RealGM

Thinking about the Concept of Starter Disadvantage

I think it is interesting to think about how well players do in situations where you’d expect their team to be at a disadvantage. There’s some minutes in a basketball game where another team has a real advantage, because they’ve got most or all of their starters out there, while the opposing team doesn’t. These are inherently difficult situations for the disadvantaged team. And I find it interesting to consider how well players do in those scenarios. It seems like an interesting lens through which to look at the concept of floor raising, since it looks at how a player elevates their team when their team probably should be losing the minutes.

Very helpfully, PBPstats has a WOWY function that allows us to look at “Starter State,” so we can actually look at what happened in just situations where the opposing team has more starters on the floor.

Case Study: Manu Ginobili at a Starter Disadvantage

I ran these numbers for some players, and I think the most interesting thing I found was Manu Ginobili.

Ginobili’s Starter-Disadvantage Numbers

In the 15 years from 2003-2017, Ginobili played 21,111 possessions where his team had fewer starters on the floor than the other team. In those possessions, they outscored opponents by 1,985 points. This is a net rating of +9.40, at a starter disadvantage with Ginobili! Which is really remarkable for a disadvantaged situation. And, of course, there are shorter time periods within that that are even higher—for instance, in the five years from 2004-2008, the Spurs were +11.26 while at a starter disadvantage with Ginobili on the floor! The Spurs really killed opposing teams when they were at a starter disadvantage, but had Ginobili on the floor. This probably is a good reason why Popovich often used him the way he did, off the bench.

Comparisons to Great Players

1. Perhaps most pertinently, the Spurs themselves from 2003-2016 were +8.01 at a starter disadvantage with Tim Duncan on the floor (and it is +8.09 if we stretch back to 2001, which is the first year PBPstats has this data).

2. From 2009-2020, LeBron’s teams were +8.25 with LeBron on the court and his team at a starter disadvantage (and other longer time periods make this go down), with a five-year peak of +11.38 in 2009-2013. If we look at 2010-2024, when LeBron was the same age as Ginobili was in 2003-2017, it is +6.80 for LeBron.

3. Giannis’s Bucks have been +8.17 at a starter disadvantage with Giannis on the floor from 2019-2024. It goes down to +5.36 if we go back to 2017.

4. Jokic’s Nuggets have been +6.76 in the last five years in this situation. However, it’s down to +4.99 if we go back to 2017.

5. Steph’s Warriors were a massive +10.11 from 2014-2024 at a starter disadvantage with Steph on the floor, with a five-year peak of +14.11 in 2015-2019.

Those are the guys I ran these numbers for. So, with the exception of Steph being slightly ahead, Ginobili’s net rating at a starter disadvantage is actually above guys that we’d consider the best of the last couple generations—including his own all-time great teammate. Which I found really interesting.

Adjusting for Ginobili not Always Being a Starter

Astute readers will probably note that Ginobili himself often not being a starter skews these sorts of comparisons, because Ginobili is effectively counting towards the purported disadvantage. And that does help Ginobili here. But we have the ability to account for that, by only counting situations where either Ginobili was a starter and his team was at a starter disadvantage or where Ginobili was not a starter and his team was at at least a two-starter disadvantage. Basically, this filters things down to situations where the Spurs were at a starter disadvantage even if we treat Ginobili as a starter for these purposes. Treating Ginobili as a starter for these purposes, from 2003-2017, the Spurs were still +8.27 at a starter disadvantage with Ginobili! This is still up there very slightly above guys like prime LeBron and Giannis, and even slightly above Duncan’s number (even though I didn’t adjust Duncan’s number for Ginobili being a bit of a faux bench player). So, when we look at how players’ teams do with a starter disadvantage, Manu Ginobili looks like an all-time great IMO!

Adjusting for Playing with Great Teammates: Ginobili’s Numbers

But, of course, people might validly point out that Ginobili had great players on his team. Is his team really at a disadvantage if they have fewer starters on the floor, but some of their starters that *are* on the floor include Duncan and Parker? Perhaps not.

So, to try to filter things down further, I looked at how the Spurs did with Ginobili on the floor at a starter disadvantage (and, as per the above, treating Ginobili as a starter for these purposes), and also without Duncan, Parker, Robinson, and 2014-onwards Kawhi on the floor. This does lower the sample size a decent bit, such that we are only looking at around five thousand possessions, but in those scenarios, the Spurs had a +6.00 net rating from 2003-2017, at a starter disadvantage and Ginobili on and Duncan, Parker, Robinson, and 2014-onwards Kawhi all off!

And let’s just remember what we are measuring at this point. This is what the Spurs did with Ginobili on the court, and the other team having more starters on than the Spurs, and the Spurs not having their other best players of the era on the court with Ginobili. So it’s basically just Ginobili + role players, against opposing lineups that are going to mostly be starters. And the Spurs still averaged a +6 net rating! This is really impressive stuff IMO!

Adjusting for Playing with Great Teammates: Comparison to Great Players

How does that compare to other players if we made the same kind of adjustment?

