I remember TCG mentioning that he was a Roy fan.
He posts on this forum from time-to-time, so perhaps he could clarify.
I am the Contrarian Goaltender, I've been a longtime lurker at the HOH boards but I'll take up the invitation to clarify a few points. I have also changed my user name for the sake of clarity.
For those interested in my background, I am a Habs fan, Patrick Roy was my favourite player growing up, and on the day I started my blog I probably would have told you he was the best goalie ever. My level of admiration for Dominik Hasek is entirely from watching him in action and from being repeatedly blown away when digging into his statistical record.
TCG is notorious for selectively picking his starting points before going into his long statistical analysis. He's not a neutral analyst at all. He's just as biased as anyone, he just covers his biases in statistical smoke. I'm surprised it isn't more well known by now.
I'm not sure what you mean by "selectively", but yes I do define the parameters before I do an analysis. I'm not sure why that would be considered "notorious". Define the criteria beforehand, run the numbers, and post the results, that's what I do, that's what people in this forum like Hockey Outsider, Overpass and Seventieslord do and that's what anyone should do when working with statistics. That is the opposite of cherry-picking or selectively choosing stats, where one is working with a pre-determined conclusion in mind and looking for evidence to help them get there.
I think it is fair to say that I have some biases, for the same reasons that it's fair to say that everyone has some biases. We all have our favorites, our blind spots, our knowledge gaps, our selective memories. I do my best not to let those things affect what I write, but I can't guarantee that they don't do so anyway. That's why I believe so strongly in the use of statistics evaluated in context. Stats are limited in what they can capture and need to be adjusted for certain flaws, but the advantage is that numbers provide a level of objectivity that is unavailable to essentially all human observers.
If someone makes a statistical argument, anyone else can replicate the study to verify it, or can introduce evidence of their own to disprove it. Like anyone I rely on feedback to keep me honest, I've been proven wrong before and I'll be proven wrong again. If I made a mistake in the numbers or you feel the model was flawed, then by all means point it out. But the numbers are what they are, whether you think I'm biased or not.
Has he been forced to acknowledge the routine undercounting of shots at the old arena in the swamp yet? I remember when an article on puck prospectus first raised the issue, he dismissed it as unimportant.
I never said it was unimportant, I said it required further study to account for all potential variables. I agree there is good evidence of undercounting of shots in New Jersey that cost Brodeur some points off of his save percentage.
The amount of mental pressure a goalie is facing is entirely different in a regular season game and in the 3rd period of a close playoff game.
I find it interesting that after criticizing me for my definition of clutch, you use exactly the same definition yourself when giving an example of a pressure situation.
The point of the Nieuwendyk article was that GWG is a poor and arbitrary definition of clutch. Pick any definition that you want of cluch play that is more expansive than GWG, and it's going to be very close between Modano and Nieuwendyk in '99.
I chose the "3rd period and OT of a close game" definition I used because I thought that was the type of thing that might be convincing to a media person or hockey fan that equates GWG with clutch play. I personally give much, much greater weight to the overall numbers than to any smaller sample, because I agree with Cognition's argument that most of the game is a clutch situation since most of the game is played with the score close.
