Can you quantify how much better Crosby makes a player? Why, yes you can.

[Jaded-Fan]

Interesting article I found.

The guy quantified exactly how much playing with Crosby increased a player's production. Everyone but oddly enough Ryan Malone had a huge bump. Yeah, that is very predictable, but having it put into raw numbers is eye opening, and likely higher than most here would have thought. I excerpted the relevant part:


I made a list of every forward who has played at least 147 minutes with Crosby from the start of the '07-08 season through Monday's games. Why 147 minutes? Because that's where I got to Michael Rupp, which seemed like far enough down the list.

Here are the scoring rates at 5-on-5 (empty net play removed) for those players in their minutes with and without Crosby over that time span:

Player TOI w/Crosby G/60 with/without A/60 with/without Pts/60 with/without Pts/60 increase

Chris Kunitz 2197:49 1.12 / 0.85 1.67 / 0.99 2.78 / 1.84 +51%
Pascal Dupuis 2064:27 1.02 / 0.80 1.66 / 0.98 2.67 / 1.79 +49%
Evgeni Malkin 1181:32 1.32 / 0.88 2.39 / 1.57 3.71 / 2.45 +51%
Bill Guerin 783:29 0.84 / 0.72 1.30 / 0.72 2.14 / 1.43 +50%
Miroslav Satan 347:15 0.86 / 0.71 1.21 / 0.83 2.07 / 1.54 +35%
Ruslan Fedotenko 346:22 0.69 / 0.59 1.91 / 0.90 2.60 / 1.49 +75%
Tyler Kennedy 343:23 1.05 / 0.76 1.22 / 0.93 2.27 / 1.69 +34%
Maxime Talbot 334:02 0.72 / 0.58 1.08 / 0.71 1.80 / 1.29 +39%
Ryan Malone 301:59 0.79 / 0.86 0.99 / 1.06 1.79 / 1.92 -7%
Matt Cooke 282:07 1.49 / 0.68 1.70 / 0.86 3.19 / 1.54 +107%
James Neal 191:27 0.94 / 1.09 2.82 / 0.97 3.76 / 2.07 +82%
Craig Adams 170:41 0.35 / 0.18 0.70 / 0.64 1.05 / 0.82 +29%
Michael Rupp 147:49 1.22 / 0.52 0.81 / 0.50 2.03 / 1.01 +101%

Four players played significant minutes with Crosby, and they all saw bumps in their scoring rate of between 49 and 51 percent. It's safe to say that's the boost we should expect for any new linemate.

http://www.sbnation.com/nhl/2014/1/...y-stats-chris-kunitz-penguins-winter-olympics

Click to read more...

 

[H382]

Surprised to see the +82% for Neal considering his chemistry with Malkin.

Edit: I could be mis-interpreting.. Too tired to read article.

 

[Big McLargehuge]

Sample size matters, and a majority of Neal's time on the ice with Sid comes on the PP, so that's not too surprising to see.

 

[Jaded-Fan]

That said, maybe the Pens should have picked Nash instead . . .

Sidney Crosby vs Rick Nash
If you had a pick, which player would you take?

http://hfboards.mandatory.com/showthread.php?t=150368&highlight=sidney+crosby

(which is another fun read)

 

[Ogrezilla]

Ryan Malone really sucked with Sid. I'd like to see that stat for him with Geno.

 

[GlassesJacketShirt]

That said, maybe the Pens should have picked Nash instead . . .

Sidney Crosby vs Rick Nash
If you had a pick, which player would you take?

http://hfboards.mandatory.com/showthread.php?t=150368&highlight=sidney+crosby

(which is another fun read)

Funny thread, although it will never top the Crosby vs. Prucha thread. At least Nash was an elite scoring winger at some point.

Edit: From another hockey board, this one stuck with me for a long time: http://www.hockeyforum.com/nhl-forum/29296-gilbert-brule-sidney-crosby-4.html

 

[orby]

That said, maybe the Pens should have picked Nash instead . . .

Sidney Crosby vs Rick Nash
If you had a pick, which player would you take?

http://hfboards.mandatory.com/showthread.php?t=150368&highlight=sidney+crosby

(which is another fun read)

I remember an old thread from Crosby's rookie season where some Rangers fans were claiming that they'd rather have Petr Prucha on their team than Sid

 

[zidell*]

Interesting article I found.

