Player Discussion Dan Girardi: Part V

[Machinehead]

Oduya was just a 50% possession player on this 2015 team that won the cup making him a -4.47 Relative possession player. Similar to Girardi, but not 45-46% raw possession because Hawks >> Rangers at puck possession.

Seabrook and Hammer were negative relative Corsi players on all 3 cup winning teams. Not by much but still by -2+ on the first two runs.

Robyn Regeher was a -4.00+ relative possession player while eating up 33% of the Kings total 5v5 ice time on that cup winning team.

None of those are similar to Girardi who was a -5.46 last year and is a -10.04 this year.

You're saying it's not a big deal because we're not great at possession anyway. It is a big deal.

Maybe before the last two seasons you could make that case. There's a MASSIVE difference between Girardi's 46.2% and his 51.4% corsi-off in 2015. This year Girardi is sitting at just over 40% and we're positive when he's not on the ice. That's not ok. That's not happening in Chicago or LA.

 

[TheRightWay]

Of course it is. So is the idea that we could replace a hugely important cog on the team with someone with better possession abilities and have it result in a Cup, without accounting for what that cog means to the team in any other aspect.

That which can be asserted without evidence can be dismissed without evidence. Unless you can provide a legitimate reason to be concerned, I have no reason to be concerned. You're basically asserting a fear of the unknown. We have strong reason to believe that replacing a bad possession player with a good one will make possession better, which will make winning more likely. We have NO reason to believe, until someone submits evidence otherwise, that Girardi's departure would have negative effects that outweigh the positive.

The constant search for objective meaning or proof in this regard is, in my opinion, misguided. There's a story involved that gets ignored.

The beauty of math is that there is no opinion involved. Only fact. That Girardi makes the Rangers a worse possession team is fact. That worse possession teams are less likely to win is fact. It's no different than you saying, "oh well scientists discovered this but it doesn't FEEL right so I'm going to ignore it." Your right to do so, but that's your problem. Not theirs.

We had a possession team in 2014. We didn't win. The answer lies elsewhere for this team.

We had a possession team in 2014, went the farthest we ever had since 1994, and then loss to the top possession team in the NHL. Somehow you've concluded from this that possession isn't the answer.

 

[TheRightWay]

The Rangers aren't and have never been a good possession team. Over the past 6 seasons, the best possession season was a mere 51.4% Corsi team.

Then, one could argue, "Yeah, that's because Dan Girardi eats up like half the ice time...."

Girardi's off-ice Corsi over the past 6 seasons.

2014: 53.18
2015: 51.43
2010: 51.25
2013: 50.80
2011: 50.26
2012: 47.36

Still, with the exception of 2014 season, barely a or just a mere even possession team without Girardi on the ice. And, this is after Girardi has tooken care of the matchups and top possession players.

Every cup winning team has a top4D that is used like Girardi and sees their relative corsi suffer like Girardi's.

But, most cup winning teams are really good possession teams in general, which the Rangers certainly are not. So, their raw Corsi numbers aren't bad like G's despite having similar relative numbers.

The best comparable to the Rangers is that 09 Pitt team who were a bad possession team and Hal Gill, their #2/3 defenseman was a 47% CF player on that cup winning team.

Girardi certainly can do better for himself from a possession standpoint but his numbers looks worse because the Rangers are an average-bad possession team in general. And, Girardi's only concern is to stop goals against in the short term at the expense of conceding shot attempts against. This piles up and accumulates to make his numbers even more inflated. He doesn't play the game like most defensemen.

I don't remember anyone asserting that the 2010-2013 teams were good possession teams.

 

[sbjnyc]

The beauty of math is that there is no opinion involved. Only fact. That Girardi makes the Rangers a worse possession team is fact. That worse possession teams are less likely to win is fact.

The plural of anecdote is not data. So I'm not sure where you're getting you're facts from.

 

[Filthy Dangles]

None of those are similar to Girardi who was a -5.46 last year and is a -10.04 this year.

You're saying it's not a big deal because we're not great at possession anyway. It is a big deal.

