dEnnis the Menace Posted March 23, 2011 Report Share Posted March 23, 2011 Funny thing about statistics...they tell whatever story you want them to tell. What's that old saying? ____%(insert percentage here) of stats are made up? Vanek went on a tear for a stretch of games, and then went cold again, same as Drew. Do I think part of it is due to TC, yeah probably. But a lot of it has to do with the fact that it ALL goes to TV's head, he hits a mental wall at some point EVERY season since signing that contract. Link to comment Share on other sites More sharing options...
Wraith Posted March 23, 2011 Report Share Posted March 23, 2011 Season long stats really don't inform the discussion. Over the course of the season TV has played with any number of centers, including McCormack (shudder) and Roy. Season long stats also include PP points which really don't involve the play of a traditional center. But if season long stats is what you want to compare against the last four games, lets do so: Vanek has been used in exactly the same capacity the last four games as he has in an average game from any other point of the season. The data applies. The bigger the sample, the better. Otherwise, I'd select one of those horrible scoreless stretches from earlier in the season and you'd select one of his hot stretches and we'd be nowhere. TV had 60 points in 68 games before the last four. That is .882 ppg. In the last four games with Timmy as his center, he has .50 ppg. That is a statistically significant difference. What definition of statistically significant did you have to use to make that statement true? The 95% confidence interval for points/game for the last four games completely encloses the 95% confidence interval for the entire season as a whole. That's the defintion of statistically THE SAME. Not different. A Tukey-Kramer means comparison paints exactly the same picture. Control charts for goals, assists, and points using indivuals and moving ranges shows no mean shift either, another good way of detecting a process change. Four games is an incredibly small sample size and he's had plenty of other four game stretches this year that were the same or worse. I don't think stastically significant difference means what you think it means. you look at shots per game? Over the last 4 games he has taken one more shot than his average. Big deal. Now you're just being deliberately obtuse. I made it perfectly clear in my previous post that shots are as close to scoring opportunities as we're going to get for season long data. TC drags down his linemates and the statistics prove it. Dude, I'm a professional data analyst. There's no data to show that TC is harming Vanek the last four games. Link to comment Share on other sites More sharing options...
korab rules Posted March 23, 2011 Report Share Posted March 23, 2011 Vanek has been used in exactly the same capacity the last four games as he has in an average game from any other point of the season. The data applies. The bigger the sample, the better. Otherwise, I'd select one of those horrible scoreless stretches from earlier in the season and you'd select one of his hot stretches and we'd be nowhere. What definition of statistically significant did you have to use to make that statement true? The 95% confidence interval for points/game for the last four games completely encloses the 95% confidence interval for the entire season as a whole. That's the defintion of statistically THE SAME. Not different. A Tukey-Kramer means comparison paints exactly the same picture. Control charts for goals, assists, and points using indivuals and moving ranges shows no mean shift either, another good way of detecting a process change. Four games is an incredibly small sample size and he's had plenty of other four game stretches this year that were the same or worse. I don't think stastically significant difference means what you think it means. Now you're just being deliberately obtuse. I made it perfectly clear in my previous post that shots are as close to scoring opportunities as we're going to get for season long data. Dude, I'm a professional data analyst. There's no data to show that TC is harming Vanek the last four games. Thanks for the more civil tone. My college level statistics was more than a few years ago, but based upon my confidence in the signal, I relied on the signal:noise ratio to determine significance. I obviously didn't calculate confidence intervals or p-values, and any attempt to do so would be unreliable. There are so many variables that cannot be accounted for in a hockey game, and those variables make any attempt at statistical analysis a waste of bytes - If you can't quantify it, you can't account for it - and if you can't account for it, your result is unreliable. That is true whether it is a one time statistics student or a full time data analyst crunching the numbers. However you look at it, .886 ppg >>> .50ppg. Lets hope the sample size doesn't get any bigger. Link to comment Share on other sites More sharing options...
