Page 1 of 1
Calculating OREB%, DREB% and REB%
Posted: Fri Mar 31, 2023 2:15 pm
by Apostadorzao
Hi! Does anyone know how nba.com calculates OREB%, DREB% and REB%?
For example in this game:
https://www.nba.com/game/uta-vs-sas-002 ... raditional
Utah rebounding stats:
OREB 8
DREB 37
REB 45
San Antonio rebounding stats:
OREB 9
DREB 30
REB 39
But when it comes to the advanced stats, that’s what it shows:
https://www.nba.com/game/uta-vs-sas-002 ... e=advanced
Utah:
OREB% 21.4
DREB% 73.1
REB% 50.0
San Antonio:
OREB% 26.9
DREB% 78.6
REB% 50.0
In basketball-reference it is how the formula says:
https://www.basketball-reference.com/bo ... 90SAS.html
Utah:
OREB% 21.1 (-0.3)
DREB% 80.4 (+7.3)
REB% 53.6 (+3.6)
San Antonio:
OREB% 19.6 (-7.3)
DREB% 78.9 (+0.3)
REB% 46.4 (-3.6)
Do you have any idea of why it happens? Because it does with everygame.
Re: Calculating OREB%, DREB% and REB%
Posted: Fri Mar 31, 2023 5:24 pm
by Crow
Comparing advanced boxscores, almost all the difference is related to Barlow getting higher OReb% at BRef and presumably that affecting Jazz Def Reb%s. Why? I don't know. Ask them if important to you.
Re: Calculating OREB%, DREB% and REB%
Posted: Fri Mar 31, 2023 10:41 pm
by Mike G
Maybe NBA.com includes "team offensive rebounds" in their OReb%
At b-r.com it's simply ORb/(ORb+oppDReb). It's a ratio of ORb to "rebounded missed shots".
In the example given, the Spurs missed 50 FGA and 4 FTA. That's 54 misses. Utah got 37 of them (DRb) and SAS got 9 (ORb). The total is 46, so 8 misses went "unrebounded" by individuals. As many as 4 FTX may have been "unreboundable" if they were the 1st of 2 (or techs).
How they come up with 26.9% ORb% is hard to figure -- 9 ORb is 27% of 33.5, a far cry from 54 (or 50) ORb opportunities.
Unfortunately, "team rebounds" are a thing still, and they don't distinguish between actual possessions recovered (as when the opponent touches the ball as it goes out of bounds) and the meaningless missed 1st-of-2. If those "dumb rebounds" are included in the total OReb, you get a dumb ORb%.
Neither nba.com nor b-r.com is getting to a real team ORb%, if they aren't utilizing the play-by-play and including true team rebounds.
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 01, 2023 12:55 am
by Mike G
OK, I have skimmed over the play by play of the game, here:
https://www.basketball-reference.com/bo ... 90SAS.html
In that telling, the Spurs had 4 legitimate (possession retaining) Offensive Rebound by Team and 3 Defensive.
The Jazz had just one of each.
Additionally, SA had 3 dumb/absurd ORBT and Utah had 1.
This accounts for the differences. ORb% = ORb/(ORb+oppDRb)
The 2nd set includes the Team Rebounds.
Code: Select all
tm ORb DRb TRb ORb% DRb% TRb%
Uta 8 37 45 .211 .804 .536
SAS 9 30 39 .196 .789 .464
tm ORb DRb TRb ORb% DRb% TRb%
Uta 9 38 47 .214 .731 .500
SAS 14 33 47 .269 .786 .500
Note that the amended SAS line has a 5th ORb added, to reach the nba.com %numbers. Either I missed one, or they included one of the dummies.
(One was a missed 1st of 2 FT; next was on a missed technical FT; finally one at 0:00 in the game. None of these count for anything.)
In general, it looks like nba.com is (mostly) doing the job, and b-r.com is slacking -- estimating.
