Where do you rank Seth Jones?

[Hawkaholic]

Some STL fans have been telling me Parayko is as good as him. Woof.

 

[North Cole]

I was literally just about to start this poll lol, good stuff.

Not an attack, but why do you rate him outside the top 31? Do you consider him a true 1D, or do you think there are more 1D in the league than teams?

I'm just curious, do the stats say hes that average?

 

[COHawk]

After that cyborg effort in 5OT? Def top 5

 

[Sidney the Kidney]

That's the issue though. The analytics don't suggest he's too 15, they suggest he's not even a top 30 defenseman, and arguably not a top pairing defenseman. It's when there's large discrepancies that it becomes tough to figure out what to believe and I think we need to start breaking down both to see what's really happening. I'm just not going to fault anyone for having doubts about whether he's actually elite or not.

Then that's an even worse condemnation of the accuracy of those advanced stats. If he's not even considered a top 30 defenseman, then there's something wrong with the interpretation of what those stats are supposed to suggest.

 

[JoeThorntonsRooster]

Not an attack, but why do you rate him outside the top 31? Do you consider him a true 1D, or do you think there are more 1D in the league than teams?

I'm just curious, do the stats say hes that average?

1D is pretty subjective, but I'm of the opinion that you can win with your roster constructed in a lot of different ways if your forward group is strong enough. I wouldn't call Seth Jones a 1D (although he obviously is Columbus' 1D by role) but I could see a team winning a Stanley Cup with him as their 1D.

The reason I have him outside of the top 31 is because his analytics are indeed rather mediocre.

I just posted his 3-year RAPM chart there because I think that's the best, most fair way to provide a snapshot of a player's analytics and I don't want to flood the page with a bunch of analytics. There's obviously a lot of uncertainty within these regression models with tons of variables, but even as you work your way back into the most rudimentary of metrics, it's extremely hard to reconcile the idea that Jones is an elite defenseman.

I mean, just looking at points, Alex Edler and like 25 other defensemen have a higher points per game than Jones over the past two seasons. I realize that Jones' forward group was unspectacular last year and downright poor this year, but Jones does play with a strong offensive partner in Werenski which balances that out to some degree and he also spent most of his minutes with Panarin last year. More importantly, he plays far more minutes than most other defensemen including minutes on the top power play which really inflate his raw point totals; his point rates both at 5-on-5 and on the power play are good but they aren't anywhere near the upper echelon of defensemen. I still think he's better offensively than the advanced stats indicate, but I don't think he's a top-30 offensive defenseman.

However, offense is obviously not his calling card and I think most of the people who rank him within the top ten would concede that maybe he ranks around 31st offensively (at least, they would after looking at points since I think most people just kind of assume he's top 10-15 in points for defensemen) or even lower, but by virtue of being the best defensive defenseman and a solid offensive defenseman, he's still top-5. Again, there's a bit more uncertainty surrounding defensive metrics than offensive metrics since offensive metrics have scoring rates to complement on-ice stats (before or after regression models have been applied to those on-ice stats) while defense just has the on-ice stats, but all of the evidence makes it very hard to reconcile the idea that Jones is this elite defensive defenseman. Take a look at the shots that Columbus allowed this season with and without Jones:

Spoiler

Even putting aside the expected goal rates in this chart and taking as rudimentary of an approach as possible here, the shot pattern that Columbus allows here is fairly similar with and without Jones: they limit net front shots all the way up to the lower hash marks and allow a large amount of shots from just above the right faceoff circle. They're a damn good defensive team. But the team clearly allows fewer shots in front of the net, and fewer shots from above the right faceoff circle, when Jones is on the bench. It's very hard to reconcile this guy being an elite defensive player when his team starts allowing more dangerous shots - both in general and specifically from the side of the ice that he plays - as soon as he steps on the ice.

I fully expect people to respond to this by saying that Jones faces tough competition (EDIT: I see it's already happened in here when people have brought up on/off-ice splits,) so it is expected that his team will be better without him, but there are two issues with that statement. One, Jones' most common forward opponents in Larkin/Staal/Huberdeau/Barzal don't look that much better than CBJ without Jones' most common forward opponents in Voracek/Giroux/Hayes/Galchenyuk; Columbus uses Savard and Murray a good deal against top competition, so Jones' usage isn't really as heavily pronounced as people make it out to be. Two, and more importantly, I can find at least 31 defensemen in the league who play against top competition and improve their team's defensive results without the need for the top competition excuse, so when comparing him to his peers at the top of the league, I don't really think the excuse holds.

I'd actually say the biggest (both literally and figuratively) excuse for Jones' on/off-ice splits is David Savard; he's an excellent defensive defenseman who doesn't get the credit he deserves, and more often than not, when Seth Jones is on the bench, David Savard is on the right side of that Columbus blue line. Because of Savard, I think that Jones' on/off-ice splits might be a tad unflattering, and it should be noted that Columbus is still allowing expected goals at a 7% lower rate than league average with Jones on the ice. I don't think Jones is bad defensively, and I realize that the relative metrics do miss certain things, but again, if he were truly the best defensive defenseman in the league or whatever people make him out to be, his team wouldn't be significantly better defensively without him.