Maybe before the last two seasons you could make that case. There's a MASSIVE difference between Girardi's 46.2% and his 51.4% corsi-off in 2015. This year Girardi is sitting at just over 40% and we're positive when he's not on the ice. That's not ok. That's not happening in Chicago or LA.

small sample size this far into season. 40% is not sustainable even for seemingly-declining Girardi. He'll progress back to 46-50% at least.

Yeah, that was a bad year. Our 1C had a really bad possession year too.

Stepan and Girardi were both 46% possession players.

#2D and #1C both 46% possession players. #1G (best in league perhaps too) misses half the year and Rangers still win presidents trophy.

 

[Machinehead]

The plural of anecdote is not data. So I'm not sure where you're getting you're facts from.

The Rangers corsi goes up when Girardi is not on the ice vs when he is on the ice just about every single year.

A top 5 corsi team wins the Stanley Cup just about every single year.

They're facts.

 

[TheRightWay]

The plural of anecdote is not data. So I'm not sure where you're getting you're facts from.

Not sure which of my statements were anecdotes.

 

[sbjnyc]

The Rangers corsi goes up when Girardi is not on the ice vs when he is on the ice just about every single year.

A top 5 corsi team wins the Stanley Cup just about every single year.

They're facts.

So what you're saying is that a top 5 corsi team/cup winner has no one with lousy corsi on the team?

In any event that's not what the poster I responded to said who made a much stronger statement. What happened in the past is fact, absolutely. How what happened in the past can be used to predict the future is not as clear cut.

 

[sbjnyc]

Not sure which of my statements were anecdotes.

I know. How many cup winners had the top corsi in the league?

 

[Machinehead]

So what you're saying is that a top 5 corsi team/cup winner has no one with lousy corsi on the team?

Not in Girardi's role, no. The Rangers are without a doubt the only contender with a bonafide corsi anchor on their top pair.

In any event that's not what the poster I responded to said who made a much stronger statement. What happened in the past is fact, absolutely. How what happened in the past can be used to predict the future is not as clear cut.

Corsi tends to be a very accurate predictor of future success, that's the whole point of it. Look at the Flames and the Kings. Last year the Kings sucked and the Flames were good even though corsi said the opposite. Now the Flames are garbage and the Kings look unstoppable.

 

[TheRightWay]

So what you're saying is that a top 5 corsi team/cup winner has no one with lousy corsi on the team?

Do please find me Cup-Winning using bad possession players on their first pairing. I'll wait.

In any event that's not what the poster I responded to said who made a much stronger statement. What happened in the past is fact, absolutely. How what happened in the past can be used to predict the future is not as clear cut.

Good point. In the past, people exposed to asbestos tended to get lung disease at a crazy rate. However, that's in the past. So let's not use that evidence to make future decisions on our usage of asbestos.

Let's be clear on something. You not knowing something does not mean we as whole do not know something. You might not know how Corsi can be used to predict future winning. That doesn't mean the answer doesn't exist.

The correlation coefficient for Corsi and future winning is roughly 0.5. For adjusted Corsi, you get closer to 0.55. That's a pretty high correlation coefficient. To compare, the correlation coefficient for goals and future winning is about 0.3. In other words, Corsi predicts future winning at a significantly better rate than actual goals will.

Again, Corsi was not just a statistic that was picked for ***** and giggles. A number of tests were run on various stats to see which were best predictors of winning. Possession stats happen to be the best that we know of right now.

 

[Tawnos]

That which can be asserted without evidence can be dismissed without evidence. Unless you can provide a legitimate reason to be concerned, I have no reason to be concerned. You're basically asserting a fear of the unknown. We have strong reason to believe that replacing a bad possession player with a good one will make possession better, which will make winning more likely. We have NO reason to believe, until someone submits evidence otherwise, that Girardi's departure would have negative effects that outweigh the positive.



The beauty of math is that there is no opinion involved. Only fact. That Girardi makes the Rangers a worse possession team is fact. That worse possession teams are less likely to win is fact. It's no different than you saying, "oh well scientists discovered this but it doesn't FEEL right so I'm going to ignore it." Your right to do so, but that's your problem. Not theirs.




