JojoTheWhale
CORN BOY
- May 22, 2008
- 33,789
- 105,380
(since Provorov had a higher GA, he couldn't be padding MacDonald's numbers).
You don't understand context or the point of Rated or Expected metrics. At all. It's actually stunning how many ways this methodology is flawed. This might be your masterpiece.
Even if we toss out the improvement in Provorov from the beginning of the year to the end (which would be insane), you're starting with the premise that it would make sense to examine Provorov's numbers away from MacDonald and ignore the rest. You've taken a sample size of ~540 minutes and attempted to use it to prove a point while ignoring a sample size of ~850 for....well I have no idea why.
You've also stunningly managed to ignore ice time completely. Of MacDonald's 10 minute+ 5v5 pairings over the same 2 year sample size, only AMac-Gudas for 80 minutes and AMac-Streit for 30 have a lower GA/60 than AMac-Provorov, with the latter being a virtual tie. To be clear, of the 5v5 minutes MacDonald has played in the last 2 years away from Provorov, ~128 minutes were spent at a lower raw GA/60 while ~730 were at a higher rate.
Then there's the part where you forget defense partners exist. Away from each other, Provorov spent ~369 minutes with Brandon Manning and Mark Streit's embalmed corpse. When paired with an NHL defenseman, albeit only in ~170 minutes, every single GA/60 was better away from MacDonald. That's not enough sample size to draw on to make any sort of conclusion, but it's in line with the backwards thinking you started with, so hey, when in Rome. Meanwhile, ~600 of MacDonald's minutes away from Provorov came with Gostisbehere. Next most common were Gudas for 80 and MDZ for 60.
What happens when you compare GA/60 rates with common partners? Let's stick to 20 minutes or more in an attempt to not be completely out to lunch. Format is Partner: Provorov / AMac.
Gostisbehere: 1.25 / 1.97
MDZ: 1.94 / 2.89
Gudas: 1.57 / 1.49
Manning: 2.39 / 2.89
Streit: 3.35 / 1.93
So what those numbers are actually telling you is that Provorov/Streit got bombed early in the year. That's it. Context. To take it a step further, I'll give you one guess what happens to those Streit numbers if you take out the Chicago game. Sample sizes are fun.
This bothers me because it's the type of throw crap at the wall attempt at analysis that gives analytics a bad name. At least care enough to understand what the numbers you're quoting are attempting to explain and what they are not. Don't just open up a hockey-ref tab and vomit out nonsense.