Defensive rating (DRtg) How you calculate it and use it

[fpliii]

Looks good to me, but I'm not going to pretend to understand it haha. You'll have to explain to me what the z-rating is and how you calculate it. I'm guessing you just add it to the original ORtg or DRtg?

z-rating is the number of standard deviations (SDs) from the mean you are

so the formula is

(sample value - sample mean) / (sample SD)

where mean and SD are calculated as normal

[magictricked]

fpliii not 100% sure what you're trying, will follow along but don't do what i did once... start when they put the 3 point shot in.. don't go further back or you'll drive yourself batty figuring out why everything's skewed funny

[fpliii]

fpliii not 100% sure what you're trying, will follow along but don't do what i did once... start when they put the 3 point shot in.. don't go further back or you'll drive yourself batty figuring out why everything's skewed funny

lol I'm doing it from the beginning of the shot clock era

since we're comparing performances from the norm in a given year, the data won't matter as much (since SD and mean are both robust enough that sampling data won't be as impactful)

[magictricked]

lol I'm doing it from the beginning of the shot clock era

since we're comparing performances from the norm in a given year, the data won't matter as much (since SD and mean are both robust enough that sampling data won't be as impactful)

haha wow that's like a career move.


Ok I see the post above what you're doing... that's cool

[KG215]

z-rating is the number of standard deviations (SDs) from the mean you are

so the formula is

(sample value - sample mean) / (sample SD)

where mean and SD are calculated as normal

Yeah, that's still a little over my head. Is the ORtg or DRtg of a given year the sample value for the for formula?

[fpliii]

Yeah, that's still a little over my head. Is the ORtg or DRtg of a given year the sample value for the for formula?

sorry

sample value is that team's ortg, drtg, or pace

sample mean is the league average ortg, drtg, or pace for that season

sample SD is the league SD of ortg, drtg, or pace for that season

so like sample value for BOS 1966-67's ORtg would be 98.8

[magictricked]

sorry

sample value is that team's ortg, drtg, or pace

sample mean is the league average ortg, drtg, or pace for that season

sample SD is the league SD of ortg, drtg, or pace for that season

so like sample value for BOS 1966-67's ORtg would be 98.8

So how do you adjust when you want to compare a team from the 80's vs a team from the 2000's?

I always wondered if a baseline number could be created. say by averaging out the defensive rating for all years then working from there you could assign a plus or minus number to a team's defensive rating.

quick and dirty would be something like, say the average is 99.7 and one team is at 96.3 because it's a defensive number that team would be +3.4 rating vs. a team that might have a 100.2 rating that team would be a -.5 team

[fpliii]

So how do you adjust when you want to compare a team from the 80's vs a team from the 2000's?

I always wondered if a baseline number could be created. say by averaging out the defensive rating for all years then working from there you could assign a plus or minus number to a team's defensive rating.

quick and dirty would be something like, say the average is 99.7 and one team is at 96.3 because it's a defensive number that team would be +3.4 rating vs. a team that might have a 100.2 rating that team would be a -.5 team

well this is done within each year, so you'd compare z-scores (also it wouldn't be -.5, since it's not an an absolute scale; it'd depend on how big the standard deviation is for each of those seasons)

so in short, it will tell you how good the 80 defense was relative to that season versus how good 00 was relative to that season

since the game has changed, I think that's the closest we can get to comparing teams across eras

BTW note to everybody in case it's not clear: higher z-score for ORtg is better, lower z-score for DRtg is better

the z-score for pace is kinda meaningless (it tells you how many plays that team ran compared to league average), but it's there for completion purposes

in short, just compare the two DRtg z-scores for the teams (say 80 LAL and 00 LAL), and the lower is better

[Yao Ming's Foot]

well this is done within each year, so you'd compare z-scores (also it wouldn't be -.5, since it's not an an absolute scale; it'd depend on how big the standard deviation is for each of those seasons)

so in short, it will tell you how good the 80 defense was relative to that season versus how good 00 was relative to that season

since the game has changed, I think that's the closest we can get to comparing teams across eras

BTW note to everybody in case it's not clear: higher z-score for ORtg is better, lower z-score for DRtg is better

the z-score for pace is kinda meaningless (it tells you how many plays that team ran compared to league average), but it's there for completion purposes

in short, just compare the two DRtg z-scores for the teams (say 80 LAL and IND LAL), and the lower is better

Are we just going to pretend that putting numbers against a league average defense during an era of video game offensive numbers league wide is equivalent to putting up numbers against a league average defense during an era where scoring is tough league wide?

