Box score stats > regression models for player comparison purposes

[3ball]

RAPM doesn't work to compare players, and it doesn't matter if you compare players of similar role..

For every ONE time you say "oh look, RAPM says Shaq > Alonzo", there will be multiple incorrect comparisons, such as Divac > Hakeem... Ostertag > Shaq... Bo Outlaw > Rodman... Terry Mills > Detlef Schrempf... Brevin Knight > Gary Payton..... Laettner > Karl Malone and every PF in the league.. Jaren Jackson > Pippen..

And on.... And on.... And on.... And on... There are way too many to list here - again, MOST comparisons done using RAPM are blatantly wrong... Just look at the top 20 of any RAPM list, and you'll see a 15, 20, 30+ bad comparisons that make no sense.

Don't get me wrong - stat bots should be proud that their regression models have value in predicting game outcomes and providing an additional lense to evaluate the best lineups... But it does NOT work for player comparisons... Again, just look at the top 20 of any RAPM list, and you'll find more bad comparisons than accurate ones.

Otoh, even a cursory glance at box score stats will USUALLY give you the correct answer in any comparison... Here's a simple example - anytime a pf averages 5 more rebounds, or a wing averages 5 more ppg on better efficiency, they're "usually" the better player... Pretty simple right??.. But if you look at RAPM, you don't get this level of reliability - RAPM doesn't "usually" result in the right result - again, player comparisons done using RAPM are wrong far more often than they're right... and usually blatantly wrong, as shown earlier.

[ShawkFactory]

The man is riled up!

[navy]

Curry > Jordan

[90sgoat]

Agreed.

[DavisIsMyUniBro]

RAPM doesn't work to compare players, and it doesn't matter if you compare players of similar role..

For every ONE time you say "oh look, RAPM says Shaq > Alonzo", there will be multiple incorrect comparisons, such as Divac > Hakeem... Ostertag > Shaq... Bo Outlaw > Rodman... Terry Mills > Detlef Schrempf... Brevin Knight > Gary Payton..... Laettner > Karl Malone and every PF in the league.. Jaren Jackson > Pippen..

And on.... And on.... And on.... And on... There are way too many to list here - again, MOST comparisons done using RAPM are blatantly wrong... Just look at the top 20 of any RAPM list, and you'll see a 15, 20, 30+ bad comparisons that make no sense.

Don't get me wrong - stat bots should be proud that their regression models have value in predicting game outcomes and providing an additional lense to evaluate the best lineups... But it does NOT work for player comparisons... Again, just look at the top 20 of any RAPM list, and you'll find more bad comparisons than accurate ones.

Otoh, even a cursory glance at box score stats will USUALLY give you the correct answer in any comparison... Here's a simple example - anytime a pf averages 5 more rebounds, or a wing averages 5 more ppg on better efficiency, they're "usually" the better player... Pretty simple right??.. But if you look at RAPM, you don't get this level of reliability - RAPM doesn't "usually" result in the right result - again, player comparisons done using RAPM are wrong far more often than they're right... and usually blatantly wrong, as shown earlier.


I dont even know much about RAPM lol.

meant to measure in role effeciency, not overall, if you can comprehend that.

but if this is what I was waiting for, then...

I doubt FPlii will deal with you, but he knows the most about RAPM I think, other than a select few.

BUT IN ANYCASE, its good that you aborted your jordan offense> curry offense arguement.

[Quickening]

Curry > Jordan
What he said

[DrakeTheSnake]

Let me guess, a regression model showed a player is better than Jordan.

[3ball]

its good that you aborted your jordan offense> curry offense arguement.



Why would I abandon the facts??... MJ scored 35% more on better efficiency:


Per 100 Possessions in Playoffs:


JORDAN:. 43.3 pts.. 118 ORtg
CURRY:... 32.6 pts.. 115 ORtg


It's not even close

[TomBrady]

Confirmed box score watcher.

We real greats understand sports on a much deeper level.

[DavisIsMyUniBro]

Why would I abandon the facts??... MJ scored 35% more on better efficiency:


Per 100 Possessions in Playoffs:


JORDAN:. 43.3 pts.. 118 ORtg
CURRY:... 32.6 pts.. 115 ORtg


It's not even close

yesh

becash we are talking about 2015 curry

[HurricaneKid]

RAPM doesn't work to compare players, and it doesn't matter if you compare players of similar role..

For every ONE time you say "oh look, RAPM says Shaq > Alonzo", there will be multiple incorrect comparisons, such as Divac > Hakeem... Ostertag > Shaq... Bo Outlaw > Rodman... Terry Mills > Detlef Schrempf... Brevin Knight > Gary Payton..... Laettner > Karl Malone and every PF in the league.. Jaren Jackson > Pippen..

And on.... And on.... And on.... And on...

In order to calculate regression you need to have data for who was on the floor for every point scored in the NBA. That information doesn't EXIST AT ALL before 1997 and it was fairly spotty until ~2002 (subs during FTs, etc).

So the foundation of your argument is a joke. Hakeem, Malone, Rodman, Schrempf, Payton WERE garbage at the end of their careers by the time the stat could even be calculated properly. And yes, it is prone to anomoly and requires a sizable data set.

Again, you posit things without even a cursory understanding of the stats you are attempting to undermine.

There is a reason everyone rolls their eyes when you make a post/thread. You need to spend more time grasping what things mean and a lot less time copying and pasting your disproven points.

[HurricaneKid]

Let me guess, a regression model showed a player is better than Jordan.

Sadly no. We don't even have regression data for any but the final few years of MJ's career.

[3ball]

So the foundation of your argument is a joke. Hakeem, Malone, Rodman, Schrempf, Payton WERE garbage at the end of their careers by the time the stat could even be calculated properly.


The comparisons posted in the OP used 1997 and 1998 RAPM data that used digitized play-by-play data.

Also, Malone was MVP in 1997... Payton was just coming off his DPOY and was perennial all-nba.. Hakeem was 1st team all-nba.

Schrempf was all-star in 1997.. So as usual, a poster who is refuting me is factually incorrect - these guys were NOT "garbage"... obviously.
.

[DavisIsMyUniBro]

The comparisons posted in the OP used 1997 and 1998 RAPM data that had play-by-play.

Malone was MVP in 1998... Payton was just coming off his DPOY and was perennial all-nba.

Schrempf was all-star in 1997.. So as usual, a poster who is refuting me is factually incorrect - these guys were NOT "garbage"... obviously... :rolleyes:


dodging curry like the matrix

[HurricaneKid]

The comparisons posted in the OP used 1997 and 1998 RAPM data that used digitized play-by-play data.

Also, Malone was MVP in 1997... Payton was just coming off his DPOY and was perennial all-nba.. Hakeem was 1st team all-nba.

Schrempf was all-star in 1997.. So as usual, a poster who is refuting me is factually incorrect - these guys were NOT "garbage"... obviously.
.

Again, we have NO DATA pre 1997. From 97-2001 the data we have is pretty lousy. For instance, when subs come in during FTs each scorekeeper did it differently (as far as who was on the floor when the FTs went down).

So the data we have pre 2001 is pretty suspect and people generally don't use that info.

Explaining basic, well understood concepts to you shouldn't be nearly so difficult.

[RRR3]

http://fotos.fotoflexer.com/34c6ed7676b9586045f094544d4c5cf6.jpg