FiveThirtyEight blog -- Utah Jazz *Carmelo player projections

[franklin]

 

[cjs55]

It seems like they are saying all our players basically played to the peak of their confidence intervals this past year and that probably won't happen again.

Not really, most of the confidence intervals (90/10 CIs) in later years go higher than where they were last year, which means there is a good chance they will have a few years in the future better than last year. It's just that the projection system expects some regression to the mean when you make a big jump in one year, and it does so because that's what best matches what other players have done in similar situations. One player might break out (although Gobert most likely), one player might regress, and that possibility is contained in the CIs. You also have to take into account how much age matters. 27 being the peak for players is a big deal, which is why you see Gordon and Favors' projections linearly drop off after they turn 27.

 

[gandalfe]

Almost every player in the NBA, they have getting worse. Notice that there's a range and the dot is in the middle of the range - so that assumes they COULD get better, they COULD get worse, and this just shows the middle of the road for that player. Since this is the first year, there's no way yet to see how accurate they are.

<EDIT> What cjs said.