Archive for the 'Statistical +/-' Category

This has been a surprisingly popular series so far, so because of reader demand I'm going to accelerate things and go ahead with the Top 10 "statistical +/-" shooting guards in NBA history (excluding seasons prior to 1951-52, when they didn't bother to track minutes). As a quick refresher, SPM is a linear regression formula that tries to predict the well-known adjusted plus-minus stat using just the conventional stats you’d find in the box score. Obviously some defensive value is going to be lost as a result, but so far the results haven't been horrible, so that's encouraging. Anyway, here are the best SGs by the method, in alphabetical order:

Read the rest of this entry »

This is a series we've been sporadically doing this month, and I figured today was as good a day as any to keep it going and check out the top 10 NBA/ABA small forwards since 1952 according to the “statistical plus-minus” method. To refresh everyone's memory, SPM is really just a linear regression formula that tries to predict adjusted plus-minus using just the conventional stats you’d find in the box score. It has its own biases like any boxscore metric, but I kinda like it, and it's pretty simple to use and understand -- the numbers that follow are estimation of a player's individual impact on his team's point differential per 100 possessions. So here’s what it has to say about the top 10 small forwards ever, in alphabetical order:

Read the rest of this entry »

Continuing our series from last week, today we're going to look at the top 10 power forwards ever by the "statistical plus-minus" method. If you don't remember what it's all about, it's basically a linear regression formula that tries to predict adjusted plus-minus using just the conventional stats you'd find in the box score. I don't think it's the ideal player rating metric or anything, but at the same time it doesn't seem to be the worst I've ever seen, either, so we're going to keep giving it a test drive by using it to rank the all-time NBA (& ABA, forgot to make that completely clear last time) players at each position. Here's what it has to say about the top 10 power forwards ever -- again, in alphabetical order:

Read the rest of this entry »

We've done several posts on statistical +/- here at the BBR blog over the past month, and it's mainly because I don't know what to make of the metric. I suppose that deep down, I very much want it to be a good, solid linear-weights method of player rating, because there's not really any fudging involved in the original regression -- it simply asks which stats best predict adjusted +/-, which itself is a method that feels "organic" to me (increasing your team's point differential being literally the purpose of the game, after all). No guesswork, no worries over how to deal with assists, defensive rebounds, the value of shot creation, or any of the usual potholes we run into when developing one of these baseball-style metrics for a sport that doesn't really lend itself to that kind of thing.

Read the rest of this entry »

The other day, I talked at some length about "statistical plus/minus," which is just a regression of pure adjusted +/- on the conventional boxscore stats. In that post, I looked into the possibility of predicting the following season using a weighted average of the 3 previous seasons' SPM scores, but I realize that I sort of skimmed over the statistical +/- metric itself -- what are its strengths and weaknesses? What kind of players does it overrate and underrate?

Read the rest of this entry »

Surely most of our readers have heard about "adjusted plus/minus" at some time or another, whether through TrueHoop, Sports Illustrated, the APBRmetrics board, the many articles at 82games on the subject, or even Michael Lewis' article about Shane Battier in last week's New York Times Magazine. But for those of you who aren't familiar with the stat, it essentially tracks a player's influence on his team's point differential by comparing the team's performance when he's on the court vs. its performance with him sitting on the bench. There are also adjustments for a player's teammates, backups, opponents, and even the location of each game, and the whole mountain of data is fed through a massive linear regression equation to try and isolate the individual impact of every player. Personally, I believe this system has a lot of promise (especially with regard to measuring defense), and with further refinement it will one day be one of the better basketball metrics out there, so it was nice to see it get some more recognition this past week -- even though there are still some wrinkles that need to be ironed out.

Read the rest of this entry »