Best all-in-one NBA player metric EPM and the Spurs

[rankingtear]

Heard from Zach Lowe Pod this is the best publicly available advance metrics according to NBA people.

https://dunksandthrees.com/epm

In percentile:




OFF
DEF
EPM


Jakob Poeltl
73
98
94


Derrick White
89
74
89


Dejounte Murray
76
84
84


Gorgui Dieng
73
83
82


Rudy Gay
45
95
78


Demar Derozan
95
5
72


Patty Mills
67
73
71


Devin Vassell
30
59
37


Keldon Johnson
55
15
35


Luka Samanic
17
73
34


Drew Eubanks
31
32
28


Keita Bates-Diop
12
45
21


Trey Lyles
31
13
18


Lonnie Walker
49
3
18


Tre Jones
15
21
12


Quindarry Weatherspoon
2
43
4

[Degoat]

Well that tells me Lonnie sucks lol

[BG_Spurs_Fan]

Interesting but I disagree with a lot of these. For one offense should matter wayyyyy more than defense. Also if your model shows Gorgui Dieng is better than DDR or that Poeltl is better than Donovan Mitchell and Beal then it's wrong.

[exstatic]

Well that tells me Lonnie sucks lol

Up until about 4 games ago, I would have agreed with you. At least on offense, he seems to have flipped a switch.

[KingKev]

Why even watch basketball when you have stats. Rudy all defence 3rd team and also probably time to give up on Keldon, Luka and Walker.

[exstatic]

Interesting but I disagree with a lot of these. For one offense should matter wayyyyy more than defense. Also if your model shows Gorgui Dieng is better than DDR or that Poeltl is better than Donovan Mitchell and Beal then it's wrong.

Ask Mike D’Antoni how that whole ‘only offense matters’ thing works out.

[BG_Spurs_Fan]

Ask Mike D’Antoni how that whole ‘only offense matters’ thing works out.

Well yeah but still, especially in today's NBA, a player with great offense and terrible defense (Harden, Lillard, Young) is a star, but players with great defense and terrible offense are borderline NBA. In a team concept it's a bit more complicated than simply individual player's defense or offense.

For this particular model they've clearly bumped some defense metrics to make up for it, e.g. Harden's and Tatum's defense rating being similar is a travesty.

[PrimeMinister]

God I miss Derrick white

[exstatic]

God I miss Derrick white

This. Probably our best two way player.

[R. DeMurre]

:lol The gap between Dieng and Eubanks...

Casual fan: advanced stats don't mean anything, DeRozan should be higher, him score a lotta points!

[Dejounte]

:lol The gap between Dieng and Eubanks...

Casual fan: advanced stats don't mean anything, DeRozan should be higher, him score a lotta points!

If you go to their blog, it states that players below a certain minutes threshold the data becomes unreliable. Stats are useful, but only when you know how to use them. I don't agree with the OP's presentation of these stats. It's leaving out a lot of context. The site also does a bad job of explaining its methodology. There's always two sides to this spectrum: one end says stats are useless, and the other end runs away with the data presented to them without first evaluating the data.

I will go over its flaws later today.

[R. DeMurre]

Great to see Daniel Gafford and Paul Reed ranked so high. Guys to watch for the future.

[R. DeMurre]

If you go to their blog, it states that players below a certain minutes threshold the data becomes unreliable. Stats are useful, but only when you know how to use them. I don't agree with the OP's presentation of these stats. It's leaving out a lot of context. The site also does a bad job of explaining its methodology. There's always two sides to this spectrum: one end says stats are useless, and the other end runs away with the data presented to them without first evaluating the data.

I will go over its flaws later today.

Everyone interested in stats knows that the smaller the sample size is, the less accurate the results. That's true of every stat.

[Dex]

Advanced stats seem to love Poeltl, so he must be doing something right.

Hopefully Dieng chooses to stick around. I could see him and Jak being pretty interchangeable for the first and second center spots depending on matchup.

I think the main reason he isn't getting minutes (aside from the shoulder injury) is he just doesn't know the system, and it's not like the Spurs have had any time in the practice gym in months.

That's something that could easily be solved with a full offseason and training camp, and then he should be a lock for the rotation with his skills.

[Dejounte]

Everyone interested in stats knows that the smaller the sample size is, the less accurate the results.

Yes, so you should have known not to gloat about Dieng over Eubanks.

