HFBoards The Official Scott Gomez Thread part tres - Siesta Edition
 Register FAQ Members List Search Today's Posts Mark Forums Read
 Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
 Notices Montreal Canadiens Prospects: Nathan Beaulieu, D» Jarred Tinordi, D» Nikita Scherbak, RW» All

# The Official Scott Gomez Thread part tres - Siesta Edition

07-22-2011, 09:45 AM
#951
MathMan
Registered User

Join Date: Jan 2006
Posts: 17,258
vCash: 500
Quote:
 Originally Posted by habsjunkie2 You should remove the 2nd *s from sometimes and type sometime, because it's only ever happened once, in a year that was a statistical anomaly for many.
Yes, it happened only once and was a statistical anomaly. That is precisely my point. Fluke statistical outliers happen, they happen at both sides of the bell curve, and they are highly unlikely to reoccur.

07-22-2011, 10:12 AM
#952
DAChampion
Registered User

Join Date: May 2011
Location: Canberra, Australia
Country:
Posts: 10,071
vCash: 500
Quote:
 Originally Posted by MathMan Yes, yes it is, given the amount of variability in hockey. I don't think people quite grasp how relatively small a sample a single season represents. What are you basing your confidence interval on? In any case, my point is that this is one of those one-in-twenty occurences. 95% seems like a lot, but it means that you will, on average, have a bit more than one player having such a season any given NHL club. Gomez fell off the left side on the bell curve on on-ice shooting percentage. These things happen. Rarely, but they do.
1*sqrt(42) gives you the 68% confidence interval.
2*sqrt(42) gives you the 95% confidence interval.

But really I disagree with the small sample argument. I don't think Gomez's problems were statistical flukes, I think they were systematic:

1) Injuries to Markov, Pacioretty.
2) He didn't click with AK46
3) Some have suggested that he gained weight, I don't know if this is true.

You make a good point at the end, 95% chance means you expect one such player per season, since there are 23 guys on the roster. However, I don't think chance is such a big role in the meaning we're using for chance here, as there so many other types of chance things: offseason training, injuries, chemistry, etc.

07-22-2011, 10:17 AM
#953
Lafleurs Guy
Moderator

Join Date: Jul 2007
Posts: 29,384
vCash: 500
Quote:
 Originally Posted by MathMan Yes, it happened only once and was a statistical anomaly. That is precisely my point. Fluke statistical outliers happen, they happen at both sides of the bell curve, and they are highly unlikely to reoccur.
Maybe if the guy could actually go to the net and score a few goals himself, he'd be less susceptible to this kind of misfortune.

Again, even if you're right and he was unlucky, he made no attempt to change his play or show any kind of effort to do things differently to try to break out of the slump. That's the difference between a guy like Gomez and a consistent scorer.

07-22-2011, 10:40 AM
#954
MathMan
Registered User

Join Date: Jan 2006
Posts: 17,258
vCash: 500
Quote:
 Originally Posted by Lafleurs Guy Maybe if the guy could actually go to the net and score a few goals himself, he'd be less susceptible to this kind of misfortune.
"He should go to the net" is such an overused cliché. It's just not a sensible strategy for everyone, especially if you play with Brian "sit on the goalie" Gionta all the time.

Quote:
 Originally Posted by Lafleurs Guy Again, even if you're right and he was unlucky, he made no attempt to change his play or show any kind of effort to do things differently to try to break out of the slump.
If it indeed was bad luck, changing his play to try to get out of it would be akin to trying to get out of a bad dice-rolling streak by throwing the dice with your left hand rather than your right hand. The only way to get out of a bad slump is to keep racking up scoring chances until probabilities reassert themselves. Changing your playstyle to get more scoring chances is always a good idea regardless of whether you're luck or not; changing your playstyle so you get less is obviously silly, slump or no slump; and changing your playstyle so you get the same number but are just doing something different is pretty much superstition.

Last edited by MathMan: 07-22-2011 at 10:54 AM.

07-22-2011, 10:52 AM
#955
MathMan
Registered User

Join Date: Jan 2006
Posts: 17,258
vCash: 500
Quote:
 Originally Posted by DAChampion 1*sqrt(42) gives you the 68% confidence interval. 2*sqrt(42) gives you the 95% confidence interval.
Err... I don't think it works like you think it does, but then I'm not sure what you think you're doing.

Quote:
 Originally Posted by DAChampion But really I disagree with the small sample argument. I don't think Gomez's problems were statistical flukes, I think they were systematic:
The problem is always the same -- it all comes down to the fact that his lack of scoring was due to on-ice shooting percentage, a value that has been shown to have little predictability and strong regression to the mean. If his possession game had declined, then the idea that he's got systematic issues would have more legs, but as it is, I'm just wondering what makes Gomez so special that he is unlike pretty much every other player before him who's run into a bad patch with the percentages.

07-22-2011, 11:15 AM
#956
DAChampion
Registered User

Join Date: May 2011
Location: Canberra, Australia
Country:
Posts: 10,071
vCash: 500
Quote:
 Originally Posted by MathMan Err... I don't think it works like you think it does, but then I'm not sure what you think you're doing.
How do you think it works assuming players are a valid statistical test (which in reality they're not).

I just assumed Poisson noise on points.

Quote:
 Originally Posted by MathMan The problem is always the same -- it all comes down to the fact that his lack of scoring was due to on-ice shooting percentage, a value that has been shown to have little predictability and strong regression to the mean. If his possession game had declined, then the idea that he's got systematic issues would have more legs, but as it is, I'm just wondering what makes Gomez so special that he is unlike pretty much every other player before him who's run into a bad patch with the percentages.
Can you give me some references to this work on shooting percentages?

07-22-2011, 11:25 AM
#957
habsjunkie2*

Join Date: Jan 2010
Posts: 4,865
vCash: 500
Quote:
 Originally Posted by MathMan Yes, it happened only once and was a statistical anomaly. That is precisely my point. Fluke statistical outliers happen, they happen at both sides of the bell curve, and they are highly unlikely to reoccur.
He seems to be the victim of such anomalies more often than a benefactor.

Forum Jump