APBRmetrics

Several people have asked me why so few SGs have positive RPM estimates (just 15) and 3/4ths of centers do so far this season. That is a question that could provoke several kinds of inquiries. Is it an imbalance in the talent pool? Are SGs trying to do too much on offense and too little on defense?

Is it typical? Last season almost twice as many SGs ended up with positive RPM estimates as thus season while less than half the centers did then, so it looks like a pretty major shift yr to yr. Don't know the longer run trends on RAPM.

My first reaction / attempt to possibly explain was: are teams going to "small ball" when behind and is it not working on average? That hypothesis could fit the above data.

I don't think I've seen anyone define small ball carefully or look at league or team level performance of lineups meeting one or another definition. Probably could be a good article or possibly Sloan Conference paper on that.

Are teams / coaches impressed with the highly visible success stories and copying a lot and possibly missing a negative overall lean? Maybe. Eyes can't sort and sum thousands of lineups and tens of thousands of tiny stints.

Could also look at "playing fast" and when small and fast overlap and when they don't.

Small ball is the label used for playing 2 PGs, playing 3 Guards, playing a PF at center or a SF at PF. Could you define small ball by a sortable rule such as anytime a lineup includes a guy at a position 2 plus inches below position average or 15 pounds? Or base it on their typical position assignment to accept their personal utilization? Do folks want to look at the results in one sum, 4 piles or both? I guess sometimes teams are small by more than one way. Is super small an adequate catch-all for these or do folks want to look at hybrid subgroups too? Might as well, I think, if you go for it.

Of the 33 most played lineups I'd count about 7 as small ball.
http://bkref.com/tiny/lg4af

It looks like 3 positive, 4 negative. These are the most played, so arguably you'd expect them to be above average on average, I think. But not that impressive as a small group, in terms of % positive. Digging down for all lineups over 10-20 minutes or all lineups would be good but ideally would be aided by query filters on what are small ball lineups.

Is that a lot compared to past? Yes, by this cut. I only count 2 small lineups in the top 33 used over 250 minutes last season.

As always, more time and angles on the analysis would be useful.


Playing fast data probably attainable from NBA.com or nbawowy.com directly or with some manipulation.

Small ball lineup or traditional, what do you expect the 4 factor changes will be on average? I assume small ball team improves on at most own efg%, own turnover rate, own ft/fga and forced turnovers and gets worse on OR%, DR%, efg% allowed and ft/fga allowed. Agree or disagree? Potentially small lineups might not get better at own turnovers (because of more dribbling and passing) or own ft/fga (because of more longish jump shots and fewer fouls generated off post-ups, perhaps off fewer pick n rolls in general or at least fouls on shots from big man rollers, and fewer fouls on offensive rebound putback attempts). If a specific small lineup looks vulnerable to losing on own turnovers or own ft/fga on its own or in a specific matchup (or weaker than expected on other factors) I might be a bit more reluctant to use it. Of course it would be good to check these guesses against real data and factor impacts are of different scales relative to each other.

Crow wrote:Several people have asked me why so few SGs have positive RPM estimates (just 15) and 3/4ths of centers do so far this season. That is a question that could provoke several kinds of inquiries. Is it an imbalance in the talent pool? Are SGs trying to do too much on offense and too little on defense?

Please correct me, I could be very wrong on this since RPM is not my thing - but doesn't RPM have some sort of statistical +/- involved in which defensive rebounding (possibly relative to something else) can be an important factor in the defensive aspect of the RPM result?

Again - could be wrong, might be confusing it with something else - but I seem to remember bigs dominating the defensive RPM, which would be half their RPM total. Right?

SGs don't rebound much to help D results, don't assist as much for bigger O results. So, I'd think Bigs (especially shot blocking) & PGs (especially efficient ones, &/or ones w/ higher steal totals) would look better than all others, on average.

Now, if I'm completely confused - & RPM does not in any way incorporate any statistical factors & is purely +/- relative to teammates & opponents - then, I haven't a clue.....

RPM does use a box score based statistical prior. Bigs dominance of rebounding, defensive and offensive, will aid their prior and their score in the play by play regression. I was talking raw 4 factors above at lineup level but it flows down to players and their RPM. But until RPM gets enough splits, you wouldn't ordinarily see small ball affects in player RPM. Perhaps in a rare case where the vast majority of a player's time was in small ball lineups, instead of just a small fraction, as it probably typical.