Additionally, while my viewings of him are limited enough that I wouldn't feel comfortable making an assessment of his play based entirely on my eye test, and while I believe the eye test is heavily swayed by confirmation bias and my eye test will always be looking to confirm assessments that I've already made from looking at stats, I've watched a good deal of him in the past, and I've been keeping an extremely close eye on him in these playoffs, and I have been impressed by his tools, but not impressed by his play. He frequently kills his team's chances at setting up legitimately dangerous cycle chances in the offensive zone by shooting the puck from low percentage areas and sending the play the other way. His straight forward line acceleration and top speed are both strong but he really struggles with turning at the blue line and because of this, he does a very poor job of defending the rush. He joins the rush frequently and gets caught deep in the other team's zone a lot when he fails to convert on a pass or win a battle down low. I think he's lucky that he has forwards who are extremely committed on the defensive side of the puck and that his goaltending has also been amazing; if not for those two factors, I think he'd have been on the ice for a lot more goals against and he'd get a lot more criticism for it.

Again, I don't put a ton of stock into my eye test because the number of viewings are still fairly limited and my assessments mostly come down to confirmation bias, but I will just add that my eye test assessment doesn't match the stats entirely. If I watched him play and then took a guess at what his RAPM chart looked like, I'd expect that he'd be below average as both a shot and expected goal suppressor, while he'd be a high end shot generator and good xgoal/goal generator. However it's possible that the stats actually do match that assessment over the small sample size of these playoffs where I've been very focused on him; I honestly haven't looked at the playoff stats as I don't feel all that confident in the stats that I use over a small sample size like the playoffs where anecdotally, score effects seem to play a much bigger role than they do in the regular season.

I've probably said too much on this topic as it is but there are other deep dives on the internet that look deeper into Seth Jones' game from the perspective of both analytics and the eye test and try to figure out where the gap is. The conclusion basically seems to be that people are impressed whenever they see him make some flashy moves that very few other players in the league can make, but that he also really struggles in certain areas that people pay less attention to, and those areas in which he struggles have a bigger influence on his actual results. I'm not 100% certain exactly what the deal is but I feel pretty confident that there are at least 31 defensemen in the league who are better than him.

 

[JoeThorntonsRooster]

I'd argue the fact he does just as well as other defenseman on his team who don't have to go up against the other team's top line is proof he IS good.

Proof he IS good...perhaps.

Proof he is an elite defenseman? Why, when there are plenty of other defensemen in the league who also go up against the other team's top line and make significant improvement to their team's on-ice results?

Analytics are certainly not the be all and end all, i know on a site like this they can reign supreme to win debates but NHL teams dont use them as the metric to determine if a player is elite. Dubas just said that Ceci's advanced stats are great, I think that tells you all you need to know. They are a good tool to use as a secondary metric but by no means are teams with strong front offices using it as a primary tool to make decisions.

Dubas is a snake oil salesman citing a black box in order to defend a terrible decision that he made last summer and the terrible results (relative to expectations) that his team posted this season.

 

[VoluntaryDom]

1D is pretty subjective, but I'm of the opinion that you can win with your roster constructed in a lot of different ways if your forward group is strong enough. I wouldn't call Seth Jones a 1D (although he obviously is Columbus' 1D by role) but I could see a team winning a Stanley Cup with him as their 1D.

The reason I have him outside of the top 31 is because his analytics are indeed rather mediocre.

View attachment 360825

I just posted his 3-year RAPM chart there because I think that's the best, most fair way to provide a snapshot of a player's analytics and I don't want to flood the page with a bunch of analytics. There's obviously a lot of uncertainty within these regression models with tons of variables, but even as you work your way back into the most rudimentary of metrics, it's extremely hard to reconcile the idea that Jones is an elite defenseman.

I mean, just looking at points, Alex Edler and like 25 other defensemen have a higher points per game than Jones over the past two seasons. I realize that Jones' forward group was unspectacular last year and downright poor this year, but Jones does play with a strong offensive partner in Werenski which balances that out to some degree and he also spent most of his minutes with Panarin last year. More importantly, he plays far more minutes than most other defensemen which really inflates his raw point totals; his point rates both at 5-on-5 and on the power play aren't anywhere near the upper echelon of defensemen. I still think he's better offensively than the advanced stats indicate, but I don't think he's a top-30 offensive defenseman.

