Which Advanced Stat is the Best/Worst to analyze player value?

[tholdren]

...

[ParadoxEN]

WS Per 48 is the best, IMO.

[dabom]

Integral of RPM. That is the summation of their(basketball players) total worth over the season. I just made this stat up right now.

Winshares per 48 and Total Winshares. Total winshares not as big as winshares per 48.

PER lastly. I rarely use this though.

You take a combination of Impact, Production per minute, and total production over a season.

[SASdynasty!]

PPG, APG, FG%

[Hoops Czar]

Worst: +/-
Best: RAPM - though only Harlem has access to this stat and I don't know where he stashes it.

[Robz4000]

PPG obviously. Nothing else matters.

[gospursgojas]

PPG obviously. Nothing else matters.
PPG

1. Stephen Curry, GS 30.0
2. James Harden, HOU 28.5
3. Kevin Durant, OKC 28.0
4. DeMarcus Cousins, SAC 27.0

2 of the 4 are out of playoffs

[Tully365]

PPG isn't an advanced stat...

[apalisoc_9]

PPG

1. Stephen Curry, GS 30.0
2. James Harden, HOU 28.5
3. Kevin Durant, OKC 28.0
4. DeMarcus Cousins, SAC 27.0

2 of the 4 are out of playoffs

Sarcasm..

:lmao

This is obviously a long running meltdown, but advanced stats if used individually without context and without a supplemental stat is always going to have major flaws.

[tholdren]

Sarcasm..

:lmao

This is obviously a long running meltdown, but advanced stats if used individually without context and without a supplemental stat is always going to have major flaws.

then does a stat have to have common context... i.e. is a player with the highest per on the best team better than a player with the highest per on a mediocre or worst team?

[UNT Eagles 2016]

Annual salary, tbh.

[ElNono]

With enough sample size, they're all useful... what's best/worst is largely dependent on what you're measuring...

For example, raw +/- is probably not that great to analyze individual player production on relatively small samples.

You also have to understand the context. ie: +/- is more valuable as a measure of individual talent within a roster when that team shuffles lineups the most. When you have a team with fairly static lineups, it's then a better measure of lineups than individual talent.

[spurraider21]

raw +/-

[Phenomanul]

WS Per 48 is the best, IMO.

Which places David Robinson #2 All-Time on the career list just behind Michael Jordan (and #8 All-Time on the Playoff list).

[GSH]

With enough sample size, they're all useful... what's best/worst is largely dependent on what you're measuring...

For example, raw +/- is probably not that great to analyze individual player production on relatively small samples.

You also have to understand the context. ie: +/- is more valuable as a measure of individual talent within a roster when that team shuffles lineups the most. When you have a team with fairly static lineups, it's then a better measure of lineups than individual talent.


I know what you're saying, and I know you know what you're saying. So I'm not going to disagree with any of it. Just add a little. All stats have to be used with a dose of common sense and context. So the best measure of player value is experience and BBIQ, much more so than the stats themselves.

One good example is Bonner's +/- over many seasons. On a team that is designed to move the ball and get wide-open shots, a guy (like Bonner) who can stand and deliver from the 3-P line has extra value. For most of the time in question, he was coming in against bench players where his defense was not out-matched. And he was playing with guys like Manu, who comprised the best benches in the league. So while the +/- stat supposedly showed how much the Spurs benefitted from Bonner, it had a whole lot to do with how much Bonner benefitted from playing with the Spurs. But Nono is absolutely right that it has more value over time, and intra-squad, and particularly when the team has more-static lineups. On an individual night? Looking at the +/- might lead you to conclude that Bonner had more "value" than Tim. Many times, people here said things just about that ridiculous, because his +/- was the best on the team over and over again.

RAPM is a good stat - it really is. Except that the usual point of stats is to use them predictively. RAPM got used/designed to find those "diamonds in the rough". The guys who were under-valued, and would be good targets for next year's roster. Unfortunately, it tends to confirm what you already know about the very good players, and throws up too many anomalies for the supposedly under-valued guys. Once again, common sense required. The people who love RAPM still insist that Brian Cardinal was a max-contract kind of guy "at the time the stat was run". I think that most good GM's would see through that pretty easily.

[Obstructed_View]

Clearly the one that makes people think Matt Bonner is useful.

[dabom]

I think RPM if used earlier would rank Bonner badly. Better stats for impact.

