Celtics Frontcourt

slamminsammya

Member
SoSH Member
Jul 31, 2006
9,152
San Francisco
This year's incarnation of the Celtics features a strong offense and a mediocre defense, representing an interesting change from last year which, though comparable in win loss record, was based on a strong defense and middling offense. The only major change from last year was the addition of Al Horford. By watching, or by a quick basketball reference lookup, it is clear that rebounding is an issue for the Celtics. The mediocre defense and bad rebounding are not independent, and in my opinion both areas are generally dominated by the play of the frontcourt.

So I thought it would be interesting to look at the various frontcourt combinations used by Brad Stevens this year, using nbawowy.com, and maybe open up a discussion about the various combinations and what people think is lacking.

I looked at the five most common two man combinations this year, which altogether account for 62% of all minutes played. Notably I didnt look at combinations involving Zeller, both because including him would have increased the number of combinations combinatorially (ha), and also because I hate Zeller and hope he does not feature in our playoff rotation. Zeller has played approximately in 16% of the minutes this season.

The most common combinations are, in order of minutes played: Johnson-Horford, Jerebko-Olynyk (!), Horford-Olynyk, Johnson-Olynyk, Jerebko-Horford.

Lets start with looking at how effective each pairing has been by looking at their net rating, which is 100 times points per possession minus points allowed per possession. On a team level, this simple statistic is a very good indicator of team strength.

Celtics Overall: +2.2
Johnson-Horford: +7.9
Johnson-Olynyk: +0.3
Horford-Olynyk: +3.6
Jerebko-Olynyk: +3.5
Jerebko-Horford: +8

Right off the bat there are some interesting numbers. The strongest pairing for net rating is Jerebko-Horford. I was surprised by this, since Jerebko has looked generally useless this year. I think I have a hypothesis for this, which I will get to below. Also interesting is how strong the Johnson-Horford pairing is, since it seems that is usually not the frontcourt that closes out the games.

Here are some other interesting traits for each pair.

Johnson-Horford
  • Play at a much slower pace than all other pairings and slower than the team overall (90.8 possessions/game), the team overall plays at 94.7. Not surprising. If they were a team this would make them the slowest in the NBA.
  • Has a much higher block rate than the team overall (6.6% vs. 4.9%)
  • Does several things slightly better than the team, but none stand out. Altogether, however, they add up to a very strong unit.
Johnson-Olynyk
  • Struggles mightily on offense. They score 101.2 points per 100 possessions, far worse than the team rating of 112.7
  • This is mostly because the shooting efficiency for this group is very poor. TS% is 52, compared to 56 for the team (true shooting percentage accounts for free throws and the fact that threes are worth 3 points)
  • However, they are also very strong defensively, allowing 100.9 points per 100. For reference, the best team defensive rating in the NBA is held by the Utah Jazz 104 points per 100,
  • They hardly block any shots (block rate is 2.2%) but cause more turnovers.
  • They defend 2 point field goals much better than the team overall (45.5% vs 49.3%). This seems incongruous with the fact that they do not block shots.
Horford-Olynyk
  • 2nd best offensive grouping, barely behind Horford-Johnson.
  • Shooting numbers, rebounding numbers, steals and blocks all very similar to Horford-Johnson.
  • Then why is this unit half as good as Horford-Johnson? This is probably the biggest standout. Opposing teams shoot a whopping 42% on threes against this group! That is on 114 attempts for those who want to think about sample sizes.
Jerebko-Olynyk
  • Cause the most turnovers of any pairing.
  • Defend the 3 very well - opponents shoot 30% against this group. Team percentage is 34.8%.

Jerebko-Horford
  • Our second best pairing. Why is that? Here are some clues.
  • Highest assist percentage (pct of made baskets resulting from an assist) at 71.4%, team is 64.4%
  • Best rebounding group grabbing 48.2% of possible rebounds, team is 47.5%.
  • Cause very few turnovers, but defend the three well - 30% opponent pct.
  • Defend 2pt fgs well as well at 46.6%
There is a lot of interesting stuff here, and I am sure that a significant portion of the differences here is due to statistical noise or attributable to other factors like who else plays with these groups. But I just want to comment on what I find the most interesting statistic, which is the Horford-Olynyk group defending threes. 42% is a crazy number. Does anyone have some insight into why this might be?
 