However, offense is obviously not his calling card and I think most of the people who rank him within the top ten would concede that maybe he ranks around 31st offensively (at least, they would after looking at points since I think most people just kind of assume he's top 10-15 in points for defensemen) or even lower, but by virtue of being the best defensive defenseman and a solid offensive defenseman, he's still top-5. Again, there's a bit more uncertainty surrounding defensive metrics than offensive metrics since offensive metrics have scoring rates to complement on-ice stats (before or after regression models have been applied to those on-ice stats) while defense just has the on-ice stats, but all of the evidence makes it very hard to reconcile the idea that Jones is this elite defensive defenseman. Take a look at the shots that Columbus allowed this season with and without Jones:

Spoiler

View attachment 360826

View attachment 360827

Even putting aside the expected goal rates in this chart and taking as rudimentary of an approach as possible here, the shot pattern that Columbus allows here is fairly similar with and without Jones: they limit net front shots all the way up to the lower hash marks and allow a large amount of shots from just above the right faceoff circle. They're a damn good defensive team. But the team without Jones clearly allows fewer shots in front of the net, and fewer shots from above the right faceoff circle, when Jones is on the bench. It's very hard to reconcile this guy being an elite defensive player when his team starts allowing more dangerous shots - both in general and specifically from the side of the ice that he plays - as soon as he steps on the ice.

I fully expect people to respond to this by saying that Jones faces tough competition, so it is expected that his team will be better without him, but there are two issues with that statement. One, Jones' most common forward opponents in Larkin/Staal/Huberdeau/Barzal don't look that much better than CBJ without Jones' most common forward opponents in Voracek/Giroux/Hayes/Galchenyuk; Columbus uses Savard and Murray a good deal against top competition, so Jones' usage isn't really as heavily pronounced as people make it out to be. Two, and more importantly, I can find at least 31 defensemen in the league who play against top competition and improve their team's defensive results without the need for the top competition excuse, so when comparing him to his peers at the top of the league I don't really think the excuse holds.

I'd actually say the biggest excuse for Jones' on/off-ice splits is David Savard; he's an excellent defensive defenseman who doesn't get the credit he deserves, and more often than not, when Seth Jones is on the bench, David Savard is on the right side of that Columbus blue line. Because of Savard, I think that Jones' on/off-ice splits might be a tad unflattering, and it should be noted that Columbus is still allowing expected goals at a 7% lower rate than league average with Jones on the ice. I don't think he's bad defensively, but again, if he were truly the best defensive defenseman in the league or whatever people make him out to be, his team wouldn't be significantly better defensively without him.

Additionally, while my viewings of him are limited enough that I wouldn't feel comfortable making an assessment of his play based entirely on my eye test, and while I believe the eye test is heavily swayed by confirmation bias and my eye test will always be looking to confirm assessments that I've already made from looking at stats, I've watched a good deal of him in the past, and I've been keeping an extremely close eye on him in these playoffs, and I have been impressed by his tools, but not impressed by his play. He frequently kills his team's chances at setting up legitimately dangerous cycle chances in the offensive zone by shooting the puck from low percentage areas and sending the play the other way. His straight forward line acceleration and top speed are both strong but he really struggles with turning at the blue line and because of this, he does a very poor job of defending the rush. He joins the rush frequently and gets caught deep in the other team's zone a lot when he fails to convert on a pass or win a battle down low. I think he's lucky that he has forwards who are extremely committed on the defensive side of the puck and that his goaltending has also been amazing; if not for those two factors, I think he'd have been on the ice for a lot more goals against and he'd get a lot more criticism for it.

Again, I don't put a ton of stock into my eye test because the number of viewings are still fairly limited and my assessments mostly come down to confirmation bias, but I will just add that my eye test assessment doesn't match the stats entirely. If I watched him play and then took a guess at what his RAPM chart looked like, I'd expect that he'd be below average as both a shot and expected goal suppressor, while he'd be a high end shot generator and good xgoal/goal generator. However it's possible that the stats actually do match that assessment over the small sample size of these playoffs where I've been very focused on him; I honestly haven't looked at the playoff stats as I don't feel all that confident in the stats that I use over a small sample size like the playoffs where anecdotally, score effects seem to play a much bigger role than they do in the regular season.

I've probably said too much on this topic as it is but there are other deep dives on the internet that look deeper into Seth Jones' game from the perspective of both analytics and the eye test and try to figure out where the gap is. The conclusion basically seems to be that people are impressed whenever they see him make some flashy moves that very few other players in the league can make, but that he also really struggles in certain areas that people pay less attention to, and those areas in which he struggles have a bigger influence on his actual results. I'm not 100% certain exactly what the deal is but I feel pretty confident that there are at least 31 defensemen in the league who are better than him.

i love you

 

[Sidney the Kidney]

Proof he IS good...perhaps.

Proof he is an elite defenseman? Why, when there are plenty of other defensemen in the league who also go up against the other team's top line and make significant improvement to their team's on-ice results?

Didn't those same charts and formulae lead you to conclude that John Carlson wasn't even a top pairing defenseman?

