garret9
AKA#VitoCorrelationi
The issue is that if Tkachuk scored a bit less he would likely fall into a much larger cohort that would also have a very high success percentage. That doesn't make his performance more "risky", it just makes it more rare. If he performed at a lower and less "rare" rate, he would automatically become less "risky" based on that criterion, which isn't very logical. Similarly, if you shave a couple of inches off of Logan Brown, he would likely fit into a larger cohort with a high success percentage. That doesn't make him more risky, just more rare.
The bottom line is that players that are performing beyond a stable and high percentage cohort are probably just as secure / stable as players within a lower cohort, and probably less risky, if anything.
Higher mean, lower confidence (ie: larger confidence intervals).
He may be less risky on average (by Grind's model), but there's more risk (by Grind's model) in whether the model describes him accurately.
I can't say much about my company, but I'm going to sneak one tidbit on why I like PLD so much:
* I just like his game more, his style, and his fit on the Jets
* I just like his scoring numbers better when adjusting for factors (age, team, linemates, secondary assists, etc)
* AND (this stays here) best relCorsi of all draft eligible players, AINEC