The guy quantified exactly how much playing with Crosby increased a player's production. Everyone but oddly enough Ryan Malone had a huge bump. Yeah, that is very predictable, but having it put into raw numbers is eye opening, and likely higher than most here would have thought. I excerpted the relevant part:


I made a list of every forward who has played at least 147 minutes with Crosby from the start of the '07-08 season through Monday's games. Why 147 minutes? Because that's where I got to Michael Rupp, which seemed like far enough down the list.

Here are the scoring rates at 5-on-5 (empty net play removed) for those players in their minutes with and without Crosby over that time span:

Player TOI w/Crosby G/60 with/without A/60 with/without Pts/60 with/without Pts/60 increase

Chris Kunitz 2197:49 1.12 / 0.85 1.67 / 0.99 2.78 / 1.84 +51%
Pascal Dupuis 2064:27 1.02 / 0.80 1.66 / 0.98 2.67 / 1.79 +49%
Evgeni Malkin 1181:32 1.32 / 0.88 2.39 / 1.57 3.71 / 2.45 +51%
Bill Guerin 783:29 0.84 / 0.72 1.30 / 0.72 2.14 / 1.43 +50%
Miroslav Satan 347:15 0.86 / 0.71 1.21 / 0.83 2.07 / 1.54 +35%
Ruslan Fedotenko 346:22 0.69 / 0.59 1.91 / 0.90 2.60 / 1.49 +75%
Tyler Kennedy 343:23 1.05 / 0.76 1.22 / 0.93 2.27 / 1.69 +34%
Maxime Talbot 334:02 0.72 / 0.58 1.08 / 0.71 1.80 / 1.29 +39%
Ryan Malone 301:59 0.79 / 0.86 0.99 / 1.06 1.79 / 1.92 -7%
Matt Cooke 282:07 1.49 / 0.68 1.70 / 0.86 3.19 / 1.54 +107%
James Neal 191:27 0.94 / 1.09 2.82 / 0.97 3.76 / 2.07 +82%
Craig Adams 170:41 0.35 / 0.18 0.70 / 0.64 1.05 / 0.82 +29%
Michael Rupp 147:49 1.22 / 0.52 0.81 / 0.50 2.03 / 1.01 +101%

Four players played significant minutes with Crosby, and they all saw bumps in their scoring rate of between 49 and 51 percent. It's safe to say that's the boost we should expect for any new linemate.

http://www.sbnation.com/nhl/2014/1/...y-stats-chris-kunitz-penguins-winter-olympics

At most it means that the players have better stats, not that they magically become better players. There are plenty of reasons for this. Playing with Crosby means the player is on the first line and takes on a more offensive role and is expected to score, while the same player on the 3rd line is expected to play defense first. Also, assists don't mean much. Being on the ice with Crosby (and Kunitz) is going to get you a lot of assists just by osmosis, that is secondary assists.

 

[Shady Machine]

At most it means that the players have better stats, not that they magically become better players. There are plenty of reasons for this. Playing with Crosby means the player is on the first line and takes on a more offensive role and is expected to score, while the same player on the 3rd line is expected to play defense first. Also, assists don't mean much. Being on the ice with Crosby (and Kunitz) is going to get you a lot of assists just by osmosis, that is secondary assists.

Stats are precisely what we are talking about

If you put a player with Crosby, how much "better" as in how much will their offensive production increase.

 

[Darth Vitale]

Would be interested to see correlations per season for some of those guys. For example I doubt Kunitz, Geno, or Neal's numbers increase by much if you compare the "Crosby year" to the one healthy year they had together. Of course that requires equal amounts of games to make a real correlation and that's part of the problem here. These aggregate numbers don't take other line mates and other seasonal factors into account.

Ideally you want to say "When player X played RW for Crosby for 20 games in season Y, his output was ____. When he played RW for 20 games with Geno / Sutter, his output was_____.

Aggregating the numbers in this way skews things a bit because you're comparing to "all the other games" in which those players had different Centers, different wingers, D pairings, different health status, etc. It's not really apples to apples, though I don't doubt generally that there's an uptick over playing with other C (except in the case of Kunitz or Neal playing with Malkin -- then I bet there's no uptick or only a small one).