I understand these analytics have value to add. I'm not saying dismiss them outright or never use them as an additional source of evaluation. But at some point, I think when the conclusions they lead you to are so far and away out of wack of reality, it's time to maybe not use them as the be-all to form a conclusion about a player?

 

[JoeThorntonsRooster]

Didn't those same charts and formulae lead you to conclude that John Carlson wasn't even a top pairing defenseman?

I understand these analytics have value to add. I'm not saying dismiss them outright or never use them as an additional source of evaluation. But at some point, I think when the conclusions they lead you to are so far and away out of wack of reality, it's time to maybe not use them as the be-all to form a conclusion about a player?

I never said that Carlson wasn’t a top pairing defenseman - just that he wasn’t elite - but more importantly, what “charts” and “formulae” are you even talking about?

Columbus’ goal shares with and without Jones over the past four seasons aren’t charts or formulae or anything of the sort. Those are the real stats that go up on the Jumbotron and determine which team wins at the end of 60 minutes of hockey. If the “reality” here is that Columbus is about as good with Jones as they are without him (which it is), then wouldn’t the conclusion that Seth Jones is the best defenseman in the league actually be the conclusion that is “so far and away out of wack of reality?”

 

[Merrrlin]

I have Hedman in his own tier.

After that, there's a tier of players including Jones, Josi, Doughty, Pietrangelo, McAvoy that is all pretty close.

Jones probably is among the four best of those players, so I'd say he's top 5. Top 3 is questionable. Maybe, but not consistently a clear top 3 defenseman.

One of these things is not like the others

 

[Sidney the Kidney]

I never said that Carlson wasn’t a top pairing defenseman - just that he wasn’t elite - but more importantly, what “charts” and “formulae” are you even talking about?

Columbus’ goal shares with and without Jones over the past four seasons aren’t charts or formulae or anything of the sort. Those are the real stats that go up on the Jumbotron and determine which team wins at the end of 60 minutes of hockey. If the “reality” here is that Columbus is about as good with Jones as they are without him (which it is), then wouldn’t the conclusion that Seth Jones is the best defenseman in the league actually be the conclusion that is “so far and away out of wack of reality?”

Well, whatever "advanced stats" that you use to draw some of your conclusions. Because whatever it is you use is looking questionable when you've got Carlson as "not elite" and Jones as not a Top 30 defenseman.

Even if one were to believe neither guy is as good as some suggest, there's a difference between "I don't think Jones is Top 5, but maybe 10-15 range" or "I don't think Carlson is Top 5, but maybe 11-15 range" and whatever methodology you're using are resulting in incredibly bad takes.

Like I'm not even questioning the math, per se. I'm sure whatever calculations using whatever data you're using are correct. What I'm questioning is how accurate they actually are at concluding what you think they're accurate at concluding.

In other words, I'm not saying 2+2 equaling 4 isn't correct. I'm questioning whether using 2+2=4 to determine what color my car is is an incorrect application of the math.

 

[JoeThorntonsRooster]

Well, whatever "advanced stats" that you use to draw some of your conclusions. Because whatever it is you use is looking questionable when you've got Carlson as "not elite" and Jones as not a Top 30 defenseman.

Even if one were to believe neither guy is as good as some suggest, there's a difference between "I don't think Jones is Top 5, but maybe 10-15 range" or "I don't think Carlson is Top 5, but maybe 11-15 range" and whatever methodology you're using are resulting in incredibly bad takes.

Like I'm not even questioning the math, per se. I'm sure whatever calculations using whatever data you're using are correct. What I'm questioning is how accurate they actually are at concluding what you think they're accurate at concluding.

In other words, I'm not saying 2+2 equaling 4 isn't correct. I'm questioning whether using 2+2=4 to determine what color my car is is an incorrect application of the math.

I don’t even have an issue with ranking Carlson 11-15, and it’s really hard to engage in good faith with anybody who can’t help themselves from bringing him up in a thread that has literally nothing to do with him, but, I’ll ignore that garbage and address the actual post...

Again, a lot of the evidence that has been posted regarding Seth Jones are not “formulae” or “charts” here. As Regal showed, over the past four seasons (a huge sample), Columbus’ proficiency at out-scoring their opponent in actual goals that go on the scoreboard has been roughly the same with Seth Jones on the ice as it has been with Seth Jones on the bench. Unless something has changed, the objective of hockey is to out-score the opponent. What is wrong with using a player’s proficiency at our-scoring the opposition in order to determine how good they are at...out-scoring the opposition? Where is the issue with the application there? It looks pretty clear to me.

