NHL Thread

[GD]

I think you guys are underestimating how good McDavid is probably going to be.

 

I think your concept of what probably means in this context might be off. see: NJH

[Nox]

Bergeron would probably have to be 25 and under. With McDavid you are getting 3 years at min. salary as well. Bergeron is making over 7.5 for the next 4 years and might only be good for another 5. A better trade offer would be Taveras, Strome and Nick Leddy from NYI and even that offer I could see Edmonton saying no to.

 

What? If Edmonton turned that down they might as well fold the franchise.

 

McDavid is maybe the best prospect ever but that does not guarantee he's even going to become a top 10 guy in the league.

[Frag]

There's just two ways of answering that type of question.

 

One way (you) is to say that I think player X is going to be this good.

 

The other way is to say that player X has a range of possible outcomes of varying likelihood. He could be so good, or so bad, or somewhere in the middle. The most rational thing to assume is that he's somewhere in the middle.

 

Do you think a 50th percentile outcome for McDavid has him being as good as Crosby?

 

If you simulate Connor McDavid's career 100,000 times, what would be the most common outcome?

 

In terms of draft year production, McDavid is in Crosby territory. The best way, of course, is to compare his production to past CHLers/NHLers that produced similar to McDavid during their draft years.

[Zaun of the Dead]

I think your concept of what probably means in this context might be off. see: NJH

 

I obviously know that he is not guaranteed to become a superstar. I'm not an idiot. I just don't see how you can see the type of talent that the kid displays and say 'oh, sports math says this so he's probably going to be average.' I think sports math is silly, I just take what I see and I use it to make the hypothesis that he's probably going to be quite good. Which will likely be the case.

[Frag]

I think you guys are underestimating how good McDavid is probably going to be.

 

There is the possibility that McDavid turns into Pavel Brendl (look up his draft year numbers). It's hard to tell with prospects, even ones as good as McDavid.

[GD]

I obviously know that he is not guaranteed to become a superstar. I'm not an idiot. I just don't see how you can see the type of talent that the kid displays and say 'oh, sports math says this so he's probably going to be average.' I think sports math is silly, I just take what I see and I use it to make the hypothesis that he's probably going to be quite good. Which will likely be the case.

 

It doesn't surprise me that you think sports math is silly because you completely misunderstand what it's saying.

[Frag]

I obviously know that he is not guaranteed to become a superstar. I'm not an idiot. I just don't see how you can see the type of talent that the kid displays and say 'oh, sports math says this so he's probably going to be average.' I think sports math is silly, I just take what I see and I use it to make the hypothesis that he's probably going to be quite good. Which will likely be the case.

 

That's not NJH's point. His point is to look at McDavid's production and potential in terms of a distribution (say, a 95% confidence interval). His average expected production could still be quite high, but ignoring variance is a dangerous game.

[Nox]

In terms of draft year production, McDavid is in Crosby territory. The best way, of course, is to compare his production to past CHLers/NHLers that produced similar to McDavid during their draft years.

 

This is true but that method tends to fall apart when you're at the tails of the distribution. If Crosby is his only comp, you obviously can't just regress to that sub-population mean to optimally project future performance.

[Frag]

This is true but that method tends to fall apart when you're at the tails of the distribution. If Crosby is his only comp, you obviously can't just regress to that sub-population mean to optimally project future performance.

 

This is true. The worst case scenario I have found is Pavel Brendl, but his PPG production wasn't as good as McDavid's during his draft year.

[Nox]

I just take what I see and I use it to make the hypothesis that he's probably going to be quite good.

 

Who on earth has said McDavid isn't likely to be really good? f***, keep up.

[Nox]

This is true. The worst case scenario I have found is Pavel Brendl, but his PPG production wasn't as good as McDavid's during his draft year.

 

The infamous Pavel Brendl!!

[GD]

That's not NJH's point. His point is to look at McDavid's production and potential in terms of a distribution (say, a 95% confidence interval). His average expected production could still be quite high, but ignoring variance is a dangerous game.

 

To translate into English: NJH is saying that you need to look at his likely best case scenario (say, Crosby and then some in your opinion) and his likely worst case (obviously anything can happen but being reasonable, I don't know, Duchene?), and find a middle point. That middle point is being reasonable for McDavid. So, somewhere in between Crosby 2.0 and Duchene or whatever. This is what these nerds call a 50th percentile projection. That's what we can reasonably expect McDavid to be.