It's possible for a team to lose all 4 of the Four Factors and still win the game! They just have to dominate in the team rebounds.
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 01, 2023 2:16 am
by Apostadorzao
That was lit! I didn’t get to count the Team Rebounds because at nba.com it was 3 for Utah and 10 for San Antonio, so it wouldn’t reach 50%/50% REB anyway. I thought it was obtained by a formula. And I thought it because I saw some pattern in a few other games.
For example:
https://www.nba.com/game/gsw-vs-mia-002 ... /box-score
GSW and MIA had the same total Rebounds and still 50%/50% REB%, plus their difference of OREB and DREB is the same: +1 in OREB for GSW and +1 DREB for MIA.
There are more with same totals and these same difference. And there are more with the same total but not the same difference, so the REB% is not 50%/50%.
The point is that I thought they were weighting any of these stats. Do you think it is a possibility?
P.S.: english is not my first language, so sorry in advice if I made any mistake.
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 01, 2023 10:32 am
by Mike G
Apostadorzao wrote: ↑Sat Apr 01, 2023 2:16 am
... to count the Team Rebounds because at nba.com it was 3 for Utah and 10 for San Antonio,..
OK, same totals I counted. I just looked at the game situation to classify them as 'real' or 'unreal' events. So they counted one from the unreal list.
The point is that I thought they were weighting any of these stats. Do you think it is a possibility?
I don't understand the question, sorry.
At b-r.com, team OReb% run from 18 to 30%; at nba.com, they are 23 to 34.5%. Defense gets most of the rebounds, but those which go out of bounds are probably close to 50-50. If there are 4-6 per game -- 2-3 per team -- that would create the difference.
Does nba.com show Team Reb somewhere? A few years ago I found a site with these listed, and I'd love to get it back. Final scores just don't make sense when this number is unknown.
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 01, 2023 11:27 am
by Mike G
This issue just fascinates me for some reason.
NBA.com gives all teams a higher OReb% than does b-r.com; all teams also have lower DReb%. And the avg tradeoff is close to 50-50, but there is significant variation.
Houston Rockets have the highest OReb% in both accounts. NBA.com gives them 4.3 more OReb/100 -- and 2.9 fewer DReb/100. Opponents also get some of those out-of-bounds calls.
Per 100 rebounds, there are (4.3 + 2.9) 7.2 which b-r does not count; and of which (4.3/7.2) .598 are gotten by the Rockets. They not only lead the league in
gotten rebounds, but also in the
un-gotten ones.
Code: Select all
tm b-rO nbaO b-rD nbaD OR+ DR- get%
Hou 30.2 34.5 75.8 72.9 4.3 2.9 .60
Ind 23.4 28.6 72.4 68.6 5.2 3.8 .58
Was 22.7 27.5 76.2 72.4 4.8 3.8 .56
Dal 18.3 23.0 75.8 72.0 4.7 3.8 .55
LAC 22.8 27.2 76.6 73.0 4.4 3.6 .55
LAL 22.7 27.0 76.1 72.4 4.3 3.7 .54
Det 24.9 29.3 73.7 69.9 4.4 3.8 .54
OKC 24.6 28.8 72.8 69.0 4.2 3.8 .53
Min 21.6 25.9 74.1 70.2 4.3 3.9 .52
SAS 25.5 29.8 74.7 70.8 4.3 3.9 .52
Den 24.6 28.6 77.0 73.3 4.0 3.7 .52
Mil 25.1 28.5 77.6 74.3 3.4 3.3 .51
Cha 23.9 27.7 75.6 71.8 3.8 3.8 .50
NOP 24.9 29.1 77.5 73.3 4.2 4.2 .50