 

[madinsomniac]

When I was arguing about Duper getting replaced on the first line I was pointing out that dupers production without Crosby was about .4 points per game and with him it was .7 to .8 ish... If you put a true scorer up there who was in the .7 to .8 range now you'd probably get a guy over a point per game. Even better if you can get a guy with any legit playmaking skills you also increase the production of kunitz and Crosby as well...

 

[terex]

Sample size matters, and a majority of Neal's time on the ice with Sid comes on the PP, so that's not too surprising to see.

Those are scoring rates at 5v5, without empty net play. No power play stats involved.

And if you roughly estimate 15 minutes of even strength time per game, the smallest sample is roughly 10 games.

 

[Darth Vitale]

I'm not going to say much more but while this is an interesting set of data, it's not very compelling if the objective is to draw a strong, "across the board" conclusion about playing with Crosby. Other than for guys like Cooke, where the effect is going to be very obvious because of who Cooke is and who he normally plays with. Example: if you took a 10 game sample from the middle of the season when Geno and Neal were tearing it up, and compared that, there's no chance Neal saw an 82% increase in productivity during his 10 games with Crosby or whatever it was.

The only real way to do this is, if all the other players lining up around the player being compared are the same, and they played a similar number of games against similar opponents, in the same season. That way you can account for playing a doormat vs. a good team 5x in a month, etc. The comparison made in the OP is way too vague in how it crunches the numbers. Too many variables not accounted for.

 

[Jaded-Fan]

I'm not going to say much more but while this is an interesting set of data, it's not very compelling if the objective is to draw a strong, "across the board" conclusion about playing with Crosby. Other than for guys like Cooke, where the effect is going to be very obvious because of who Cooke is and who he normally plays with. Example: if you took a 10 game sample from the middle of the season when Geno and Neal were tearing it up, and compared that, there's no chance Neal saw an 82% increase in productivity during his 10 games with Crosby or whatever it was.

The only real way to do this is, if all the other players lining up around the player being compared are the same, and they played a similar number of games against similar opponents, in the same season. That way you can account for playing a doormat vs. a good team 5x in a month, etc. The comparison made in the OP is way too vague in how it crunches the numbers. Too many variables not accounted for.

Doesn't the fact that the article uses around 7000 aggregate minutes between all of his examples satisfy your quibbles about breaking down data regarding when Sid was on the ice with these guys?

There is no way to address all of your points, but that huge aggregate sample size and limiting the data to 5 on 5 makes the concerns you raise to be unlikely to make any difference, no?

 

[Darth Vitale]

It's not the aggregate data set (7000 minutes among all players) that bothers me, it's how it's divided when making a comparison. It's not a quibble if it's about apples and apples, generally.

I think the most obvious candidates for that comparison are Kunitz and Dupuis, especially Kunitz. But you have to dig a little deeper even there. The interesting comparison (in my mind) would be taking minutes from the two seasons where Kunitz played LW on the first line for a healthy Sid, and when he played LW on the first line for a healthy Geno (i.e. when Sid was hurt). There are a number of factors that would make this comparison more valid than virtually all of the other comparisons:

1) You could get about a full season's worth of games from each scenario
2) He's going up against the other team's best D in both scenarios, and his role is the same in both.
3) He's probably got the same D pairings on the ice with him during many of those games

I don't know for sure but I'd guess the increase would be more like 20%, not 50% (for sure he's scored more G with Crosby the last year or so), but I forget the exact numbers from that season with Geno and Neal.

Point is Item 1 ensures that you have a large data set, eliminating the effect of aberrantly good or bad games, and erasing the bias of certain samples (used elsewhere in this example) being more full of good or bad opponents.

Points 2 and 3 ensure that the number of opportunities he gets are going to be roughly equivalent in both seasons, that he's not hampered by playing with a D one season that wasn't as good a passer as the D he played with the prior season, or not as good on the boards, or pinched a lot less or whatever. Basically you're eliminating the effect of personnel on the ice (other than Crosby or Geno) who could substantially change the offensive opportunities that arise during a game.

For most of those players, including Neal, you can't really make these kinds of comparisons because the number of games at a stretch and who they were playing with, who was hurt, etc... is too random.