[fpliii]

Are we just going to pretend that putting numbers against a league average defense during an era of video game offensive numbers league wide is equivalent to putting up numbers against a league average defense during an era where scoring is tough league wide?

I'm not comparing individual numbers, this is for comparing the effectiveness of team defenses (and team offenses) with regard to league averages

btw what specific era are you talking about? I can explain more clearly/differently if I'm speaking in specifics instead of generalities

[KG215]

Pretty sure he's talking about 80s and early 90s compared to early 2000s or so. mainly, the eras Kobe and Jordan played.

[fpliii]

Pretty sure he's talking about 80s and early 90s compared to early 2000s or so. mainly, the eras Kobe and Jordan played.

ah

well, I didn't get to them yet

but the z-scores won't tell us much about Kobe or Jordan in that case, he'd want to just compare league average DRtg if he's speaking in general

if he's talking playoff numbers though, then the z-scores are useful to him

in order to make use of these numbers though, you need to adjust scoring for pace, which actually helps Kobe significantly (Jordan played on a slow team, but in a faster era)

he'd want to calculate points per 100 possessions for both guys, and then compare to league average DRtg (in general comparing pace-neutral stats is dangerous, but both guys had similar roles while playing similar minutes)

EDIT: the formula he'd want is

PP100 = (100 / team pace) * ((team MP) / 5) / player MP) * (player total points / team games)

Yao Ming's Foot - if you're interested in any particular results for PP100, let me know

[Yao Ming's Foot]

I'm not comparing individual numbers, this is for comparing the effectiveness of team defenses (and team offenses) with regard to league averages

btw what specific era are you talking about? I can explain more clearly/differently if I'm speaking in specifics instead of generalities

This thread was originally bumped again after I mentioned the differences in defenses faced between prime Jordan (1st threepeat) and prime Shaq (threepeat with the Lakers).

I understand that you are interested in ranking the strength of the defenses relative to eachother and can see why it makes to compare them to league averages in that case but if we are comparing the raw offensive numbers derived against those defenses by Jordan and Shaq can you justify a reason why league averages would be used if the offensive raw numbers are not adjusted by league averages?

[fpliii]

This thread was originally bumped again after I mentioned the differences in defenses faced between prime Jordan (1st threepeat) and prime Shaq (threepeat with the Lakers).

I understand that you are interested in ranking the strength of the defenses relative to eachother and can see why it makes to compare them to league averages in that case but if we are comparing the raw offensive numbers derived against those defenses by Jordan and Shaq can you justify a reason why league averages would be used if the offensive raw numbers are not adjusted by league averages?

see my response in the post just above you

EDIT: if you want a general quantitative comparison, then using the stat above

adjusted scoring per 100 possessions = pp100 * (100 / league average DRtg) * (100 / league average ORtg) = pp100 * (100 / league average ORtg)^2 (since they're the same across the league)

for a playoff series, you have to calculate pp100 using numbers from the playoffs only, and you can't simplify (since he's playing against one specific defense)

this won't be perfect, but this is the application I think you want

[Yao Ming's Foot]

see my response in the post just above you

EDIT: if you want a general quantitative comparison, then using the stat above

adjusted scoring per 100 possessions = pp100 * (100 / league average DRtg) * (100 / league average ORtg)

I see that you would suggest accounting for pace and league average defensive rating.

My question is why? By adjusting for league averaging defensive rating you are making the false declaration that offensive performances against all league average defenses are identical. Its simply not true. Scoring well against a league average D in 2001 for example is much more impressive than scoring well against a league average in 1991.