generally shared by the general public. Now that you've been provided with the results of these models, I'm looking forward to you objectively shooting these results down; I'm not sure why you haven't already.