[exstatic]

If you go to their blog, it states that players below a certain minutes threshold the data becomes unreliable. Stats are useful, but only when you know how to use them. I don't agree with the OP's presentation of these stats. It's leaving out a lot of context. The site also does a bad job of explaining its methodology. There's always two sides to this spectrum: one end says stats are useless, and the other end runs away with the data presented to them without first evaluating the data.

I will go over its flaws later today.

I’ve been doing software development for almost 35 years. Here are two statements. They’re both true.

Models are useful.

Models are wrong.

[R. DeMurre]

Yes, so you should have known not to gloat about Dieng over Eubanks.

I can gloat because he's clearly a better player. You probably agree, but would prefer to be pedantic.

[SAGirl]

Well yeah but still, especially in today's NBA, a player with great offense and terrible defense (Harden, Lillard, Young) is a star, but players with great defense and terrible offense are borderline NBA. In a team concept it's a bit more complicated than simply individual player's defense or offense.

For this particular model they've clearly bumped some defense metrics to make up for it, e.g. Harden's and Tatum's defense rating being similar is a travesty.
Agree. Watching the Justine Winslow experiment play out in Memphis definitely proves this point. He’s awful offensively. On the level of worst player in the league getting minutes this season (and he contributed a great deal to a share of recent Grizzlies losses.) his numbers by any stat reflect it. He’s been a hustler and defensive player but nothing he does on defense can compensate for how terribly he tanks his team offensively.

Maybe the injuries he’s had have taken away too much but he was never good offensively to begin with and is now worse. Meanwhile horrible black holes defensively like Bryn Forbes get minutes regularly for a playoff team and I am sure what he does offensively in small doses can help a playoff team. Yeah if you overplay him then he’s going to get exploited and cause you to lose, but he’s playable in the right situation. Current Winslow is not playable unless you want to win the tankathon.

[Dejounte]

I can gloat because he's clearly a better player. You probably agree, but would prefer to be pedantic.

Dieng could "clearly" be a better player, but not directly in the manner you chose. Being pedantic in this circumstance is not possible since I was merely holding you accountable for the misuse of the data presented. I know it's hard for people on this board to own up to being wrong, but it can be done. I've done it numerous times. Shit, I was wrong about Vassell. It's okay.

I do believe Dieng is better than Eubanks in certain situations. There's room for all three and Dieng seems to love it here. I haven't seen a player come in and have so good of a chemistry as fast as Dieng has had with his teammates. The certain situations I speak of is if the matchup calls for it. Just like how people complained about DJ's shot attempts last night in the first half, the situation called for it. He was taking what the defense was giving him, and it's usually his best shot. Unfortunately, they did not go in. Players will have their days where they should be taking a lot of shots because the defense is giving them looks. So for Dieng, he should be playing when there's not a strong big in the inside for the other team. There may have been games where this was applicable, but I point that to maybe his unfamiliarity with the offensive/ defensive schemes and he was not ready. I'm like 95% sure he'll be back next season, and he will be playing more.

[R. DeMurre]

I'm like 95% sure he'll be back next season, and he will be playing more.

It's certainly possible, but I'm not going to claim I can tell by seeing him be friendly on the sidelines with his teammate... the vast majority of players behave appropriately during game time, and interact positively with teammates. But if he does leave, I would understand why.

[Chucho]

:lol The gap between Dieng and Eubanks...

Casual fan: advanced stats don't mean anything, DeRozan should be higher, him score a lotta points!


Flip side of the coin is a self-appointed "expert fan" that believes in player potential based on advanced stats, despite said player(s) never fulfilling said potential.

lled.

[Ed Helicopter Jones]

So if DDR could play a lick of defense he'd be far and away our best player. As it is, his defense negates his superior offensive talent...basically what we all recognize already.

This metric doesn't paint a very favorable picture of Lonnie, does it? :lol






And who did Patty pay for his defensive rating? :lol

[exstatic]

So if DDR could play a lick of defense he'd be far and away our best player. As it is, his defense negates his superior offensive talent...basically what we all recognize already.

This metric doesn't paint a very favorable picture of Lonnie, does it? :lol





And who did Patty pay for his defensive rating? :lol

Lonnie didn’t deserve any favorability until the last 4-5 games. He played scared until very recently. That defense is DeRozan level, though. Yikes.

[timvp]

Good stuff, tbh :tu

The two best offensive players on the team are DeRozan and White and the two worst defenders are DeRozan and Walker -- so that checks out. Gay's numbers are a bit surprising. It seems like he's a better offensive player than defensive player. Then again, advanced stats have always graded his defense a lot higher than an eye-test would predict. That's somewhat understandable because his length can usually make up for his slow feet -- even though the slowness of his feet stands out.