Also, you are making the very issues with your assessments that you inaccurately accuse the analytics people of making. You tell people that they need to take a step back from the models, that they can’t treat them as the be all, despite posts like my response to North Cole in here making it very clear that I don’t do that, yet you assume that it’s just an absolutely certain conclusion that both Carlson and Jones must both be top-15 defensemen at the absolute worst, and that anybody who concludes otherwise must be making an error in their process. Why are you absolutely 100% certain that Jones/Carlson must be top-15 defensemen? Expert or consensus opinion? Your eye test? Points? Whatever it is, it seems that you are more guilty of treating that as the be all than any of the analytics people in here are of treating their “charts and formulae” as the be all.

 

[Sidney the Kidney]

I don’t even have an issue with ranking Carlson 11-15, and it’s really hard to engage in good faith with anybody who can’t help themselves from bringing him up in a thread that has literally nothing to do with him, but, I’ll ignore that garbage and address the actual post...

Because he's an example of, just like Jones, some methodology resulting in conclusions about him that seem very off the mark. The thread itself has nothing to do with Carlson, but the METHOD you're evaluating Jones (the subject of the thread) is the same method that resulted in your conclusion about Carlson. So he does have relevance to this thread in that he's an example of how the evaluation of Jones isn't just an outlier.

Again, a lot of the evidence that has been posted regarding Seth Jones are not “formulae” or “charts” here. As Regal showed, over the past four seasons (a huge sample), Columbus’ proficiency at out-scoring their opponent in actual goals that go on the scoreboard has been roughly the same with Seth Jones on the ice as it has been with Seth Jones on the bench. Unless something has changed, the objective of hockey is to out-score the opponent. What is wrong with using a player’s proficiency at our-scoring the opposition in order to determine how good they are at...out-scoring the opposition? Where is the issue with the application there? It looks pretty clear to me.

Because that's using one single factor in determining Jones' effectiveness when it's not even just Jones himself that impacts those results. Obviously a simplistic example, but it's like assuming a player's play is reflective in him being on the ice for a goal against when he might have played his role perfectly but one of the other 4 players on the ice with him were the one that caused the goal against. Yet Jones would still be considered to have "been on the ice when a goal was scored against him".

How many top pairing defensemen consistently have the best GF/GA ratio on the team? Is Pittsburgh better off using Justin Schultz and not trading away Jamie Oleksiak and using them against the opposition's best rather than Letang because over the last three seasons, their GF/GA ratio is better than his? Letang's barely a positive GF/GA at 5on5 these past three years (188 GF vs. 181 GA).

So it seems weird that the stat you're clinging to is this whole "look at how many goals for and against he's been on the ice for compared to other defensemen on his team" is an issue for me. Otherwise, do you suggest Schultz is better than Letang?

(Note: I'm using the Pens because it's a pretty clear example of one legitimate top pairing defender in Letang having "worse" numbers than someone whose play is abysmally bad in Schultz)

Also, you are making the very issues with your assessments that you inaccurately accuse the analytics people of making. You tell people that they need to take a step back from the models, that they can’t treat them as the be all, despite posts like my response to North Cole in here making it very clear that I don’t do that, yet you assume that it’s just an absolutely certain conclusion that both Carlson and Jones must both be top-15 defensemen at the absolute worst, and that anybody who concludes otherwise must be making an error in their process. Why are you absolutely 100% certain that Jones/Carlson must be top-15 defensemen? Expert or consensus opinion? Your eye test? Points? Whatever it is, it seems that you are more guilty of treating that as the be all than any of the analytics people in here are of treating their “charts and formulae” as the be all.

No, because I don't just look at what numbers pop out of a spreadsheet to determine my opinion on a player, and I don't adhere to those numbers even when they start giving really questionable results.

I use my eye test, I use stats (both simplistic and advanced) and yes, I'm not so arrogant that I don't listen to what experts say about these players. I also WILL change my opinion if someone presents something that actually makes clearer sense or points out a flaw in what I'm evaluating. I don't see the same "budge" from you, though. You just keep doubling down on questionable evaluation of players all because you think your numbers trump anything else.

 

[aufheben]

I don't know, he looks great (or exciting) to me, but the stats—while not perfect—just straight up do not add up to a top-5 defenseman. Closer to Ekblad probably.

The Athletic just posted a great article on this debate. While stats are the basis of it, they include about 20-30 clips from the game. So it's the stats test and the eye test, with Jones not really acing either.

Video Room: He played over an hour but was Seth Jones actually good in Game 1?

All of these sequences surely speak to freakish athleticism. That Jones still had legs, after playing more than a full 60-minute hockey game all on his own, is a tremendous accomplishment of his fitness, his conditioning, and his ability as a skater.
But ask yourself this: What really did they create? Were there scoring chances to show for it or did a lot of the decisions look like the sequence below, which amounts to a smart pinch but never more than the puck rattling around along the boards?

But you also have to try to remind yourself that the game isn’t played exclusively off of the rush and that while Jones may be an asset there, and those plays may grab your attention, there are little details elsewhere that also determine whether a player is really having a positive impact.
And in some of those areas, the ones we notice less, Jones’ Tuesday night marathon performance highlighted where he has often been, at the very least, a little lacking.

