Quote:
Originally Posted by SebC
Fine. Post something and I'll evaluate it. Point is, I'm being criticized for having "an argument based on nothing". That can't be true, as the opposing arguments are based on even less.
|
Can't be bothered. I have read all of this stuff. I know what data I would like to see available as inputs for a model and I know that nobody has measured it adequately. Again, garbage in garbage out, and these guys generally conclude that phenomena are not statistically significant, based on the wrong input data.
Consider the 2 on 1 where Bouw pinches and Brodie overplays. All these guys have to go on is shot location. No two shots from the same location are alike. And the attempt to correlate to available data means stuff like game situations such as score muddy the issue and add further noise. No wonder these guys never calculate that anything really has influence. It is incredibly limited and flawed.
Reality is that to propose a model you would have to watch way more hockey to compile your own data because nobody measures data that captures situations well enough. There are always multiple layers and inferences.
So no. Given the data available is unsuitable to generate a meaningful model I'm not writing a thesis for you.