FGM FGA FG% 3PM 3PA 3P% minutes +/-
Smart with Horford 37 86 43.0 17 39 43.6 350 +155
on floor
Smart w/out Horford 26 116 22.4 11 58 19.0 265 +1
per 36 with Horford 3.8 8.8 43.0 1.7 4.0 43.6
per 36 w/out Horford 3.5 15.8 22.4 1.5 7.9 19.0
Above is Smart's shooting with and without Horford on the floor. Smart's shot volume almost doubles when Horford is not on the floor and his shooting % is cut in half. When Marcus is on the floor as a 4th or 5th option he has been very successful. When he is stuck with the bench units that are devoid of quality offensive options and have much poorer spacing he has displayed some dreadful shooting.
Great find, good thinking, and your theory is plausible. TP.
There's been a debate in several threads over Baynes versus Morris as the fifth starter. Smart has not been included as a third option in those discussions, but in fact the three most used lineups, by far, have been:
Irving/Brown/Tatum/Horford, plus one of...
Morris, Baynes, or Smart.
And, in fact, those three lineups have been on the floor together an almost identical amount of time.
1. With Morris (84.4 minutes): Offense 1.06 points per possession; Defense 1.20 ppp
2. With Smart (82.5 minutes): Offense 1.16 ppp; Defense 1.02 ppp
3. With Baynes (79.7 minutes): Offense 1.03 ppp; Defense .92 ppp
There are a lot of moving parts here, including Morris' recovery (he still had a minutes restriction as of last week, iirc), widely different matchups, etc. And the sample size is small enough even at the end of the season...
So, with a grain of salt,
the net of the Morris lineup is -.14
the net of the Smart lineup is .14
the net of the Baynes lineup is .11
[The next most used lineup, fwiw, is Irving/Smart/Brown/Ojeleye/Horford (27.0 minutes): Offense 1.27 ppp; Defense 0.93, for a net .34 points per possession]
The numbers may seem tiny and the differences minuscule, but remember that we're talking about 100 possessions in a game, typically (actually Boston has been playing at a relatively slow 98.5 per game), so the difference between the Smart lineup and the Baynes lineup approaches 3 points per game - that translates to a big difference in wins over the course of a season.
There's an argument here for Smart starting instead of either Baynes or Morris. But maybe Brad Stevens is heeding the old NBA wisdom of Red Auerbach: "It's not who starts, but who finishes."
I disagree with this conclusion, since there's such a big difference in absolute ratings. Relatively I think the net values are approximately equal. Because with a higher score there's also more absolute (and probably relative) variance.
Say as a team every game you score between 104 and 128 points, your opponent scores between 92 and 112 points (based on the ratings of the line-up '2' with Smart and a 20% interval). To simplify math we assume there's an even distribution of possible outcomes among that interval, which leads to a 92,3% chance of winning.
We do the same for line-up '3' with Baynes. Our outcome between 93 and 113 (20% interval) and for our opponents a score between 83 and 101. Now we've got a 89,8% chance of winning.
If we had to choose between the line-ups over a whole season for 82 games that is an exact difference of 2 won games (Baynes' goes 74-8 and Smart's 76-6). Not that big a difference. Although if you see it from another standpoint, it's a relative huge increase of losses from 6 to 8 games. Of course it's all nonsense since you can't play every minute of every game with the same 5 players and obviously then in a vacuum the direct effect of the choice between Smart and Baynes becomes marginal.