 

[ColePens]

Good read, actually. I can see the PP stats working in his favor so I will definitely take this with a grain of salt. I also think it's important we realize the numbers. 30% of 30 points is only 9 extra points. So we take a player and make him a 39 pt player instead of a 30. That's not really that insanely good. So taking the 3rd/4th liners of world (Dupers) and sticking him PERMANENTLY with Sid for his career is bad. But if you take a guy like Hossa... then the numbers look huge.

So this article does explain just how damn good Sid is. It also makes a great case for getting him a strong winger instead of just throwing random players in there. Kunitz is a good winger to give him. Imagine that line w/ another big time winger and using our draft idea of playing young talented players mixed w/ good solid vets. That is how the team should be structured.

And I think my opinion will come to a bit of shock to the vets around this board. I've never been a "GET WINGERZ" guy. In fact, I have been against that until our defense developed. Here we are. Our defense is developing before our eyes and we are sticking them in the box each night while throwing great 4th line players on Sid's wing (Vitale) and hoping magic happens. Sid is the best player in the NHL... he's not a magician.

 

[Jaded-Fan]

It's not the aggregate data set (7000 minutes among all players) that bothers me, it's how it's divided when making a comparison. It's not a quibble if it's about apples and apples, generally.

I think the most obvious candidates for that comparison are Kunitz and Dupuis, especially Kunitz. But you have to dig a little deeper even there. The interesting comparison (in my mind) would be taking minutes from the two seasons where Kunitz played LW on the first line for a healthy Sid, and when he played LW on the first line for a healthy Geno (i.e. when Sid was hurt). There are a number of factors that would make this comparison more valid than virtually all of the other comparisons:

1) You could get about a full season's worth of games from each scenario
2) He's going up against the other team's best D in both scenarios, and his role is the same in both.
3) He's probably got the same D pairings on the ice with him during many of those games

I don't know for sure but I'd guess the increase would be more like 20%, not 50% (for sure he's scored more G with Crosby the last year or so), but I forget the exact numbers from that season with Geno and Neal.

Point is Item 1 ensures that you have a large data set, eliminating the effect of aberrantly good or bad games, and erasing the bias of certain samples (used elsewhere in this example) being more full of good or bad opponents.

Points 2 and 3 ensure that the number of opportunities he gets are going to be roughly equivalent in both seasons, that he's not hampered by playing with a D one season that wasn't as good a passer as the D he played with the prior season, or not as good on the boards, or pinched a lot less or whatever. Basically you're eliminating the effect of personnel on the ice (other than Crosby or Geno) who could substantially change the offensive opportunities that arise during a game.

For most of those players, including Neal, you can't really make these kinds of comparisons because the number of games at a stretch and who they were playing with, who was hurt, etc... is too random.

If you are saying 'take these numbers with a grain of salt, the methodology is imperfect' I agree. But given the large sample size and the limitation to 5 on 5 I disagree with you about the variance. I can not see the variance being from 50 percent all the way to 20 percent as in your example. The methodology should produce a fairly close approximation. If you said maybe 5 percent one way or another I could agree.

 

[Pancakes]

Our defense is developing before our eyes and we are sticking them in the box each night while throwing great 4th line players on Sid's wing (Vitale) and hoping magic happens. Sid is the best player in the NHL... he's not a magician.

I'm sure Shero is aware of this too. Fact is it's really hard to make trades in today's NHL. The dollars have to fit, and the other team has to want to make the trade. We can sit here and bang the wingers drum all we want, but that doesn't mean Shero is magically going to be able to make a deal that would help us.

Plus every team thinks they are in the playoff races these days. Everyone thinks they are one winning streak away from getting in. Every year it seems like there are less and less teams willing to sell off their players.

Don't get your hopes up on a great young winger coming here. We were beyond fortunate to get Neal. Most likely if we get anything at all it will be another rental winger, and even that's dicey since there will be so few sellers.

 

[PensFreak]

The only constant is Sidney Crosby here. And even strength, I suppose. Over 147 minutes, all the other variable would likely even out to create an accurate representation of the effect that Crosby has with a player.

 

[Big McLargehuge]

Those are scoring rates at 5v5, without empty net play. No power play stats involved.

And if you roughly estimate 15 minutes of even strength time per game, the smallest sample is roughly 10 games.

10 games still qualifies as a small sample size in my book.