We had a possession team in 2014, went the farthest we ever had since 1994, and then loss to the top possession team in the NHL. Somehow you've concluded from this that possession isn't the answer.

We continue to have this issue of people not understanding that math is objective, but the application of statistics or analytics is not. For one thing, we don't have possession stats, we have possession proxy.

How can their be evidence of something that hasn't happened yet? I mean, I'll tell you that based on the actions of the coaches, other players comments and what we've been able to find out from people who cover the team, the evidence indicates that Girardi is the leader of the defense and losing a player with that impact on the rest of the team is likely to be a problem. The whole team would win less games, purely from the psychological effect. The wins might be more satisfying, but there would be less of them. And before you say it, this is not a Callahan situation where the player had been somewhat marginalized on the ice by the coach. When I say things like that, coming from a perspective of someone who has dissected and analyzed the game for a long time, it's not an opinion without evidence to back it up. It's just that the evidence I have is qualitative. You, and plenty of others, only seem to want to accept quantitative evidence, and that's fine. It's not a reason to dismiss other types of evidence as being valueless.

On your last point, it's not "somehow" that I came to that conclusion. The Rangers played that Kings team to a complete standstill, despite the fact that they didn't match up well with their roster. Bounces go towards one team or the other for several games in a row quite often. The Kings outplayed the Rangers in 3rd periods, but not really in overtimes. The bounces went all their way. Again, that there is another opinion besides yours doesn't make that opinion baseless

 

[TheRightWay]

We continue to have this issue of people not understanding that math is objective, but the application of statistics or analytics is not. For one thing, we don't have possession stats, we have possession proxy.

Whether it's actual possession or not is irrelevant in discovering its relation to winning. That being said, people have sat down with a stop watch and calculated actual possession and then evaluated it against "possession proxy." It's pretty close to a 1:1 ratio.

How can their be evidence of something that hasn't happened yet?

Based on this assertion, we should have zero ability to predict the weather. Because how could we possible predict what the weather will be on Saturday if it hasn't even happened yet?

I mean, I'll tell you that based on the actions of the coaches, other players comments and what we've been able to find out from people who cover the team, the evidence indicates that Girardi is the leader of the defense and losing a player with that impact on the rest of the team is likely to be a problem.

This is exactly, word-for-word, the argument that many made for keeping Ryan Callahan. Now he's an afterthought.

The whole team would win less games, purely from the psychological effect. The wins might be more satisfying, but there would be less of them. And before you say it, this is not a Callahan situation where the player had been somewhat marginalized on the ice by the coach. When I say things like that, coming from a perspective of someone who has dissected and analyzed the game for a long time, it's not an opinion without evidence to back it up. It's just that the evidence I have is qualitative. You, and plenty of others, only seem to want to accept quantitative evidence, and that's fine. It's not a reason to dismiss other types of evidence as being valueless.

[/QUOTE]

No, I think we just have different ideas of what "evidence" means. You're using observation and hypothesis. "I see this happening, and so I think this other thing is true." Congratulations on completing the first step of the scientific method. You have a long ways to go before coming up with anything that anyone should take seriously.

On your last point, it's not "somehow" that I came to that conclusion. The Rangers played that Kings team to a complete standstill,

Okay...

despite the fact that they didn't match up well with their roster.

Once again, conjecture.

Bounces go towards one team or the other for several games in a row quite often. The Kings outplayed the Rangers in 3rd periods, but not really in overtimes. The bounces went all their way. Again, that there is another opinion besides yours doesn't make that opinion baseless

No, this is exactly the point. Eric Tulsky asserted that the Rangers and Kings were pretty much neck-and-neck when accounting for possession, goaltending, special teams, etc. The series showed it with multiple OTs. A bounce either way, a detected goaltender interference, and maybe the Rangers win it. This is all true. I still don't see how this takes away from anything I've said. A very good possession team could absolutely beat a better possession team thanks to other influences. Definitely. For sure. Unfortunately, the Rangers aren't even at THAT point right now.