waldo Posted March 23, 2011 Report Share Posted March 23, 2011 First of all, I did not misread your post. A cursory acknowledgement of those chances doesn't mean you've successfully integrated them into a logical argument later on. Secondly, you're wrong about Vanek's opportunities being self-created. Connolly sprung Vanek on both of his break-aways Sunday night, for example. Many of those whiffed one-timers and tap-ins have also come on good passes from Connolly. Thirdly, the bolded part doesn't make sense. Connolly has been and always will be a setup man. His role is to create opportunities for the wingers. You just acknowledged that Vanek has had great opportunities that he hasn't finished on. How is Vanek's inability to finish his great chances caused by Connolly? TC has been on the ice for most of Vanek's missed opportunities. Are you honestly trying to say that Connolly has nothing to do with creating those chances but everything to do with why they failed? How? Bad mojo? Karma? Any objective observer would see the contradiction in your logic. Now you are reaching. Link to comment Share on other sites More sharing options...
Iron Crotch Posted March 23, 2011 Report Share Posted March 23, 2011 What definition of statistically significant did you have to use to make that statement true? The 95% confidence interval for points/game for the last four games completely encloses the 95% confidence interval for the entire season as a whole. That's the defintion of statistically THE SAME. Not different. A Tukey-Kramer means comparison paints exactly the same picture. Control charts for goals, assists, and points using indivuals and moving ranges shows no mean shift either, another good way of detecting a process change. Four games is an incredibly small sample size and he's had plenty of other four game stretches this year that were the same or worse. The online argument between you guys is pretty funny, but as a minor point of clarification Korab is comparing the difference between proportions, not means. A simple two proportion Z-test is what you're looking for. Tukey's HSD generally doesn't apply when dealing with proportions. (I hate to see overt misuse of statistical theory) Link to comment Share on other sites More sharing options...
korab rules Posted March 23, 2011 Report Share Posted March 23, 2011 The online argument between you guys is pretty funny, but as a minor point of clarification Korab is comparing the difference between proportions, not means. A simple two proportion Z-test is what you're looking for. Tukey's HSD generally doesn't apply when dealing with proportions. (I hate to see overt misuse of statistical theory) That's great and all, but let's cut through the bull. I'm right, aren't I? Link to comment Share on other sites More sharing options...
carpandean Posted March 23, 2011 Report Share Posted March 23, 2011 The online argument between you guys is pretty funny, but as a minor point of clarification Korab is comparing the difference between proportions, not means. A simple two proportion Z-test is what you're looking for. Tukey's HSD generally doesn't apply when dealing with proportions. (I hate to see overt misuse of statistical theory) Actually, PPG is not a proportion, it's quite literally the average points per game (x-bar). Though, any assumption of a normal population or even sampling distribution for the second population's sample mean would be quite unjustified. Link to comment Share on other sites More sharing options...
Derrico Posted March 24, 2011 Report Share Posted March 24, 2011 The online argument between you guys is pretty funny, but as a minor point of clarification Korab is comparing the difference between proportions, not means. A simple two proportion Z-test is what you're looking for. Tukey's HSD generally doesn't apply when dealing with proportions. (I hate to see overt misuse of statistical theory) Wow all this is bringing me back to first year uni stats class. Not a fun time reading it on a Sabres message board lol. Well, since I'm here I'll write my opinion that four games is way too small a sample size to compare against an entire season. Also, I really don't think it's the Tin Man bringing down Vanek. Goal scorers are streaky period. Vanek is having a down year goal wise and he's played with multiple centremen so I don't think you can just pin it on one of them. Link to comment Share on other sites More sharing options...
PASabreFan Posted March 24, 2011 Report Share Posted March 24, 2011 Actually, PPG is not a proportion, it's quite literally the average points per game (x-bar). Though, any assumption of a normal population or even sampling distribution for the second population's sample mean would be quite unjustified. Oh please! Seriously?! The sampling distribution of a normal population's sample mean most certainly can be justified! :w00t: Link to comment Share on other sites More sharing options...