[Brazil]

There is not a single advanced stat good enough to analyze player value tbh... You have to mix it up ponderate them if not already with usage and avoid comparing positions or at least use the guard, center, forward... Then you have to consider teams, roles, specifics of a team that influence a lot a bunch of them...

Take for instance S Nash and compare his last year with Dallas and first in Phoenix, both year he was in his prime, usage there roughly the same... now you see quite a difference on his advanced metrics: per, ws per 48, vorp... (did not check rapm)... some of his metrics benefitiated greatly of the run and gun Phoenix style but that was also visible on adjusted pace metrics so it is not just a question of volume...

In terms of player he was not that different even though you could argue he had a down year in 03/04...

[DMC]

Worst: +/-
Best: RAPM - though only Harlem has access to this stat and I don't know where he stashes it.


This. +/- is useless. It's one of those "the truck asks" Bill Land and Sean Elliott queries.

[GSH]

There is not a single advanced stat good enough to analyze player value tbh... You have to mix it up ponderate them if not already with usage and avoid comparing positions or at least use the guard, center, forward... Then you have to consider teams, roles, specifics of a team that influence a lot a bunch of them...

Take for instance S Nash and compare his last year with Dallas and first in Phoenix, both year he was in his prime, usage there roughly the same... now you see quite a difference on his advanced metrics: per, ws per 48, vorp... (did not check rapm)... some of his metrics benefitiated greatly of the run and gun Phoenix style but that was also visible on adjusted pace metrics so it is not just a question of volume...

In terms of player he was not that different even though you could argue he had a down year in 03/04...


:tu Excellent example. Advanced metrics can't be used in a vacuum.

[lilbthebasedgod]

best Points.

Worst Rebounds.

Really tho, dumb question. Sure there may be an answer. But win shares per 48 is my favorite but it says Enes Kanter is the no.8 player, Kawhi is the no.2, and Valenciunas is no.10, Aldridge is no.9, which are absurdly high.

Defensive win shares it shows Tim is the best defender on the league, which is absurd. Real Plus Minus is skewed to how good your record is heavily, etc.

[Kawhitstorm]

Defensive rating b/c you can hide shitty defenders on shitty offensive players.

Offensive rating is different b/c hiding shitty offensive players is the equivalent of playing 4-on-5.

[ParadoxEN]

Which places David Robinson #2 All-Time on the career list just behind Michael Jordan (and #8 All-Time on the Playoff list).

Not an unreasonable result, IMO. D-Rob put up stats.

[tholdren]

Not an unreasonable result, IMO. D-Rob put up stats.
hes also 4th in career PER

[313]

Worst: +/-
Best: RAPM - though only Harlem has access to this stat and I don't know where he stashes it.
:lmao

[Brazil]

:tu Excellent example. Advanced metrics can't be used in a vacuum.

BTW and on a side note, you can see that those advanced metrics are more stable for bigs than for guards... Take for instance Shaq his advanced fluctuate a bit but were quite stable during his prime. Bigs production are less dependant of the system or the team they play tbh... Rebounds are a big part of numerous metrics, you can see that per 100 poss a common advanced metrics adjust the rebound numbers are very stable during a big career... Rodman who reached his prime quite late after 30 y/o piled up from 24 to 22 rpg per 100 since 91 till the end of his career. Now take a guard known for being an elite rebounder like J Kidd (6 rpg career), per 100 his numbers fluctuate from 8 to 12 ! which is quite significant, obviously his production in this category is highly dependant of the team and the roster.

[Chinook]

1) Plus-minus isn't an advanced stat.

2) The issue I have with advanced stats is that people treat them like they're objective, when they're actually mathematical opinions.

[SPURt]

1) Plus-minus isn't an advanced stat.

2) The issue I have with advanced stats is that people treat them like they're objective, when they're actually mathematical opinions.
That is the best description I've seen of advanced statistics.

[SAGirl]

I look at them all. Frankly they usually will coincide by a matter of degrees, I suppose depending on their formula. But you have to put them in context. Guys roles, do they play against benches or starters, do they get hidden on defense, assigned to non threats, do they play mostly garbage time putting up gaudy numbers like Bonner--making one or two threes in a couple of minutes of garbage time does not mean you are our best offensive player, etc.

If you don't put a context behind those stats they can be very misleading. However, if you put them in their proper context they will be useful.