[TD 21]

There's no logical explanation for Eubanks over Dieng and if there wasn't a prearranged deal to extend him in the off season, I don't know why he'd bother re-signing with these senile idiots.

The latter is a much more well rounded, high IQ player (it was refreshing to watch the final 8 minutes yesterday and see a group with decent versatility, who knew how to play the game and were doing so without agenda), two things they've developed an aversion to.

[Slippy]

Good stuff, tbh :tu

The two best offensive players on the team are DeRozan and White and the two worst defenders are DeRozan and Walker -- so that checks out. Gay's numbers are a bit surprising. It seems like he's a better offensive player than defensive player. Then again, advanced stats have always graded his defense a lot higher than an eye-test would predict. That's somewhat understandable because his length can usually make up for his slow feet -- even though the slowness of his feet stands out.

Just curious. Do these advanced stats take into acoount the opposition man you're guarding becsuse we all see Demar gets hidden mostly since he terrible on d while Lonnie guarding the oppositions best man on man happens more often than not.

[R. DeMurre]

Flip side of the coin is a self-appointed "expert fan" that believes in player potential based on advanced stats, despite said player(s) never fulfilling said potential.

lled.

? Not sure what you mean... are you saying Dieng hasn't lived up to being being a 21st pick?

[timvp]

Just curious. Do these advanced stats take into acoount the opposition man you're guarding becsuse we all see Demar gets hidden mostly since he terrible on d while Lonnie guarding the oppositions best man on man happens more often than not.

Depends on which advanced stats you're talking about. Some of them do. I haven't dug into this one.

What you said is true -- but it's clear why that's the case. DeRozan is a terrible individual defender and a not-quite-as-terrible help defender, so hiding him is always the best idea. Walker is an atrocious help defender (tbh, probably worse than DeRozan). Walker isn't a great individual defender but the best way to "hide" him is by putting him on a top-notch scorer and telling him not to rotate off of him. That allows him to think less and limits any complexities.

[Jsmythe]

For plus-minus stats, I find the ESPN RPM (Real Plus-Minus) stat more closely aligned to reality. The ESPN one accounts for who was else on the court (both on your own team and the opponent), which means that bench players are rated appropriately for playing against lesser competition.

These are the Spurs ratings in ESPN RPM:

Derrick White (1.34)
Dejounte Murray (1.27)
Jakob Poeltl (0.33)
Keldon Johnson (0.16)
Rudy Gay (-0.20)
Patty Mills (-0.66)
DeMar DeRozan (-0.79)
Trey Lyles (-1.10)
Gorgui Dieng (-1.27)
Devin Vassell (-1.77)
Drew Eubanks (-1.89)
Luka Samanic (-1.94)
Keita Bates-Diop (-2.09)
Lonnie Walker IV (-2.14)
Quinddary Weatherspoon (-2.61)
Tre Jones (-3.70)

Source: http://www.espn.com/nba/statistics/rpm

[DAF86]

Interesting but I disagree with a lot of these. For one offense should matter wayyyyy more than defense. Also if your model shows Gorgui Dieng is better than DDR or that Poeltl is better than Donovan Mitchell and Beal then it's wrong.

Why?

[Allan Rowe vs Wade]

Well that tells me Lonnie sucks lol

lonnie’s advanced stats are :vomit:

[spurraider21]

Ask Mike D’Antoni how that whole ‘only offense matters’ thing works out.
Great coach who was ahead of his time

[Chinook]

Great coach who was ahead of his time

Damned straight. Just because the Spurs had a GOAT candidate coach doesn't mean MDA is bad. Probably a top 5-10 coach of the past 25 years.

[Jsmythe]

I was curious about how EPM differs from RPM, and I found this blog article where the EPM creators compared EPM with other all-in-one stats to show that EPM had better predictive value than the other stats. But given the fact that EPM was basically designed to win this metric, and that RPM came a close 2nd, I still think RPM is more realistic. But I'll add EPM to the my list of "other stats to check" (such as PER, Wins Produced, etc).

Blog article comparing all-in-one stats: https://dunksandthrees.com/blog/metric-comparison

Wins Produced per 48, if you're interested: https://www.boxscoregeeks.com/teams/sas?direction=desc&sort=per48_wins_produced