Outside wrote:dhsilv2 wrote:Owly wrote:Unless I'm doing something seriously wrong here, Brand is ahead. Brand 110 for career (only 105 in Dallas), Melo is 108 as Outside has it.
Brand: https://www.basketball-reference.com/players/b/brandel01.html
Anthony: https://www.basketball-reference.com/players/a/anthoca01.html
And this, measured in raw terms, gives a slight pro-Carmelo bias. In Brand's four seasons before Carmelo, the league average Ortgs were 104.1; 103; 104.5 and 103.6 - in 2017, post Brand, Carmelo played in a league that was 108.8. Then too Brand peaked and played his most minutes in the lower norm Ortg league. Using relative Ortg would mitigate some of the problems (though probably not easy to get a minutes weighted career average for that) or comparing too Drtgs would mitigate bias.
well I just give up then, lol. I vividly remember 105 and 108, but thought Brand was ahead. Now I'm seeing 108 and 110 still with brand ahead.
Brand behind ahead however is what I would expect. Melo's "value" is in his ability to be that high usage guy. I however have been debating here if that's a plus or a minus from him.
I think I'm in the same boat -- I would've sworn in court that I saw Carmelo ahead for RS, 108-105. Only had two tabs open for this. Geez.
Now I have a follow-up question -- why is Brand's ORtg higher? I believe posters in both this and the POY threads have proposed skipping listing TS% and TOV% separately and just listing ORtg as a single efficiency stat. However, assists are also part of the mix, and it's a complicated calculation.
https://www.basketball-reference.com/about/ratings.html
However, with the career RS numbers of Carmelo and Brand, it doesn't make sense that Brand comes out ahead.
TS% Carmelo 54.3, Brand 54.8
TOV% Carmelo 10.8, Brand 11.8
USG% Carmelo 31.1, Brand 22.8
AST% Carmelo 15.8, Brand 11.7
APG Carmelo 3.0, Brand 2.1
The only category Brand leads in is TS%, and that's by 0.5%. Carmelo leads in all the other categories by more significant margins than that.
So how does Brand wind up with a higher ORtg?
Looking at the formula more closely, it also includes offensive rebounds, and Brand is considerably better in that area. That seems to be the difference.
Not that I've done a deep dive into the ORtg formula, but I've at least identified the components -- shooting efficiency, turnovers, assists, and offensive rebounds. So while I can see how it's trying to measure a player's overall contribution to the offense, I'm not sure I equate that with offensive efficiency in the way that most people think of it. A relatively low usage player can end up with a really high ORtg by being a good offensive rebounder, and that's why Steven Adams, Clint Capela, and DeAndre Jordan are in the top five for ORtg this season. But those aren't the guys who drive their offenses, and while they have very good TS percentages, that's because their offense consists of finishing lobs and converting putbacks.
Offensive rebounds are great and obviously help an offense by extending a possession, but when trying to gauge a player's efficiency, I think most people think of what a player does when he has the ball -- scoring per attempt, turnover rate, assist rate. Knowing what I do now, I find listing the component parts of efficiency as more informative than just deferring to ORtg.
You'll note that I separate usage % when discussing ORtg because that is a key driver in the context we should look at it with.
ORtg is attempting to tell us that when that player is a part of the box score metrics on offense, what will that team score per 100. It tells us the efficiency of a theoretical offense where that player was involved in each play.
Low usage guys clearly can't create and thus we should not use this metric to compare them to volume guys, but to their peers.
You'll also note that there are some team based scale in the metric too which I can't comment on if that helps or hurts Brand.
In short this metric tells me that Brand utilizes possessions better than Melo does. He does it at a high usage rate, but not nearly as high as Melo's rate which is really staggeringly high.
As with anything, individual pieces are useful, but it is nice to have metrics that logically mix them together to paint that broader picture. That is why PER imo is still a useful tool, even if I don't go to it often.