avg. 24.0 28.1 76.0 71.9 4.1 4.1 .50
Sac 22.1 26.1 77.5 73.5 4.0 4.0 .50
Uta 27.1 31.2 74.9 70.8 4.1 4.1 .50
Por 22.7 26.7 75.0 70.9 4.0 4.1 .49
Mia 22.8 27.4 77.7 72.9 4.6 4.8 .49
Orl 23.6 28.0 77.5 72.8 4.4 4.7 .48
Atl 25.1 29.0 75.7 71.5 3.9 4.2 .48
Phx 26.8 30.9 76.0 71.5 4.1 4.5 .48
Tor 28.0 31.8 76.6 72.1 3.8 4.5 .46
Chi 20.5 24.1 77.9 73.6 3.6 4.3 .46
Mem 26.7 30.5 75.6 70.9 3.8 4.7 .45
Phl 21.6 25.8 77.1 71.9 4.2 5.2 .45
GSW 23.8 27.1 76.1 72.0 3.3 4.1 .45
Bos 21.8 25.0 78.7 74.7 3.2 4.0 .44
Cle 23.6 27.4 76.4 71.6 3.8 4.8 .44
Brk 19.6 23.3 73.7 69.0 3.7 4.7 .44
NYK 28.3 31.7 76.8 72.4 3.4 4.4 .44
avg. 24.0 28.1 76.0 71.9 4.1 4.1 .50
Just a handful of teams are above 55% or below 45% in the fraction of Team Rebounds they get.
The avg is a bit under 2 per game per team, and a few teams are winning or losing by more than 0.4 per game.
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 01, 2023 6:37 pm
by Apostadorzao
Mike G wrote: ↑Sat Apr 01, 2023 10:32 am
Does nba.com show Team Reb somewhere?
https://www.nba.com/game/gsw-vs-mia-0022200104
Under LINESCORES - TREB.
Mike G wrote: ↑Sat Apr 01, 2023 10:32 am
I don't understand the question, sorry.
Let me try to explain. Like the examples I gave. GSW and MIA their OREB and DREB differences are the same, so the percentage kept the same. But Utah and San Antonio had different differences (SAS had +1 OREB, Utah had +7 DREB), so they were valuated differently and made the final REB% the same.
Another example:
https://www.nba.com/game/phi-vs-was-002 ... /box-score
PHI:
OREB 10 (-4)
DREB 29 (+4)
REB 39
WAS:
OREB 14 (+4)
DREB 25 (-4)
REB 39
Their REB% are 50%.
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 01, 2023 7:39 pm
by Mike G
OK, it says TREB: GSW 8, MIA 9
And I assume that includes non-events which are not even included in their Reb%.
They aren't even saying how many ORb and DRb. damn
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 01, 2023 10:08 pm
by Apostadorzao
You were right all the time. It’s all about the Team Rebounds. But I just found what we were missing. They use the rebounds that were counted on NBA gamebooks, they don’t use the ones in the summary.
The first example of UTA and SAS.
https://statsdmz.nba.com/pdfs/20230329/ ... S_book.pdf
They counted 2 for Jazz and 8 for Spurs, what gives a total of 47 for both.
The point now is: which do you think is the most correct percentage?
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 01, 2023 10:11 pm
by Apostadorzao
Mike G wrote: ↑Sat Apr 01, 2023 7:39 pm
OK, it says TREB: GSW 8, MIA 9
And I assume that includes non-events which are not even included in their Reb%.
They aren't even saying how many ORb and DRb. damn
But they counted 7 for GSW and 9 for MIA.
https://statsdmz.nba.com/pdfs/20221101/ ... A_book.pdf
What gives a total of 48 for both.
Re: Calculating OREB%, DREB% and REB%
Posted: Sat Apr 29, 2023 12:15 am
by Mike G
Code: Select all
REBOUND PERCENTAGES
OFF. - Percentage of a given team's missed shots which that team rebounds.
DEF. - Percentage of opponents' missed shots which a given team rebounds.
TOT. - Average of offensive and defensive rebound percentages.
That Total Reb% is the same formula I use in eWins.