One thing to look for in Jones’ game is that despite his A-level skating, he’s a little sluggish on pivots, which can result in some lost races when pucks are put in behind him. You’d like to see him win this race and be able to make a play with the puck, given his length and his positioning when it’s chipped.
Despite his length and skating, quick players regularly expose his ability to gap up. When he gives too much space, players can spin him around with that added room. When he tries to play them tight, his size limits his ability to stick with them on stops and starts.

He can also be a little passive in his own zone. Instead of using his length and strength advantage to close on forwards, he often tries to simply keep them to the outside. Some of that is likely part of John Tortorella’s playbook for his team. Some of it, on Tuesday, could have been fatigue.

 

[ClevelandUke]

Per IneffectiveMath, Cody Ceci had the best impact on 5v5 xGA/60 of all TOR D this season, and per EvolvingWild Cody Ceci led all TOR D in EV Defensive xGAR, and was second in total EVD GAR. Cody Ceci: analytics darling.




Lets stop using advanced stats to make any type of sane argument

 

[ClevelandUke]

1D is pretty subjective, but I'm of the opinion that you can win with your roster constructed in a lot of different ways if your forward group is strong enough. I wouldn't call Seth Jones a 1D (although he obviously is Columbus' 1D by role) but I could see a team winning a Stanley Cup with him as their 1D.

The reason I have him outside of the top 31 is because his analytics are indeed rather mediocre.

View attachment 360825

I just posted his 3-year RAPM chart there because I think that's the best, most fair way to provide a snapshot of a player's analytics and I don't want to flood the page with a bunch of analytics. There's obviously a lot of uncertainty within these regression models with tons of variables, but even as you work your way back into the most rudimentary of metrics, it's extremely hard to reconcile the idea that Jones is an elite defenseman.

I mean, just looking at points, Alex Edler and like 25 other defensemen have a higher points per game than Jones over the past two seasons. I realize that Jones' forward group was unspectacular last year and downright poor this year, but Jones does play with a strong offensive partner in Werenski which balances that out to some degree and he also spent most of his minutes with Panarin last year. More importantly, he plays far more minutes than most other defensemen including minutes on the top power play which really inflate his raw point totals; his point rates both at 5-on-5 and on the power play are good but they aren't anywhere near the upper echelon of defensemen. I still think he's better offensively than the advanced stats indicate, but I don't think he's a top-30 offensive defenseman.

However, offense is obviously not his calling card and I think most of the people who rank him within the top ten would concede that maybe he ranks around 31st offensively (at least, they would after looking at points since I think most people just kind of assume he's top 10-15 in points for defensemen) or even lower, but by virtue of being the best defensive defenseman and a solid offensive defenseman, he's still top-5. Again, there's a bit more uncertainty surrounding defensive metrics than offensive metrics since offensive metrics have scoring rates to complement on-ice stats (before or after regression models have been applied to those on-ice stats) while defense just has the on-ice stats, but all of the evidence makes it very hard to reconcile the idea that Jones is this elite defensive defenseman. Take a look at the shots that Columbus allowed this season with and without Jones:

Spoiler

View attachment 360826

View attachment 360827

Even putting aside the expected goal rates in this chart and taking as rudimentary of an approach as possible here, the shot pattern that Columbus allows here is fairly similar with and without Jones: they limit net front shots all the way up to the lower hash marks and allow a large amount of shots from just above the right faceoff circle. They're a damn good defensive team. But the team clearly allows fewer shots in front of the net, and fewer shots from above the right faceoff circle, when Jones is on the bench. It's very hard to reconcile this guy being an elite defensive player when his team starts allowing more dangerous shots - both in general and specifically from the side of the ice that he plays - as soon as he steps on the ice.

I fully expect people to respond to this by saying that Jones faces tough competition (EDIT: I see it's already happened in here when people have brought up on/off-ice splits,) so it is expected that his team will be better without him, but there are two issues with that statement. One, Jones' most common forward opponents in Larkin/Staal/Huberdeau/Barzal don't look that much better than CBJ without Jones' most common forward opponents in Voracek/Giroux/Hayes/Galchenyuk; Columbus uses Savard and Murray a good deal against top competition, so Jones' usage isn't really as heavily pronounced as people make it out to be. Two, and more importantly, I can find at least 31 defensemen in the league who play against top competition and improve their team's defensive results without the need for the top competition excuse, so when comparing him to his peers at the top of the league, I don't really think the excuse holds.

I'd actually say the biggest (both literally and figuratively) excuse for Jones' on/off-ice splits is David Savard; he's an excellent defensive defenseman who doesn't get the credit he deserves, and more often than not, when Seth Jones is on the bench, David Savard is on the right side of that Columbus blue line. Because of Savard, I think that Jones' on/off-ice splits might be a tad unflattering, and it should be noted that Columbus is still allowing expected goals at a 7% lower rate than league average with Jones on the ice. I don't think Jones is bad defensively, and I realize that the relative metrics do miss certain things, but again, if he were truly the best defensive defenseman in the league or whatever people make him out to be, his team wouldn't be significantly better defensively without him.

