Quote:
Originally Posted by Enoch Root
What? No he didn't. You're a doctor, right? So I assume you have a strong working knowledge of statistics. If so, you should be able to clearly see that he did no such thing. Not at all. Not even close. What he showed is that 3 really good teams also had good corsi numbers. Are there Stanley Cup winners with bad corsi numbers? (yes) Are there terrible teams with good corsi numbers? (yes). There is no strong correlation for him to show. There is a fairly weak correlation between winning and corsi, but there is no strong correlation between corsi and winning Stanley Cups.
As to your last sentence, I don't think you really believe that. A team 'on the right side of parity' but underperforming its potential, is not doing well enough in my books.
|
To your last point...how can you objectively measure potential again? Right.
Listen, I get that he didn't illustrate it without any doubt, but if you look at the data on the whole over multiple seasons, it becomes clear that it
usually indicates teams that should do well. The teams that do not fall into that correlation are called "outliers" and are common when you analyze something with correlation. Not EVERY data point is going to explain your correlation perfectly, because then it would just be causation. Confidence intervals indicate how reliable the data is within a few points either way, but you don't often get those posted as part of the analysis. That's what would indicate strong correlation, but there isn't enough data yet to determine that trend repeatedly year after year...not yet anyway. However, it does seem to be trending in that direction.
Stick to your guns if you wish. As I said, I don't necessarily disagree that I might see those things too, but I've often seen stuff that isn't supported by data, and I had to let go of my personal perceptions when I was proven wrong in the past.