I'm not using shooting percentage alone; McDavid beats him in shooting percent above expected by every model I've looked at.
Fair enough. But, if a player takes a higher percentage of shots from areas that he's good at, would it not skew the numbers? McDavid seems to wait to shoot until he's in close the majority of the time, and he has a quick, accurate release in tight. I'd say he's better than MacKinnon close to the net (although part of that is moves on the goalie rather than just shot), but MacKinnon is much better 5-10 feet out and beyond. It's subjective then which shot we would consider better in a vacuum (though I would argue traditionally shot quality tends to be judged from a bit further out), but McDavid would look better on paper in terms of shooting percentage/expected shooting percentage, because he shoots a greater percentage of his shots in the area he's best in, which also has a higher percentage of going in.
As an example, if the average player can score on 10% of their shots in close and 5% on their shots from further away, I can score at 25% on my shots in tight, and 5% on my shots further away, and Joe can score at 20% in tight and 10% of his shots further away, if I take 100 shots in tight and 100 shots further away, I'd have 30 goals and a 15% shooting percentage, and an expected shooting percentage of 7.5%. If Joe took 50 shots in tight and 200 shots further away, he'd have 30 goals and a 12% shooting percentage and an expected shooting percentage of 6%. So I would have a better shooting percentage above expected in terms of amount of percentage above expected (7.5 to 6.0), and the same if we're looking at percentage above expected (both twice the expected percentage), but am I really the better shooter? Or even the same? Joe is 80% as good as me in close but twice as good as me from further out. I would say Joe is the better shooter.
Their shot charts seem to support this idea of where they're shooting and scoring:
Also, I'm kind of curious what numbers these are, because just a brief look at some sites doesn't suggest this for me. Looking at evolving hockey's xFSh%, we see McDavid with %10.25 and MacKinnon with 6.96%, and their actual FSh%s are 12.22 and 9.16 respectively. This would put MacKinnon with better numbers relative to expected.
Also, according to NaturalStatTrick, over the past three years, MacKinnon has an individual expected goals of 81.06 and McDavid has an individual expected goals of 85.51. Evolving hockey puts them at 87.3 for MacKinnon and 97.3 for McDavid. Considering McDavid only has one more goal than MacKinnon over that time period (116 to 115), would it not follow that MacKinnon scoring a greater amount/percentage over his expected goals suggest he was the better shooter?