Additionally, while my viewings of him are limited enough that I wouldn't feel comfortable making an assessment of his play based entirely on my eye test, and while I believe the eye test is heavily swayed by confirmation bias and my eye test will always be looking to confirm assessments that I've already made from looking at stats, I've watched a good deal of him in the past, and I've been keeping an extremely close eye on him in these playoffs, and I have been impressed by his tools, but not impressed by his play. He frequently kills his team's chances at setting up legitimately dangerous cycle chances in the offensive zone by shooting the puck from low percentage areas and sending the play the other way. His straight forward line acceleration and top speed are both strong but he really struggles with turning at the blue line and because of this, he does a very poor job of defending the rush. He joins the rush frequently and gets caught deep in the other team's zone a lot when he fails to convert on a pass or win a battle down low. I think he's lucky that he has forwards who are extremely committed on the defensive side of the puck and that his goaltending has also been amazing; if not for those two factors, I think he'd have been on the ice for a lot more goals against and he'd get a lot more criticism for it.

Again, I don't put a ton of stock into my eye test because the number of viewings are still fairly limited and my assessments mostly come down to confirmation bias, but I will just add that my eye test assessment doesn't match the stats entirely. If I watched him play and then took a guess at what his RAPM chart looked like, I'd expect that he'd be below average as both a shot and expected goal suppressor, while he'd be a high end shot generator and good xgoal/goal generator. However it's possible that the stats actually do match that assessment over the small sample size of these playoffs where I've been very focused on him; I honestly haven't looked at the playoff stats as I don't feel all that confident in the stats that I use over a small sample size like the playoffs where anecdotally, score effects seem to play a much bigger role than they do in the regular season.

I've probably said too much on this topic as it is but there are other deep dives on the internet that look deeper into Seth Jones' game from the perspective of both analytics and the eye test and try to figure out where the gap is. The conclusion basically seems to be that people are impressed whenever they see him make some flashy moves that very few other players in the league can make, but that he also really struggles in certain areas that people pay less attention to, and those areas in which he struggles have a bigger influence on his actual results. I'm not 100% certain exactly what the deal is but I feel pretty confident that there are at least 31 defensemen in the league who are better than him.

Advanced stats are also a joke in hockey

 

[Pia8988]

Advanced stats are also a joke in hockey

the publicly available ones aren’t great yet

 

[aufheben]

the publicly available ones aren’t great yet

They're not completely random though. Like a multitude of different websites, different stats, different methods, none of them support Jones being an elite player. And his regular stats don't either.

 

[ClevelandUke]

They're not completely random though. Like a multitude of different websites, different stats, different methods, none of them support Jones being an elite player. And his regular stats don't either.

yikes, thats a bad take

 

[Sidney the Kidney]

They're not completely random though. Like a multitude of different websites, different stats, different methods, none of them support Jones being an elite player. And his regular stats don't either.

There's "not Top 5" and then there's results that lead one to conclude he's not even a Top 30 defender in the league. I won't argue if someone wants to say he's closer to the 10-15 range or whatever. I don't necessarily agree, but I'm not going to die on that specific hill. But when one starts concluding that Jones isn't even basically a #1D (ie. outside of the Top 31), then that starts to put into question how accurate these "methods" are at evaluating what they're supposed to be evaluating.

 

[JoeThorntonsRooster]

Per IneffectiveMath, Cody Ceci had the best impact on 5v5 xGA/60 of all TOR D this season, and per EvolvingWild Cody Ceci led all TOR D in EV Defensive xGAR, and was second in total EVD GAR. Cody Ceci: analytics darling.




Lets stop using advanced stats to make any type of sane argument

Cody Ceci is the furthest thing from an analytics darling. Some biased Leafs fan misinterpreting numbers on Twitter and making an inaccurate argument isn't an indictment of analytics.

 

[ClevelandUke]

Cody Ceci is the furthest thing from an analytics darling.

Well except for the fact that he is , you dont get to pick and choose what advanced stats to believe, its either all or nothing
but as i've proven, advanced stats dont matter anymore than the average persons opinon, they are incredibly flawed and mostly ignored.

 

[JoeThorntonsRooster]

Because he's an example of, just like Jones, some methodology resulting in conclusions about him that seem very off the mark. The thread itself has nothing to do with Carlson, but the METHOD you're evaluating Jones (the subject of the thread) is the same method that resulted in your conclusion about Carlson. So he does have relevance to this thread in that he's an example of how the evaluation of Jones isn't just an outlier.



