Quote:
Originally Posted by Bingo
It will be interesting to see how this contract works out.
I've seen the argument about coaching style, and I've seen the argument about missing Klefbom - reason for Koskinen's decline of late.
So here's a table ...
Code:
Row Labels Count of Date Saves Save% CA/Game SCA/Game HDA/Game
McLellan 8 211 0.913 56.8 28.8 11.1
Healthy 8 211 0.913 56.8 28.8 11.1
Hitchcock 20 505 0.908 55.7 27.2 12.4
Healthy 7 196 0.942 59.1 28.9 12.9
Klefbom 13 309 0.888 53.8 26.2 12.1
Grand Total 28 716 0.910 56.0 27.6 12.0
In chapters ...
Chapter One - McLellan as coach - .913 save percentage
Chapter Two - Hitchcock with Klefbom healthy - .942 save percentage
Chapter Three - Hitchcock Klefbom hurt - .888 save percentage
The assumption ... Klefbom out means Oilers are no good and it's not Koskinen's fault.
But look at the team metrics at the same time.
Chapter One - McLellan as coach - CA 56.8 SCA 28.8 HDCA 11.1
Chapter Two - Hitchcock with Klefbom healthy - CA 59.1 SCA 28.9 HDCA 12.9
Chapter Three - Hitchcock Klefbom hurt - CA 53.8 SCA 26.2 HDCA 12.1
The Oilers have given up less in terms of shot attempts, scoring chances and high danger chances without Klefbom but Koskinen has been shredded just the same. |
I’ll add one more layer. In the statistical / research world, absolutely all results are accompanied by margins of error. It’s a little more mathy and not easily digestible for the general hockey fan base. But if you are making hockey decisions based on results, you can’t just compare two numbers without considering ‘what if this is just random chance’?
Some things you need for accurate estimates are large differences and large sample size. In this example, we do have a relatively large difference in estimates (0.942 vs 0.888). We do have a relatively huge difference in context of hockey goalies, but a difference of 0.054 in a test of proportions is quite small. This is where the number of games played (more specifically, number of shots faced) comes into consideration.its not even close to reaching statistical significance. Plus, we know that there are plenty of other factors beyond Hotch an Klefblom that we need to adjust for (score effects, shot danger, etc). There simply isn’t a body of evidence to make a claim based on in-game statistical evidence.
That leaves us with hockey sense. For example, if a PeeWee goalie was thrown into an NHL game and let in the first shot of the game, while there isn’t enough body of evidence to ‘statistically’ predict that he will continue to be lit up, we all can see that he’s not going to be a good goaltender. That’s because of factors that we can appreciate, but that the numbers themselves might not.
So without statistical backing, the Oilers are either A) misinterpreting basic statistics or B) using their hockey sense to make these decisions. Either option is very reassuring to all Flames fans!