Quote:
Originally Posted by edslunch
Sorry, I misread your example, agreed it can't be neither. I still don't see that as a shorthand it's very useful. A good team will tend to be above 100 a bad team will tend to be below, but neither says why. True extreme values can indicate a non-sustainable situation but there should be no expectation that numbers revert to mean. A team with a great goalie has an automatic bump on PDO.
|
Well, we can argue about whether there's even such a thing as a great goalie, but what I mean by a shorthand is that you can quickly glance a league table in PDO and see, "okay, these two teams are at 1050+ and this team that's chasing them is a 977, that probably means they're going to see some bounces even out over the next 30 games". You could have, for example, used this as a basis to predict that the Lightning were still going to have a shot at the Atlantic, when there were a bunch of stories about how they had fallen apart after a first round sweep and their captain was criticizing their play style. If you actually wanted to do any sort of in depth analysis of a team's play I agree you'd discard it pretty quickly.
Quote:
|
Even worse IMO is individual PDO. Player X has a PDO of 99 - what does that mean? They don't control save percentage and it's not even a good indicator of defensive performance, which is anyway masked by adding in smaller sample shot %. Garbage stat.
|
Agreed. Although individual stats of any kind are somewhat tough to eliminate the noise from. Which is why we now have RelTM, the most sophisticated attempt so far to deal with all of those effects.