 

[Tawnos]

Now observation isn't evidence. Done!

 

[sbjnyc]

A top 5 corsi team wins the Stanley Cup just about every single year.

How often does a top 5 corsi team miss the playoffs?

Not in Girardi's role, no. The Rangers are without a doubt the only contender with a bonafide corsi anchor on their top pair.

Do please find me Cup-Winning using bad possession players on their first pairing. I'll wait.

Seidenberg was sub 50% when the bruins won? I just checked and 4 of the top 6 TOI players were sub 50% that season (I mean none as bad as Girardi's but still).

Corsi tends to be a very accurate predictor of future success, that's the whole point of it. Look at the Flames and the Kings. Last year the Kings sucked and the Flames were good even though corsi said the opposite. Now the Flames are garbage and the Kings look unstoppable.

This doesn't make sense. Should we be looking at last seasons corsi to see where the Rangers will wind up at the end of this season?

Good point. In the past, people exposed to asbestos tended to get lung disease at a crazy rate. However, that's in the past. So let's not use that evidence to make future decisions on our usage of asbestos.

Hence my comment. How much data was analyzed to determine that there is a link between asbestos exposure and cancer? What kind of statistical evidence was required to establish that link? Now compare that to what you posted.

You also wrote:

That worse possession teams are less likely to win is fact.

How often has the #1 corsi team won the cup?

Let's be clear on something. You not knowing something does not mean we as whole do not know something. You might not know how Corsi can be used to predict future winning. That doesn't mean the answer doesn't exist.

The correlation coefficient for Corsi and future winning is roughly 0.5. For adjusted Corsi, you get closer to 0.55. That's a pretty high correlation coefficient. To compare, the correlation coefficient for goals and future winning is about 0.3. In other words, Corsi predicts future winning at a significantly better rate than actual goals will.

Again, Corsi was not just a statistic that was picked for ***** and giggles. A number of tests were run on various stats to see which were best predictors of winning. Possession stats happen to be the best that we know of right now.

I'm not sure what the point of this is (and a 0.5 correlation is modest at best). And I'm not sure what you mean by future winning (ie what MH said about last year's corsi correlating to this year's results or something else).

The Rangers have been winning with middling corsi and those who highly value corsi often imply that this is why the Rangers haven't won a cup. So my objection to your comments regarded prediction of corsi and cup winners, not future winning. For cup winners there is really only anecdotal evidence that a top corsi seems to give you a better chance (since it's relatively new), but there is nothing objective about it. It is indeed what happened in the past but there is not enough data to make that an objective statement.

 

[Filthy Dangles]

you're simply uninformed or misinformed about the predictive value of analytics
Skip to 2:30

[NHL]637723[/NHL]

Also, the video implies a distinction between possession and Corsi. You can have possession but low Corsi, like the AVs. Corsi might not correlate as well to actual possession time in different players/teams cases. Girardi is probably one of these.

But, Colorado actually fell apart the next year despite having the top5 offensive zone time in the league last year, but they were bottom in the league in corsi. So did Corsi predict their fall?

It's unclear if Colorado was actually able to sustain the zone time with possessoin they had the prior year because that data isnt public. But their corsis remained the same and they missed the PO and were a bottom feeder in the league.

 

[sbjnyc]

you're simply uninformed or misinformed about the predictive value of analytics

I don't think I said anything about the predictive value of analytics. I said there isn't enough data to use corsi to say anything meaningful about cup winners.

BTW I'm unable to view the video until later.

 

[Filthy Dangles]

I don't think I said anything about the predictive value of analytics. I said there isn't enough data to use corsi to say anything meaningful about cup winners.

BTW I'm unable to view the video until later.

This doesn't make sense. Should we be looking at last seasons corsi to see where the Rangers will wind up at the end of this season?

In fact, yes. You do look at a teams corsi the following year. The fall of teams have been predicted by looking at future Shot Differential % for years. Again, you were either uninformed or misinformed about that.

