Quote:
Originally Posted by Enoch Root
I don't know how much you use stats to explain things in your field, but those correlations are really low for anyone trying to explain results. It's just not enough to have any kind of confidence to draw conclusions from. 70% is really ambiguous for explanatory purposes, never mind 35%.
But again, it's not just the lack of clarity in the numbers, it's the conclusions people draw from them. If your performance sucks, and an analytical number that explains about 35% of the results is telling you that you're actually doing better than the results say, you should be looking for other explanations.
|
You know what? I've made a huge mistake.
The correlation with winning is in the second column so I don't get why Vollman said that. Maybe I misheard him or I don't remember properly. As you can see, I make a lot of mistakes.
The correlations for 5 v 5 CF%, 5 v 5 FF%, 5 v 5 SF% are actually between 52% and 54%. 5 v 5 PDO (sh% + save %) is 54.6%.
I think you have to add up the effects of all the stats to be able to predict success as well as possible.
In general, I think the best teams have a lot of things going well for them: great goaltending (save %); the ability to transition the puck quickly from the defensive zone and keep the puck in the offensive zone (CF%); the ability to actually get pucks through on net (SF%), generate chances (SC%) and high-danger chances (HDSC%); and capitalize on chances (sh%).
The individual stats are flawed. For example, just because you have the puck close to the net doesn't mean you actually got a great shot on net so HDSC% is flawed.
But if you add all them up, I think there's a good chance of predicting success.
I think its funny because if I wrote that paragraph about how the best teams have a lot of things going well for them and just didn't include the stats in the brackets, would you disagree with me?