Why are RPM projections low on the Clippers?

The Clippers’ 53.5-54 O/U set by Vegas seems laughably low to most fans, but to stats-reliant prognosticators and RPM-heavy predictive models, the near-unanimous verdict is that it’s set too high.

FiveThirtyEight has the Clippers at 48 wins (5th in the West), while ESPN’s RPM model puts them at about 46 (4th). Kosta Medvedovsky (who came second in last season’s APBRmetrics prediction contest) had the Clippers at 51 wins (3rd). Here’s where some other models place them, via Nylon Calculus:

Andrew Johnson’s: 53 (4th, 3rd West). A combination of PT-PM (a SportVU player tracking metric) blended with RAPM. Two-time reigning champion of the APBRmetrics board predictions contest.

Nick Restifo’s: 52 (T-4th, T-3rd West). A simulation using BPM and RPM for player value, which includes game effects like altitude and rest.

Kevin Ferrigan’s: 50 (T-6th, 4th West). A player projection system with inputs from RAPM, BPM, height, and age.

Many of these come from extremely reputable individuals with proven track records, so it should be a little bit disconcerting to see that the consensus from the analytics community is low on the Clippers this year. What’s going on here?

Justin Willard explains some of this in his Clippers preview for Nylon Calculus:

Numbers-based projections are consistently under the Vegas expected win total for the Clippers, and it’s not easy to see why at first. Chris Paul, JJ Redick, Blake Griffin, and DeAndre Jordan are arguably the best quartet outside of Golden State, but age is a negative factor for two of them and neutral for the others...

Age alone isn’t enough to push the Clippers down as much as some projections see, so there’s a missing component here few are considering: the Clippers lost Cole Aldrich, and two of the most popular metrics for prediction, RPM and BPM, adored him. By VORP, which measures total production from BPM, he was the fourth most productive player, even though he played just 800 minutes. Per 36 minutes, he averaged 3.1 steals and 2.1 blocks per game. He alone doesn’t dramatically change LA’s projection, but it’s another effect that must be accounted.

Although you could argue that both CP3 and Redick have games that will age well, statistical projections looking at average age curves don’t make case-by-case judgments. It’s fair to expect both guys to regress after coming off great seasons, and Father Time comes for us all eventually.

Blake Griffin’s injury history the last two years also weighs heavier in models like FiveThirtyEight’s CARMELO, and he’s never done as well in RPM as Chris Paul or DeAndre Jordan. That metric also undersells Redick’s all-around contributions.

Aldrich’s presence off the bench will be sorely missed, but are all of these factors enough to knock a healthy Clippers team down to 50 wins or less? It seems unlikely. Last year’s Clippers were projected for around 53 games if healthy, and their performance without Blake Griffin likely exceeded even the the rosiest of projections. For what it’s worth, guys like Kevin Pelton disagree with their models when it comes to the Clippers.

While none of these models claim to be perfect, it’s unwise to dismiss them out of hand, especially if they’re all drawing similar conclusions. RPM and its ilk have underestimated the Clippers in the past, but in the aggregate they tend to be more accurate than Vegas and any layman you know.