Cesar Crespo

79
SoSH Member
Dec 22, 2002
21,588
Are they really mediocre defensively? Their real problem seems to be that they give up more FTs than 24 other teams. Also, Kelly Olynyk is a very hard player to judge and you'll find opinions all over the spectrum.
 

slamminsammya

Member
SoSH Member
Jul 31, 2006
9,152
San Francisco
I forgot to mention that it seems like Jerebko lineups are good at defending the three, which intuitively makes sense as Jerebko, while certainly not a rim protector, seems good at moving his feet out on the perimeter. Also, box score stats are bad at accounting for things like defensive positioning and awareness. Perhaps the Swede is a very solid team defender.
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY
There is a lot of interesting stuff here, and I am sure that a significant portion of the differences here is due to statistical noise or attributable to other factors like who else plays with these groups. But I just want to comment on what I find the most interesting statistic, which is the Horford-Olynyk group defending threes. 42% is a crazy number. Does anyone have some insight into why this might be?
The influence a player has on opponent Three Point Percentage is so minuscule that it probably should never be mentioned. While it's obviously a skill, it's too variable to show up in single year data. You should look at opposing 3PT attempts instead if you want to determine who is a good perimeter defensive unit.
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,617
I don't think these front court rotations hold the back court constant so I don't really know how to evaluate them. is there a real plus minus for pairs?
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY

Big John

New Member
Dec 9, 2016
2,086
How does defensive rebounding (or lack thereof) affect these numbers? My eyeballs tell me that whichever combination of bigs Stevens uses, the other team gets far too many second and third shot opportunities.
 

HomeRunBaker

bet squelcher
SoSH Member
Jan 15, 2004
30,096
I forgot to mention that it seems like Jerebko lineups are good at defending the three, which intuitively makes sense as Jerebko, while certainly not a rim protector, seems good at moving his feet out on the perimeter. Also, box score stats are bad at accounting for things like defensive positioning and awareness. Perhaps the Swede is a very solid team defender.
I don't have the numbers in front of me but it seems as though the majority of Jerebko's minutes come against second units at the end of the first quarter and the first 6 minutes of the 2nd quarter when opponents best scorers are resting. The Jerebko/Olynyk pairing could certainly benefit from these matchups especially with opponents turning the ball over more with their second units.
 

Cellar-Door

Member
SoSH Member
Aug 1, 2006
34,457
I was wondering how each does in the small lineup.
Here they are solo (that means all other bigs are off-court)
Bigs are considered: Horford, Olynyk, Johnson, Zeller, jerebko, mickey

Horford- (This is mostly the IT and D lineup as some called it that many thought would close out games) -5.8, 116.4 Ortg, 122.2 Drtg.

Olynyk- (only 55 minutes) +11, 125.5 Ortg 114.5 Drtg

Amir (33 minutes)- -10.3, 114.7 Ortg, 125 Drtg

nobody else got real minutes in the small lineup.


As to the 3pt% on the Horford/Olynyk pairing. I agree it's likely mostly noise, but one thing that popped out.... it's 36.6% when IT is on the bench (on 41 attempts in 160 possessions) with that pairing, 45.2% when he's on the floor (on 73 attempts in 259 possessions). So teams are taking and making more 3s when IT shares the court with those 2. Now could be that IT defends 3s poorly, or that IT's overall bad D means more need for bigs to rotate, leaving a shooter open.


This is interesting.....
Overall teams are taking 3s against Boston on 27.7% of possessions when IT is on the bench and hitting at 30.4%. When he's on the court, take 3s on 28.6% of possessions and shooting 37.2%. Pretty interesting.
 

slamminsammya

Member
SoSH Member
Jul 31, 2006
9,152
San Francisco
How does defensive rebounding (or lack thereof) affect these numbers? My eyeballs tell me that whichever combination of bigs Stevens uses, the other team gets far too many second and third shot opportunities.
This was another interesting thing in the numbers. The rebounding rate is pretty stable across all combinations, and stably bad. Around 47% for pretty much all of them. This makes sense, as no big man is an above average rebounder at their position. The only plus rebounder we have this year may be Bradley.
 

Sprowl

mikey lowell of the sandbox
Dope
SoSH Member
Jun 27, 2006
34,437
Haiku
This was another interesting thing in the numbers. The rebounding rate is pretty stable across all combinations, and stably bad. Around 47% for pretty much all of them. This makes sense, as no big man is an above average rebounder at their position. The only plus rebounder we have this year may be Bradley.
Rebounding may be be the best part of Rozier's game, but he lost the competition with Smart to be the backup point guard, so he no longer gets the minutes to show off his ballhawking talents.
 

slamminsammya

Member
SoSH Member
Jul 31, 2006
9,152
San Francisco
Not sure where else to post this but this website is great: http://alexwainger.github.io/NBASubstitutionPatterns/

One conclusion from the patterns for the Celtics: The two best big man pairings by net rating, namely Johnson-Horford and Jerebko-Horford, are quite rare in the fourth quarter.

