- Sep 28, 2014
- 76,751
- 123,307
And in relation to the stats thing... all those guys do great work.
But I always feel that working in definites in this world of ours is both commendable in terms of confidence (especially when know going to get backlash)... but often can come across as idealistic and self-righteous. (I say that as someone who has both of those attributes ingrained in my personality...) But the world is a thousand shades of grey and not black and white. That is the same with numbers.
Underlying stats are a great tool. Everyone here knows I utilise them and enjoy delving into them.
But there are a lot of variables in hockey, and if you are running correlation coefficients, linear regressions etc on that many moving parts... then a fantastic result is that your model accounts for 50%+ of variance off the top of my head.
Having studied that area but in relation to crime, yeh, things are a great "indicator"... and you will for sure realise things that you probably never thought of when plug them into a model... and also that the stats are a better overall indicator than one set of eyes could be... but that sometimes for specific data points a different model would actually work better.
For example I have never believed it is sheer coincidence that when you combine: High QoC with below 40-45% zone starts the result is it being basically impossible to get over 50% Corsi and ExGF... and that every player who has played such minutes suffers. No model really accounts for that right now, as they look at the bigger picture, and over big samples of a lot of players in different situations... yeh... both zone starts and QoC are not "that" important. In turn I would guess a lot of models "unfairly" deflate those players results while inflating those with joke minutes. Yeh, you are talking maybe ~30 players a year who will be affected by that over a model (and overall such models are great, predictive etc)... but speaking in definites when there are areas that are more nuanced.
And with that: Draisaitl should almost certainly not be in the Hart race or particularly close. He is a net positive and amazing offensively but McDavid is better on the same team... and one of the worst forwards in the NHL inside his own half of ice.
This is perfectly stated. Don't be an absolutist and you won't get as much backlash and you will actually do a better job at increasing interest in the topic.
Also, you brought up a great example with the QoC + Zone Starts. According to Micah and Evolving Wild QoC is not important due to QoT and neither are zone starts due to how many shifts start and end on the fly.
And regarding Drai, i'm positive he will get nominated for the Hart even though he doesn't deserve to be. Which also shows that Hockey Guys have a long way to go as well.