On Real Plus Minus: Pieces, Pieces Everywhere

Last Monday, ESPN took basketball media in its firm and assertive hands and shook mightily; rolling out Real Plus-Minus (RPM). This metric, presented by Steve Ilardi and Jeremias Englemann, is an extension of their previous work on Regularized Adjusted Plus-Minus (RAPM) and this new iteration is now proprietary to ESPN.

The initial explanatory article, and several accompanying articles using the metric for some entry-level analysis, were a huge infusion of gasoline reigniting all the your old, familiar tire fires. I spent a big chunk of Monday watching straw men meeting their unfortunately flammable ends up and down my Twitter timeline. It was, quite literally, a remastered release of greatest hits from the super-group Analytic Arguments. There were the disagreements between proponents of plus-minus based statistical models and those who prefer systems derived directly from counted box score statistics. There were even feathers ruffled among adherents of different plus-minus based statistical models, some who found fault in the particular quirks of ESPN’s new system. And there was, of course, plenty of shouting about the role of analytics in general and what it means for The World Wide Leader to be pushing themselves to the theoretical cutting edge.

I let this wash over me all week but couldn’t muster much more than apathy.

I’m not here to critique RPM, or of ESPN’s investment in it. RPM appears, at a rudimentary glance and from the parties involved, to be a splendid refinement of the existing plus-minus models. ESPN’s involvement lends credibility to not just this particular school of statistical models, but to sports analytics in general and to the idea that we should always be looking for new and better ways to understand basketball. Whether or not RPM is the greatest basketball stat known to man is of far less importance than the fact that it appears to be a very good one which will help create more educated basketball fans. By being presented to such a large audience, with the considerable media might of ESPN, should help raise all ships in the analytics arena, which is, to my mind, a very good thing.

But even as someone with plenty of familiarity with, and curiosity about, basketball statistics, RPM didn’t spark much of anything for me.

My issue with RPM, and really all of the various plus/minus models, is that they are increasingly complex methods for stripping away the context of a player’s production, trying to measure it in a vacuum. It’s an admirable pursuit to some degree and these intricately designed techniques have become, in many ways, the basketball analytics arms race. The problem is that I’m just not that interested in the result. The context and the noise, which these models work so hard to control for, are exactly the things I’m interested in. I don’t just want to know which player is better. I want to know why and in what ways. I want to know what that implies about both the player and team, his teammates and opponents, and basketball as a whole. As constructed and presented, I typically find precious little of that information in plus-minus statistics.

This problem is not unique to RPM, or even to the entire family of plus/minus models. Win Shares, Wins Produced, PER, also chase the same goal–generalizing the “why” to highlight the “what.” But the “why” is what I find most interesting, the “why” is the reason I watch and write about basketball.

The why is not permanently removed from RPM. There are ways to combine it with other statistics, to manipulate and array other metrics around it to create a more complete picture of what is happening on a basketball court. And ultimately that is the point. There is no single basketball metric for me which feels indisputably complete. I don’t want one number to measure basketball production. I want the nuance, the details and the nitty-gritty. I want all the numbers.

A new attempt at refining measures of player production is exciting, and seeing it being developed and promoted by the biggest of sports media entities, ESPN, is a hugely important step forward. But I would hate for the implication to be that basketball analytics is a funnel, with everything being drawn down to a single, most accurate number for representing player performance. I prefer to see analytics as a delta, with creativity and curiosity at the mouth and techniques, statistics and questions fanning out in all directions.

So by all means, please look at RPM. Read the articles that have already been written about it and continue to read those written in the days to come. Learn as much as you can about how it is calculated and about the information it imparts. But please, don’t stop there.

Ian Levy

Ian Levy (@HickoryHigh) is a Senior NBA Editor for FanSided and the Editor-in-Chief of the Hardwood Paroxysm Basketball Network.

  • thecity2

    “My issue with RPM, and really all of the various plus/minus models, is that they are increasingly complex methods for stripping away the context of a player’s production, trying to measure it in a vacuum.”

    I take issue with the last part. It’s exactly because most other box-score stats are measured in a vacuum, that RPM (RAPM and its ancestors) actually do try to account for the context that is most often missing.

    “I don’t just want to know which player is better.”

    Ok you don’t “just” want to know that. But you do want to know it, right? I mean, shouldn’t we start there?

    • I really appreciate the comments Evan. On a reread I realize I’ve grossly mischaracterized the way RPM and it’s predecessors handles this context. While other non-plus-minus systems try to strip it away, this is an attempt to get it all in. One is a vacuum the other is throwing everything in the pot. My point is that in both cases, for me, the why is difficult to parse out from the end result.

      To your second point, I do want to know which is player. But the nature of my own curiosity means that I don’t always want to start there. Sometimes I want to start there. Sometimes I want to finish there. Sometimes it is a mid-point or a fork in the road on my way to some place else. Like I said in the piece my concern with a single metric for assessing player value is that (probably, mostly, unintentional) implication that this single number is the only, or even the most important thing to know about a player.

      One of the things that I know sets me apart from a lot of the basketball analytics community is that I don’t always approach things from an evaluative standpoint. To me RPM is, at it’s core, an evaluative metric. It’s dominant function is to reveal how players compare to one another in overall productive value. But I feel like I often approach analytics from a descriptive or narrative standpoint. That final characterization of good/bad and to what degree is often of secondary interest for me.

  • Pingback: GotBuckets | SportsVU and Rim Protection()