Quote:
Originally Posted by Bingo
The Flames were a tire fire of a possession team relying on blocked shots to stay in games.
That wasn't a recipe for success.
His process had him look for coaches that could convince him that they had the ability and structure to coach an NHL hockey team to reverse the possession crater that had become the Flames.
He hired Gulutzan after many interviews, an odd hike, and then a 5 hour lunch meeting where they grilled him on the structure.
They liked it. They hired him.
He delivered on the possession ... as the Flames are 4th in CF% at 52.7% and 4th in scoring chances at HDCF% 54.3%. Those are great numbers.
I think Gulutzan whiffed on his assistant coaches as they special teams have let them down but the team is losing almost exclusively on finish (they are ranked 8th last at a shooting percentage of 7.1%.
I've said recently that the team's inability to respond to adversity may and perhaps should be Gulutzan's downfall, but the organization as a whole set out on a path to improve their play, and for most metrics they did.
|
This is where you (and many others) are led astray. You think there's a clear correlation between possession and winning hockey games. There isn't. If it were up to me statistics 101 would be a required course in high school so that we could shake this overfitting problem we see in modern advanced stats in hockey.
For example, I could make the case that Hartley's stretch passing created more odd man rushes and as a result correlates to higher shooting percentages. The numbers are clear, Hartley coached teams had significantly higher shooting percentages than Gulutzan coached ones. But I'd be mistaking a pattern in the noise for the signal. When I hear CF% thrown around, I just shake my head for the very same reason.
Hockey is an extremely difficult sport to model, and assuming you were successful at modeling it, your model would have a very short shelf life. The fact that CF% continues to be the de facto metric of choice after all these years is baffling to say the least.