We are not just talking about predicting the SC winner here. No one stat can do that, but Corsi is the best thing right now for it.

There's no arguing the predictive ability of Corsi, like you seemingly are.

 

[Revel]

In fact, yes. You do look at a teams corsi the following year. The fall of teams have been predicted by looking at future Shot Differential % for years. Again, you were either uninformed or misinformed about that.

We are not just talking about predicting the SC winner here. No one stat can do that, but Corsi is the best thing right now for it.

There's no arguing the predictive ability of Corsi, like you seemingly are.

What I don't get is how Corsi is being used as a means of predicting the future, but past results are being ignored.

For instance, LA had the #1 Corsi rating in the NHL last year, and they didn't make the playoffs.

This year, they are still one of the top Corsi teams, and are doing well (though after a rough start).

It smells like confirmation bias. Picking and choosing...

Also, I believe Carolina and Toronto are right there in the top 10 as well. They're not winning.

Don't get me wrong. CF is cool and I have enjoyed going through the stats recently. It seems like something is missing, though - like we're one piece away from figuring out a more perfect statistic for determining present and future success.

 

[sbjnyc]

In fact, yes. You do look at a teams corsi the following year. The fall of teams have been predicted by looking at future Shot Differential % for years. Again, you were either uninformed or misinformed about that.

We are not just talking about predicting the SC winner here. No one stat can do that, but Corsi is the best thing right now for it.

There's no arguing the predictive ability of Corsi, like you seemingly are.

I was referring to the SC winner, which seemed to me to be at the heart of the discussion. With lots of data you can make general observations. I never argued against that. I have no doubt that better teams will in general out chance other teams. But some people take that general concept too far and make specific claims that I don't believe are really supportable (cup winners).

 

[Raspewtin]

What I don't get is how Corsi is being used as a means of predicting the future, but past results are being ignored.

For instance, LA had the #1 Corsi rating in the NHL last year, and they didn't make the playoffs.

This year, they are still one of the top Corsi teams, and are doing well (though after a rough start).

It smells like confirmation bias. Picking and choosing...

Machinehead previously said it, two things are the bane of corsi. Goaltending and special teams. Jonathan Quick is an average NHL goaltender.

They also lost 11 shootouts.

The Kings were the 6th best ES goal differential in the NHL. Were the best possession team. It was literally Quick and the shootout that stopped them from the playoffs.

 

[Revel]

Machinehead previously said it, two things are the bane of corsi. Goaltending and special teams. Jonathan Quick is an average NHL goaltender.

They also lost 11 shootouts.

Thanks. Must have missed MH's comments. It does show that Corsi is only one piece and not necessarily an accurate predictor of future success.....but I'm going to assume that has been brought up countless times as well. Corsi = A worthwhile stat, but is not Gospel.

 

[True Blue]

The Rangers corsi goes up when Girardi is not on the ice vs when he is on the ice just about every single year.

A top 5 corsi team wins the Stanley Cup just about every single year.

They're facts.

That is fine. But teams can have excellent Corsi and still have players whose Corsi counts are average at best. That does not define if said player is good or bad. Your point is valid on which teams have done well. But that is an overall statistic. They can have excellent offensive players. They can also have excellent defensive players whose Corsi is not so hot. But the players do in preventing goals helps the offensive players pad their Corsi.

 

[True Blue]

Don't get me wrong. CF is cool and I have enjoyed going through the stats recently. It seems like something is missing, though - like we're one piece away from figuring out a more perfect statistic for determining present and future success.

What is missing is that statistical analysis is done in a vacuum and is predicated on certain things. The issue with that, off course, is that hockey is played on the ice and many such predications are not factual. The Corsi vacuum is looked upon in absolutes, and does not take into consideration other factors. Like how often AV uses Girardi on defensive draws against other teams top players. There is no room in Corsi to quantify and take that into account. Like most statistics, it has flaws.

 

[SSOne718]

How come the Rangers are in the very bottom of the league in Corsi but we are 8-2-2?

What is the explanation for that?