[Zaun of the Dead]

Who on earth has said McDavid isn't likely to be really good? f***, keep up.

 

ok sweet, than there isn't anything to argue about.

[Frag]

This is true but that method tends to fall apart when you're at the tails of the distribution. If Crosby is his only comp, you obviously can't just regress to that sub-population mean to optimally project future performance.

 

Of course, you could instead look at the probability of McDavid's NHL production conditional on his draft year production (like what Chris Mitchell did when devising KATOH), but that's besides the point.

[Nox]

NJH just padding erryone's thank ratio ITT.

[Laika]

Duchene might be below expectations but he'd be much better than worst case scenario. Duchene is a stud!

[GD]

Duchene might be below expectations but he'd be much better than worst case scenario. Duchene is a stud!

 

I don't know these things you dickbutt I just pulled a name

[Nox]

Of course, you could instead look at the probability of McDavid's NHL production conditional on his draft year production (like what Chris Mitchell did when devising KATOH), but that's besides the point.

 

I might be misreading what you said here but did Mitchel actually do this for hockey?

[GD]

I might be misreading what you said here but did Mitchel actually do this for hockey?

 

No, for baseball. Frag didn't specify. He did it on Fangraphs Community Research, where work goes to never see the light of day

[Nox]

No, for baseball. Frag didn't specify. He did it on Fangraphs Community Research, where work goes to never see the light of day

 

I thought that was Breaking Blue? Ziiiinnngg.

 

(I actually like Breaking Blue)

[GD]

I thought that was Breaking Blue? Ziiiinnngg.

 

(I actually like Breaking Blue)

 

http://38.media.tumblr.com/tumblr_m7pkcx6ZeK1qgawlzo1_500.gif

[NorthOf49]

I thought that was Breaking Blue? Ziiiinnngg.

 

(I actually like Breaking Blue)

Have you seen the new content?

[Nox]

This seems contrived.

 

http://www.nhl.com/ice/news.htm?id=758665

[Nox]

Have you seen the new content?

 

I really don't like projections that basically just amount to running a regression. At least the inputs are reasonable and don't have a bunch of retarded exponents in them like SIERA.

[Laika]

This seems contrived.

 

http://www.nhl.com/ice/news.htm?id=758665

 

I will be so upset if they go to 3 on 3...

 

If the NHL wants less shootouts, then they should remove incentives to go to overtime in the first place. That means get rid of the 3 point games, and the point system all together. You win or lose and that's it!

 

They should also make overtime longer. Five minutes of 4 on 4 is stupid as it is, and too brief. I would want 10 minutes of 5 on 5 and then 10 minutes of 4 on 4, and if it's still tied, fine. Shootout. That's just me though... could also do 5 and 5 minutes if they don't want four hour games.

[Zaun of the Dead]

they really don't want to extend overtime because of TV scheduling and the like. If it was up to me there would be continuous OT until a winner is decided. But that will never happen.

[NorthOf49]

I really don't like projections that basically just amount to running a regression. At least the inputs are reasonable and don't have a bunch of retarded exponents in them like SIERA.

Yeah I tried to keep it fairly simple and the analysis should improve with experience. Thanks for reading, means a lot.

[Nox]

Yeah I tried to keep it fairly simple and the analysis should improve with experience. Thanks for reading, means a lot.

 

Oh that was you that wrote that? I actually didn't know!

 

I liked your writing and explanation of what you did. It was very clear. Also you get a big +1 for ensuring that you tested out of sample.

[GD]

Yeah, North's writing style is a little jumpy, but it's certainly very clear.

[Nox]

I will be so upset if they go to 3 on 3...

 

If the NHL wants less shootouts, then they should remove incentives to go to overtime in the first place. That means get rid of the 3 point games, and the point system all together. You win or lose and that's it!

 

They should also make overtime longer. Five minutes of 4 on 4 is stupid as it is, and too brief. I would want 10 minutes of 5 on 5 and then 10 minutes of 4 on 4, and if it's still tied, fine. Shootout. That's just me though... could also do 5 and 5 minutes if they don't want four hour games.

 

I think I heard the NHLPA would piss and moan or something about OT being too long. I guess that makes sense from a labour standpoint.

 

From a pure entertainment standpoint, I'd much rather see 3 on 3 than another f***ing shootout.