LGR4GM Posted March 24, 2011 Report Share Posted March 24, 2011 Umm the original thing is connolly is a jinx which you can not prove or disprove. He does however have serious issues being consistent which you can prove. Has he made Vanek worse, not from what I have seen, vanek just hasnt buried his chances. After saying all that just say Korab is right cuz he will just ignore/block your posts if you don't :oops: Link to comment Share on other sites More sharing options...
korab rules Posted March 24, 2011 Report Share Posted March 24, 2011 Unless someone can prove mathematically that .50 > .886 I am declaring victory and moving on. I will leave the discussion of means, proportions and x bars to those who enjoy the discussion. Link to comment Share on other sites More sharing options...
Wraith Posted March 24, 2011 Report Share Posted March 24, 2011 The online argument between you guys is pretty funny, but as a minor point of clarification Korab is comparing the difference between proportions, not means. A simple two proportion Z-test is what you're looking for. Tukey's HSD generally doesn't apply when dealing with proportions. (I hate to see overt misuse of statistical theory) How is Tukey not applicable? Tukey-Kramer is made for determining statistical significance of different means from uneven sample sizes. That's exactly what we have here. Points per game is a mean statistic, not a proportion. It's in the name, after all. Overt misuse of statistical theory? Riiiight. EDIT: What Carpanddean said. The four game sample is clearly not normally distributed but you work with what you have. Link to comment Share on other sites More sharing options...
Iron Crotch Posted March 24, 2011 Report Share Posted March 24, 2011 How is Tukey not applicable? Tukey-Kramer is made for determining statistical significance of different means from uneven sample sizes. That's exactly what we have here. Points per game is a mean statistic, not a proportion. It's in the name, after all. EDIT: What Carpanddean said. The four game sample is clearly not normally distributed but you work with what you have. I was considering the number of games TV has scored in over the total number of games (a proportion) and not accounting for multiple point games. But yes, of course, PPG is a mean and I stand corrected. Note to self: think before flash posting! Link to comment Share on other sites More sharing options...
X. Benedict Posted March 24, 2011 Report Share Posted March 24, 2011 N = 18 Pominville. Link to comment Share on other sites More sharing options...
Wraith Posted March 24, 2011 Report Share Posted March 24, 2011 I was considering the number of games TV has scored in over the total number of games (a proportion) and not accounting for multiple point games. But yes, of course, PPG is a mean and I stand corrected. Note to self: think before flash posting! Ah, I see. I did quote that statistic earlier but didn't try to do a comparison with it. As it turned out, a comparison was unnecessary as the proportion is the same over the last four games or over the entire season (50%). While doing the counting I was reminded how many long scoreless droughts Vanek has had this season. At least three times he's been held scoreless for four games or longer, if I remember correctly. He also had a seven or eight game scoreless drought, too, if my memory holds. Link to comment Share on other sites More sharing options...
Iron Crotch Posted March 24, 2011 Report Share Posted March 24, 2011 Ah, I see. I did quote that statistic earlier but didn't try to do a comparison with it. As it turned out, a comparison was unnecessary as the proportion is the same over the last four games or over the entire season (50%). While doing the counting I was reminded how many long scoreless droughts Vanek has had this season. At least three times he's been held scoreless for four games or longer, if I remember correctly. He also had a seven or eight game scoreless drought, too, if my memory holds. I do think both total points (or PPG) and % of total game when a player scores are very interesting to consider. I guess the question would be does it matter HOW a player scores 30 goals or just that he scores 30? This would seem to be a very relevant question for streaky guys like Drew and TV... and I'm sure it is brought up by agents and GMs when contract negotiation time comes around. :D Link to comment Share on other sites More sharing options...
