Analyzing the Celtics

[Crow]

This is a first pass:

Compared to last season and the league now they are really succeeding on defensive rebounds and limiting fouls. They are have a lower 3pt fga rate than last season but it is still above league average. They have great pair and full lineup results and presumably chemistry. They aren't dominating as much at PG but are doing better / winning most of the rest of the positions. Their big men are shooting very well and defending shots well. This is only a modestly better than average Moreyball team and a poor overall shooting team (especially the bench guards). That might get them in the end.

They have faced a modestly easier than average strength of schedule. This is particularly true with regard to opponent rebounding strength.

They have been very good in both shot clock crunch time and in the clutch (and / or lucky) at both ends of the court.

Stevens hadn't looked like as much of a genius previously in NBA to me but he does right now and the distance between then and now is large. Call the earlier high praise from others pre-mature or prescient.

Limiting fouls given makes them a strong defensive anti-Morey ball team. Without that they are good but not special.

Do they care about RPM estimates? I dunno, will major media ever ask or ever get a forthcoming answer? From the guys they let go it sorta appeared no. But they haven't misses them and most of these are doing worse on RPM in their new places.

Horford is by RPM (and perhaps other tools) the leading player impact story. But it I also notable that no one is really causing drag.

What else do you see? I haven't looked at player tracking or play type data, yet. Their pace is 2 possessions below league average. Does anyone have the offensive / defensive splits? From what 82games has, it looks like the offense is slower.

[Crow]

Break-even on average in first half, big winner in second half, especially 3rd quarter.

Doesn't do much on fastbreak. Bench plays more minutes than average but scores less than average and is dead last on 3pt fg%. Still wins mildly though. Pretty good team A/TO. 6th highest on isos but terribly inefficient. Meh on pick n roller ball handler, great on rolls. post-ups, pretty frequent an drelatively good but it is still a below average shot. Spot-ups, handoffs- well below average. Cuts and putbacks, 2nd worst in league. Screens, most efficient but only average in use. High on contested shot %. A bit lower than average on passes, touches (including elbow and post) and dribbles. Just one guy over 25% usage. League average (no minute qualifier) is 2.2.

[Crow]

187 lineups used (down 17% from same time last season), 41% positive (that's high and up from 34% last season). But none over 5 minutes per game. Only 3 over 2 minutes per game and only 8 over 1 minute. So it is still improv, not standard, nothing really proven. Only winning 21% of the super dink lineups used less than 1 minute per game. 10 biggest lineups getting 37% of the minutes; bottom 177, 73%.

[josecarlos]

Sorry ... but what is or what is the meaning of ... Moreyball Team?

[watto84]

Named after rockets gm Daryl morey. Basically there are only 3 shots allowed...3’s, dunks/layups and free throws. No shots from the midrange. Daryl is 1 of the pioneers in terms of implementing an analytic driven game.

[Rd11490]

187 lineups used (down 17% from same time last season), 41% positive (that's high and up from 34% last season). But none over 5 minutes per game. Only 3 over 2 minutes per game and only 8 over 1 minute. So it is still improv, not standard, nothing really proven. Only winning 21% of the super dink lineups used less than 1 minute per game. 10 biggest lineups getting 37% of the minutes; bottom 177, 73%.

Just to add some context, they've lost :
Kyrie - 1 game - facial fracture
Horford - 1 game - concussion
Smart - 1 game - ankle
Morris - 9 games - knee
Tatum - 1 game - ankle
Hayward - season - ankle

I'd expect a lot of line up experimentation to try to patch up these holes.

[Rd11490]

187 lineups used (down 17% from same time last season), 41% positive (that's high and up from 34% last season). But none over 5 minutes per game. Only 3 over 2 minutes per game and only 8 over 1 minute. So it is still improv, not standard, nothing really proven. Only winning 21% of the super dink lineups used less than 1 minute per game. 10 biggest lineups getting 37% of the minutes; bottom 177, 73%.

Do you believe that coaches have a set of valid heuristics used to cut a dink lineup before it can do any more damage? For example, 10 possessions may not be enough to be statistically significant, but do you believe that a coach can learn something in those 10 possessions that tells them that it is highly unlikely the lineup will improve?

[Crow]

If the minutes are very small, in general I don't think they have a basis for saying it is highly unlikely to improve. If minutes increase and it stays terrible, then they'd have a better case. But the case against dink and super dink lineups is more that you know more about the non-dink lineups so select from them and probably max the best of them and not try to guess on super thin evidence about the lesser used lineups. Try 10-20, increase use of best 5-10 unless or until the results turn bad. More likely to learn about a small group than hundreds.

[josecarlos]

Named after rockets gm Daryl morey. Basically there are only 3 shots allowed...3’s, dunks/layups and free throws. No shots from the midrange. Daryl is 1 of the pioneers in terms of implementing an analytic driven game.

Thank you very much for your answer!!!

[Crow]

Celtics (& Stevens) winning 37% of lineups. 224 lineups used, on average a bit more than twice. 7 of 8 used over 20 minutes for season are positive or 88% positive. Among dink lineups with less minutes than that, the win% is 35%. 66% of all minutes went to dink lineups. No lineup has been used a full 4 minutes per game for season. Very good overall... but probably could be even better with more lineup concentration and maybe more selectivity.

[Crow]

Over 10 games there has been considerable slippage in both kinds of rebounding, steals and net turnovers. Offset by stronger 3pt fg% but net point margin down from previous 7 to 5.5. The slippage on defense has been about twice as much as the improvement of offense. Overall they've changed dramatically from very good offense / great defense to great offense / maybe fairly good defense (for last 3-5 games defense has been bad.) Has the shift been noticed by major media? Where? Last night crushed on boards but saved by 3 point shooting.


Celtics deep bench- Theis has done well. The rest? Not much to terrible on box score stats / metrics.

[Crow]

Celtics finish season with 21 of last 35 games on road. Doing well on road so far but will they at the end? Will they tire more than others?

[Crow]

In first 15 games, Celtics averaged 102.7 points per game and gave up just 94.5. In last 10 games they've scored 107.9 but given up 102.

Defense might be tired already, or just less motivated, effective and / or lucky.

Still haven't seen this discussed elsewhere but I don't get everywhere.

[Rd11490]

I think there is a good amount of luck. The difference in their defensive points per shot and expected points per shot has halved since earlier in the season and their expected points per shot has increased from .99 to 1.005.

As of 11/18
https://i.imgur.com/xqtPFYQ.png
As of 12/3
https://i.imgur.com/XhWhmxP.png

[Crow]

Tracking change on that differential adds insight into performance vs. "luck" or randomness. Should be a common practice.