Another: Brad likes to end games primarily with IT-Smart-Bradley-Crowder-Horford. This unit has played 63 minutes together, and they are terrible! Net rating is -9.1, with a horrific 133.1 defensive rating. Maybe we should start a Brad Stevens thread?
 

mcpickl

Member
SoSH Member
Jul 23, 2007
4,546
Not sure where else to post this but this website is great: http://alexwainger.github.io/NBASubstitutionPatterns/

One conclusion from the patterns for the Celtics: The two best big man pairings by net rating, namely Johnson-Horford and Jerebko-Horford, are quite rare in the fourth quarter.

Another: Brad likes to end games primarily with IT-Smart-Bradley-Crowder-Horford. This unit has played 63 minutes together, and they are terrible! Net rating is -9.1, with a horrific 133.1 defensive rating. Maybe we should start a Brad Stevens thread?
Not necessary. He experimented with that lineup closing the game, but last time he did was vs Houston on 12/5. That was 20 games ago. Since then he's only used it in the middle of a 4th quarter once, for 8 possessions, against Indiana on 12/22.

Don't worry, Brad has it under control.
 

wade boggs chicken dinner

Member
SoSH Member
Mar 26, 2005
30,482
Another: Brad likes to end games primarily with IT-Smart-Bradley-Crowder-Horford. This unit has played 63 minutes together, and they are terrible! Net rating is -9.1, with a horrific 133.1 defensive rating. Maybe we should start a Brad Stevens thread?
One thing about these stats is that it doesn't really take into account the substitution patterns of the other team. For example, Brad plays Amir versus some guys and KO in other situations to maximize effectiveness. If he didn't have all of these options, some of the combinations would be worse.

And as for the "IT & D" lineup, as MP says, Brad recognized that they were getting smoked (particularly on the boards) and has moved away from playing that lineup except for certain situations.
 

Lose Remerswaal

Experiencing Furry Panic
Lifetime Member
SoSH Member
Not sure where else to post this but this website is great: http://alexwainger.github.io/NBASubstitutionPatterns/

One conclusion from the patterns for the Celtics: The two best big man pairings by net rating, namely Johnson-Horford and Jerebko-Horford, are quite rare in the fourth quarter.

Another: Brad likes to end games primarily with IT-Smart-Bradley-Crowder-Horford. This unit has played 63 minutes together, and they are terrible! Net rating is -9.1, with a horrific 133.1 defensive rating. Maybe we should start a Brad Stevens thread?
A: It's only 63 minutes total
B: as pickle said, it's been a few weeks since he last did this
C: haven't you started enough threads already?
 

slamminsammya

Member
SoSH Member
Jul 31, 2006
9,152
San Francisco
A: It's only 63 minutes total
B: as pickle said, it's been a few weeks since he last did this
C: haven't you started enough threads already?
Its exciting starting a thread! And the majority of the basketball discussion in this subforum has been about the Nets doing badly.

In last night's win over Charlotte the Celtics took over the game during a 5 minute stretch in the third quarter with a 14-4 run, extending the lead to 13 points. That was with Horford and Olynyk out there. The same 5 man unit closed out the game on an 8-2 run.
 

Sprowl

mikey lowell of the sandbox
Dope
SoSH Member
Jun 27, 2006
34,437
Haiku
haven't you started enough threads already?
A: No. The Port Cellar needs more threads, not fewer.

Its exciting starting a thread! And the majority of the basketball discussion in this subforum has been about the Nets doing badly.

In last night's win over Charlotte the Celtics took over the game during a 5 minute stretch in the third quarter with a 14-4 run, extending the lead to 13 points. That was with Horford and Olynyk out there. The same 5 man unit closed out the game on an 8-2 run.
Now that Olynyk's shooting is finally starting to heat up to 2016's pre-injury levels, the Horford-Olynyk tandem is starting to show what serious matchup problems it presents.
 

JCizzle

Member
SoSH Member
Dec 11, 2006
20,530
A: No. The Port Cellar needs more threads, not fewer.



Now that Olynyk's shooting is finally starting to heat up to 2016's pre-injury levels, the Horford-Olynyk tandem is starting to show what serious matchup problems it presents.
If we can get the Kelly back from the first half of last year, it would solve a ton of our problems.