ESPN’s Real Plus-Minus unveil on Monday was a major story in the sports analytics world. To celebrate, Sports Analytics Blog had its own special unveil: its first-ever roundtable post, featuring a back-and-forth discussion between founder Jonathan Gordon and contributor Jacob Rosen. We’ll also add to the ongoing RPM conversation with this interview below.
Kevin Hetrick is one of the smartest basketball guys out there. He’s been writing over at Cavs: The Blog (an ESPN True Hoop affiliate) for years, and I finally had the special opportunity to meet the fellow Cavs blogger before a game earlier this season.
Over the past several weeks, I started chatting with Kevin about his new project, gotbuckets.com. The website aimed to do something very similar to what ESPN just did yesterday: Add a bit more to the statistical conversation of basketball with adjusted plus-minus numbers.
Kevin's website includes regularly updated APM and RAPM leaderboards. It features a stat called SWAg (Summed Wins de-Aggregated) and a regularized version called SWAgR. That clever name is about as sexy as sports analytics gets.
After the Real Plus-Minus announcement, I reached out to Kevin with a few questions. He admitted that he’s not a RAPM expert of any kind, but just a guy that likes the stat and wanted to build a cool website. He now assuredly knows he’s not the only one with such interests.
JACOB: What are the benefits of looking at RAPM, in general?
KEVIN: Many things happen on a basketball court that are not reflected in the box score. Of the "catch-all" stats, RAPM better provides for a means to credit the traditionally non-tracked basketball activities.
JACOB: What are some things that will stand out immediately to the average fan in the Real Plus-Minus standings?
KEVIN: One of the primary items is that the defense splits "make sense" so much more than any of the other defensive stats available. Given the complexities of NBA defense, contributions at that end are very poorly reflected in the box score stats of defensive rebounds, steals and blocks. RAPM tends to give strong defensive players their due, and they will show up higher in overall rankings. Conversely, players that are one-sided towards offense are typically rewarded by box-score stats, but tend to show up lower on RAPM rankings.
JACOB: What do you see as ESPN's significant deviations to the formula to create RPM? Good or bad changes?
KEVIN: As noted in the introductory article at ESPN, the RPM is based on Engelmann's xRAPM which has been available at his site for a few years. I am sure the variations from more traditional RAPM to RPM are solid, but ESPN doesn't provide a lot of detail or information on the strength of the "out of sample" predictions.
JACOB: How do you think ESPN's investment in the RPM metric will affect GotBuckets?
KEVIN: Hopefully it helps shine a light on these type of stats and adds some credibility. Those were some of the primary purposes of gotbuckets. I have been a strong believer in these +/- regression based stats for several years.