The issue with Smart's CARMELO projection, which has not been emphasized enough, is the variance.
That's capture by the gray shaded "confidence interval" region on the projection.
Take a look at Smart's - it's huge:
https://projects.fivethirtyeight.com/carmelo/marcus-smart/
Smart's confidence interval goes from 0 to 10 WAR. For those counting at home, that's the difference between Nik Stauskas and Kevin Durant.
Now, if you take other players of similar vintage, their ranges are smaller and in many cases much, much smaller
https://projects.fivethirtyeight.com/carmelo/dante-exum/
https://projects.fivethirtyeight.com/carmelo/kelly-olynyk/
https://projects.fivethirtyeight.com/carmelo/jabari-parker/
There are even much more recent draftees with less uncertainty according to these projections:
https://projects.fivethirtyeight.com/carmelo/brandon-ingram/
Why the difference? I'm not sure, but I would guess that the most of the range for Smart is a function of possible variation in his future offensive capabilities, and in particular whether he becomes a plus 3 pt shooter or remains atrocious.
tl;dr: This tells us nothing we don't know.
When people were into CARMELO last year I tried to caution them about variance. That said, while you can certainly find players with less variance, generally the higher the mid-point (which is essentially the numerical value they report) the greater the variance. This is largely due to the fact that CARMELO uses modeling based on past performance from "similar" players. The better the player, the fewer comparable players from the past, so the wider the error bars.
Smart is an interesting case -- he has a lot of comparable players up until this point in his career (with similarity scores near and above 50), but their career paths after this year are widely divergent. You've got a couple of guys who peak around All-NBA status (Gus Williams and Mike Conley), a few players with a few years of sustained production around his current level, and a bunch of players for whom this is as good as it gets, and it's time to sell now. (The model uses more than the top 10 comparables shown, of course, but weights each subsequent player less based on similarity.)
Some of it certainly has to do with shooting, but most of Smart's top comparablea were not known for shooting, and were much more known for a combination of defense and traditional point guard skills. If Smart can continue to develop his offensive playmaking, without including his shooting, while maintaining his defense, he has a chance to reach the higher portion of those error bars. If he adds shooting, he's the top of those bars. On the other hand, if his physical defense takes a toll on him, he's going to be on the lower end.
A lot of people the past couple years have compared him to Dennis Johnson, and if he pretty much just stays at his current level for a decade, that looks like a really good comp.
Topic: The next Celtics Super Star......."Marcus Smart" (Read 4601 times)