Because that's using one single factor in determining Jones' effectiveness when it's not even just Jones himself that impacts those results. Obviously a simplistic example, but it's like assuming a player's play is reflective in him being on the ice for a goal against when he might have played his role perfectly but one of the other 4 players on the ice with him were the one that caused the goal against. Yet Jones would still be considered to have "been on the ice when a goal was scored against him".

How many top pairing defensemen consistently have the best GF/GA ratio on the team? Is Pittsburgh better off using Justin Schultz and not trading away Jamie Oleksiak and using them against the opposition's best rather than Letang because over the last three seasons, their GF/GA ratio is better than his? Letang's barely a positive GF/GA at 5on5 these past three years (188 GF vs. 181 GA).

So it seems weird that the stat you're clinging to is this whole "look at how many goals for and against he's been on the ice for compared to other defensemen on his team" is an issue for me. Otherwise, do you suggest Schultz is better than Letang?

(Note: I'm using the Pens because it's a pretty clear example of one legitimate top pairing defender in Letang having "worse" numbers than someone whose play is abysmally bad in Schultz)



No, because I don't just look at what numbers pop out of a spreadsheet to determine my opinion on a player, and I don't adhere to those numbers even when they start giving really questionable results.

I use my eye test, I use stats (both simplistic and advanced) and yes, I'm not so arrogant that I don't listen to what experts say about these players. I also WILL change my opinion if someone presents something that actually makes clearer sense or points out a flaw in what I'm evaluating. I don't see the same "budge" from you, though. You just keep doubling down on questionable evaluation of players all because you think your numbers trump anything else.

The Letang/Schultz thing is precisely why you don't just use GF% alone. Schultz has had some very fortunate on-ice shooting and Letang has had some unfortunate on-ice goaltending over that sample. Pittsburgh's CF% and xGF% are both about 4% higher with him on the ice than with him on the bench. More robust methods of analysis (or "charts and formulae") would lead you to conclude that Letang is the far better player.

Most high end players generally have a positive influence on their team's GF%. It is not always present even through a 3-year sample of on/off-ice GF% or what not, but it almost always is once you look into more robust methods of analysis. If you can't find it, it might be that the player isn't actually elite.

I don't think numbers trump all else. I think that in this case, the numbers rather comfortably trump any argument that I've seen in favor of Jones being elite, since I haven't really seen any actual arguments as to why Jones is elite. Speaking of what makes Jones elite, I asked previously and didn't get any response, so I'll ask again: what is it that makes Jones (not Carlson, stop talking about him, this thread isn't about him) definitively top-15? Why is that a 100% bonafide fact?

Also, I think you're completely off on the arrogance/absolute certainty/lack of "budge". If anything, that's coming more from you than anybody else. I'm not in here saying that anybody who has Jones in their top-5 or top-15 or whatever is automatically wrong, or using a questionable process to evaluate him because they think their eye test trumps everything else. I'm just stating my case as to why I don't think he's elite. You're the one that's really coming off arrogant with a lack of budge in this discussion. Not that I care, and I wouldn't bother bringing it up since pretty much everybody on an internet forum has a lack of budge, but since you brought it up I recommend you look in the mirror there.

But again, I feel I need to conclude my post by asking this again because I don't think I'll get answer otherwise: what makes Seth Jones definitively top-15? Why is that fact so 100% certain that anybody who disagrees is automatically using a flawed process of evaluation?

 

[aufheben]

Cody Ceci is like the posterboy for shitty stats.

 

[JoeThorntonsRooster]

Well except for the fact that he is , you dont get to pick and choose what advanced stats to believe, its either all or nothing
but as i've proven, advanced stats dont matter anymore than the average persons opinon, they are incredibly flawed and mostly ignored.

This isn't how things work. You can't just say "one idiot called Cody Ceci an analytics darling, so all analytics are incredibly flawed." I could make a Twitter account and say "I watched every Leafs game this season, and I think Cody Ceci is the best player in the league." By your logic, that would mean that the eye test is incredibly flawed, because one idiot improperly used the eye test to lead to an inaccurate conclusion.

Cody Ceci was literally below replacement level this season by GAR and xGAR (two of the three models that were referenced in that Tweet that called him an analytics darling). The guys who make those models have both been extremely critical of Ceci throughout his career and the analytics community absolutely slammed Dubas when he traded for and signed Ceci. 3-year RAPM is the analytic that I use the most, and here is Ceci's 3-year RAPM chart:

Spoiler

Contrary to what you've suggested, you absolutely can pick and choose what analytics you do and don't value (I'm sure you place plenty of value on points) and the analytics community is not a monolith that all thinks the same. There are very many analytics out there, each with their own nuances, which means that very few players around the league would qualify as either an "analytics darling" or an "analytics villain." Cody Ceci is one of the very few players who I would say comfortably qualifies as an "analytics villain."

Cody Ceci is like the posterboy for shitty stats.

This is a better and more brief way of saying exactly what I just said.