Hollinger's Kevin Martin Projection Has Befuddled Me

[This post is long and blustery. A shorter, less blustery statement is also available.]

I'm a big fan of John Hollinger. I use his terms, his metrics, his philosophies, his analytical style. As close as I'd associate myself to any School of NBA Thought, it's the Hollinger school. Anyone who reads me here or at FanHouse knows this. Know this: I appreciate Hollinger's methods.

Hollinger, in his '08-09 projections, figures a drop in Kevin Martin's production is coming, from a PER of 21.07 in '07-08 to 19.98.

Is it because Martin's shooting will suffer under the weight increased shot attempts? I'd understand that -- there's a general expectation that forcing increased shots on a player decreases the player's efficiency.

But Hollinger projects Martin's already sterling True Shooting percentage (61.8%) to get even better (a scandalous 64.2%).

Is it because Martin will tally more turnovers and fewer assists registering as the sole focal point of the offense? That might make sense.

Hollinger projects Martin will earn more assists per possession and fewer turnovers.

So how on Earth will Kevin Martin get worse (according to PER) in his fifth season?

He'll ... ahem ... he'll shoot less.

Hollinger's system for projection is built on skill curves, age curves and (mostly) similarity scores. So apparently some statistically similar player (Reggie Miller? Rip Hamilton?) saw their usage rates go down at age 26. And this, apparently, has more relevance than the fact Martin has been knighted the sole offensive focal point and the team's other '07-08 weapon was traded for a 20-year-old and a back-up PG and the coach, GM and players in Sacramento have all repeatedly referred to Martin as the no. 1 option.

I mean, it's not like Hollinger doesn't know this stuff. In the profile, he writes:

Martin is 25, he's improved every year, and he's about to take on a much larger offensive role (um, right Reggie?) now that Artest and Mike Bibby have been traded. All signs point to a big scoring year from Martin, and given how efficient he's been percentage-wise it could translate into his first-ever All-Star appearance.

[Emphasis mine.] How does that acknowledged fact square with Hollinger's projection that Martin's usage will fall? It doesn't. And while it's admirable that Hollinger apparently keeps all subjectivity out of his projections, isn't all the stuff about taking on a greater share of the offense really a rather important factor? Like, if Gregg Popovich said "Bruce Bowen will not take more than 2 FGAs per game this season, or else he will feel 5,000 volts of electricity course through his body" ... wouldn't that be an important factor you'd want to take into account if you were projecting how many points Bowen would score per game?

All of this isn't to bash Hollinger -- again, I'm a fan -- but to point out the problems with pure objectivity. Hollinger has some facts which could improve the accuracy of one of the major factors in his formula, but he cannot use those facts because they aren't data. You can't use regression analysis to predict how many minutes per game or how many shots a player will take. Well ... you can. But it's not going to be more accurate than common sense.

This post shorter and less blustery: Hollinger has projected the Kings will win 23 games. He has also projected Kevin Martin will be a smaller part of the offense this season. Take John Hollinger's projections with a grain of salt.

UPDATE: Hollinger also projects that literally every single King who played in the NBA last season will become a worse rebounder ... in some cases (Hawes, Martin, Miller) a substantially worse rebounder. The Kings finished #22 and #27 in offensive and defensive rebounding percentage last season. I'm eager to either a) see how much they can possibly get, which is what Hollinger expects, or b) watch Jason Thompson average 25 rebounds a game.