Are you implying JFresh is an expert?
I used to think these dudes were all know it all dummies that couldn't form an opinion against what the stats said to save their life.
Now that I've had the chance to work and learn in a more corporate environment at a fortune 500 company doing problem solving for different types of problems...
Most of the "self made" analytic gurus are extremely advanced and effective problem solvers beyond. Far beyond what our business would consider normal.
And over the years the data has gotten more tested compared to 10 years ago. And there seems to be a more general understanding of the limitations.
Point I'm trying to make is if NHL clubs are like the environment I'm familiar with, staffed by smart, regular people that do a good job, make a good living and go home to their families... these online guys are likely just as good and could be even better as whole analytic departments working for teams.
Coaches aren't statisticians nor can they be predictable robots. To me, you can get a pretty good idea of what a player is made up of and how they can best be utilized from these profiles. You can lump them into buckets accurately and once you are within 10-20% it's too close to call.
If a guy is bottom 10% in these models it's because they suck and not because the data is wrong. If the data is off its likely they could be bottom 30% in which case they still kinda suck....
That's my take anyway, it's so complicated and nuanced a topic it will never get consensus anyway.
Are you implying JFresh is an expert?
I used to think these dudes were all know it all dummies that couldn't form an opinion against what the stats said to save their life.
Now that I've had the chance to work and learn in a more corporate environment at a fortune 500 company doing problem solving for different types of problems...
Most of the "self made" analytic gurus are extremely advanced and effective problem solvers beyond. Far beyond what our business would consider normal.
And over the years the data has gotten more tested compared to 10 years ago. And there seems to be a more general understanding of the limitations.
Point I'm trying to make is if NHL clubs are like the environment I'm familiar with, staffed by smart, regular people that do a good job, make a good living and go home to their families... these online guys are likely just as good and could be even better as whole analytic departments working for teams.
Coaches aren't statisticians nor can they be predictable robots. To me, you can get a pretty good idea of what a player is made up of and how they can best be utilized from these profiles. You can lump them into buckets accurately and once you are within 10-20% it's too close to call.
If a guy is bottom 10% in these models it's because they suck and not because the data is wrong. If the data is off its likely they could be bottom 30% in which case they still kinda suck....
That's my take anyway, it's so complicated and nuanced a topic it will never get consensus anyway.