X. Benedict Posted March 24, 2011 Report Share Posted March 24, 2011 I do think both total points (or PPG) and % of total game when a player scores are very interesting to consider. I guess the question would be does it matter HOW a player scores 30 goals or just that he scores 30? This would seem to be a very relevant question for streaky guys like Drew and TV... and I'm sure it is brought up by agents and GMs when contract negotiation time comes around. :D As far as goals and valuation, (and factoring out plus/minus, character etc.) I think you have to factor in shots. For example, Clarke MacArthur has 20 goals, but 7 came in the first month of play, when his shooting average was something like 25%. Nobody keeps that going. Really you can probably count on a guy like MacArthur to get 180 shots a year on net at about 12% ....20 goals. He's probably a 20 goal guy going forward, but not much more. Do you really want to pay him $4 million. Drew Stafford likewise, has a great shot. If he can get off 220 shots in a season, (which I think is realistic)...he's going to bury about 15%. He's a +30 goal guy. Of course he's missed time....but he probably worth $4 million or better to the Sabres or somebody. Link to comment Share on other sites More sharing options...
SwampD Posted March 24, 2011 Report Share Posted March 24, 2011 It's a half past a monkey's ass, quarter to his balls. Link to comment Share on other sites More sharing options...
BuffalOhio Posted March 24, 2011 Report Share Posted March 24, 2011 Speaking of stats (of which I know nothing), did it piss anyone off that the Canadiens had 5 shots on their first power play and the Sabres had 1 on their first 2? Why can't Lindy just watch the Habs power play and implement it in Buffalo? How hard can that be? Their sucky power play is killing them! Oh, and Subban is really freaking good. Gerbe beat him for the EN goal because Gerbe had energy (always does!) and Subban was exhausted, but he's gonna be really good for a long time. Link to comment Share on other sites More sharing options...
nucci Posted March 24, 2011 Report Share Posted March 24, 2011 Speaking of stats (of which I know nothing), did it piss anyone off that the Canadiens had 5 shots on their first power play and the Sabres had 1 on their first 2? Why can't Lindy just watch the Habs power play and implement it in Buffalo? How hard can that be? Their sucky power play is killing them! Oh, and Subban is really freaking good. Gerbe beat him for the EN goal because Gerbe had energy (always does!) and Subban was exhausted, but he's gonna be really good for a long time. For that matter, why doesn't every team do it? Link to comment Share on other sites More sharing options...
PASabreFan Posted March 24, 2011 Report Share Posted March 24, 2011 Speaking of stats (of which I know nothing), did it piss anyone off that the Canadiens had 5 shots on their first power play and the Sabres had 1 on their first 2? Why can't Lindy just watch the Habs power play and implement it in Buffalo? How hard can that be? Their sucky power play is killing them! Oh, and Subban is really freaking good. Gerbe beat him for the EN goal because Gerbe had energy (always does!) and Subban was exhausted, but he's gonna be really good for a long time. I'm sure Lindy will spend some of Pegs' money on a top-notch power play coach. Link to comment Share on other sites More sharing options...
X. Benedict Posted March 24, 2011 Report Share Posted March 24, 2011 Speaking of stats (of which I know nothing), did it piss anyone off that the Canadiens had 5 shots on their first power play and the Sabres had 1 on their first 2? Why can't Lindy just watch the Habs power play and implement it in Buffalo? How hard can that be? Their sucky power play is killing them! Oh, and Subban is really freaking good. Gerbe beat him for the EN goal because Gerbe had energy (always does!) and Subban was exhausted, but he's gonna be really good for a long time. Montreal is streaky with the powerplay as well...they are going to miss Pacioretty. Montreal has 51 PPG.....Buffalo 48. Link to comment Share on other sites More sharing options...
wjag Posted March 24, 2011 Report Share Posted March 24, 2011 Montreal is streaky with the powerplay as well...they are going to miss Pacioretty. Montreal has 51 PPG.....Buffalo 48. Pittsburgh last I saw was 2-46. I've actually thought the Sabres PP has looked better recently (initial win on faceoff, zone presence, chances) it's just not finishing. I don't hate it this year so I guess that's an improvement. Link to comment Share on other sites More sharing options...
LabattBlue Posted March 24, 2011 Report Share Posted March 24, 2011 I'm sure Lindy will spend some of Pegs' money on a top-notch power play coach. Nope. He will just give McCutcheon a raise since he is doing such a great job. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.