1. Well, let’s first take Duncan, since he seems like the most pertinent one. From 2003-2016, the Spurs were only +2.64 when at a starter disadvantage with Duncan on the floor and no Ginobili, Parker, Robinson, or 2014-onwards Kawhi. They did extremely well in those sorts of minutes in 2002, so if we extend it back to 2001-2016, that actually goes up to +5.85. But, even then, it’s still slightly below what the Spurs were doing at a starter disadvantage and no other stars, with Ginobili! And I’ll note that I also ran these numbers just focusing on Duncan and Ginobili themselves. In 2003-2016 with Duncan on with a starter disadvantage and Ginobili off, the Spurs had a +3.26 net rating. In those same years, with Ginobili on with a starter disadvantage (always counting Ginobili as a starter) and Duncan off, the Spurs had a +5.01 net rating (and it’s +5.15 if we add in 2017, when Duncan was no longer there; note also that these numbers are +6.39 and +6.93 if we didn’t always count Ginobili as a starter). Duncan’s number goes up to +4.74 if we counted 2001 and 2002 as well, but obviously that’s pre-Ginobili.

What about other guys?

2. Well, Steph is an interesting one, because his raw numbers without this sort of adjustment are the highest of anyone. However, if we adjust for just these situations where Draymond and Durant weren’t on the floor, we get +4.26 from 2014-2024. And even in that peak 2015-2019 period, it is just +4.72 (though the number of possessions is below a thousand, so that’s really noisy).

3. LeBron is another interesting one. If we look at LeBron on the Heat and second-stint Cavs (i.e. 2011-2018), but without Wade, Bosh, Kyrie, or Love, his teams were just +0.97 at a starter disadvantage with LeBron. The Lakers did super well in 2020 with LeBron at a starter disadvantage without Anthony Davis though, so if we throw that year in too (to basically sweep in all prime years with great teammates), it goes up to +3.00. Including the first-stint Cavs in this particular analysis is a bit weird since there’s honestly just no obvious teammates to adjust for, but doing no adjustment at all probably skews the resulting lineups in those years as being overly good for these purposes since we aren’t taking away the other couple best players on the team). But if we take 2010-2024 (i.e. the same age as Ginobili from 2003-2017) and don’t filter out any teammate on the first-stint Cavs, we get +4.30 for LeBron at a starter disadvantage. If we add 2009 to that because it was a good year for these purposes and not including it basically just punishes LeBron for the fact that Ginobili didn’t come to the NBA earlier, then we get +4.83. And going earlier than that just makes the number go down. We can pick out shorter time periods for LeBron that are better for these purposes (for instance, 2009-2020 is +6.62). But, Ginobili has smaller time periods that are higher than his 2003-2017 average too (for instance, 2006-2017 is +7.81).

4. With Giannis, he actually has gone a fantastic +8.17 in the 2019-2024 time period, at a starter disadvantage without Middleton, Holiday, or Lillard. But the sample size there isn’t all that big, and it goes down to +4.85 if we go back to 2017.

5. Similarly, for Jokic, he’s been a fantastic +7.11 at a starter disadvantage in the last 5 years without Murray, but if we go back to 2017 then that’s down to +4.35.

So, overall, I’d say that if we look at how players’ teams did with them with a starter disadvantage and other stars off the court, Manu Ginobili actually looks right at home amongst the very best players of the last couple generations! Indeed, I’d say he’s arguably the top guy in this analysis!

Conclusion and Some Caveats

The conclusion is that I think this is an interesting lens to look through, to see how players raised the floor of role-player-centric lineups in tough situations. And Manu Ginobili looks great with it! And this perhaps shows why Popovich used him so much with bench lineups. He did great with them!

I’d add a few caveats to this analysis though:

First of all, the sample sizes here aren’t massive, particularly when we start filtering out possessions with other stars on the floor. Once we drill down like that, we’re often talking about only a few thousand possessions, which is around the noisiness of a single-season “on” value. So I would take the exact numbers here with an appropriate grain of salt.

Second of all, obviously none of this analysis corrects for the precise quality of the role players or bench guys on each team. The Spurs did often have very good role players (especially towards the end of Manu’s career), so that helps. It’s hard to really gauge the effect of that, but it’s something to keep in mind. We could perhaps try to look at how these teams did in these same sorts of situations but without these players on the court, but then that would be highly affected by whether stars were staggered in those sorts of minutes, and if we try to correct for that I think we’d be looking at really tiny “off” samples. I’m also just less concerned with “impact” per se with this sort of thing (since, especially with role player lineups, that probably gets more to whether they have gaping lineup holes that the guy fills than it does exactly how good those role players actually are), and more just how players actually did when on the court in these sorts of situations.

Relatedly, obviously this is just getting at something that is part of the data that goes into RAPM. So it’s not really independent of that. It’s just zeroing in on one particular type of scenario, to look at how players did at a disadvantage.

Finally, as it relates to Ginobili, there’s always the question of whether things would’ve been as good with him on the court if he played more minutes than he did.

So, this is the first I've heard of this on PBP and I'm exploring, but your post makes it sound like you've got like a table of numbers you're looking that we can't see. If you have something you've constructed, I'd love to see it ltj!

So Manu is a known great player kinda kept down by his lower minutes totals. None of this should be too surprising.

One factor to look at him vs Duncan is that Duncan was always a starter where as Manu was frequently a bench player. The bench units being both something Manu more familiar with playing with, but also you structure your starting lineup to maximize its effectiveness whereas Duncan with bench guys is going to be less optimized than it is for Manu.

To take nothing away from Manu, just some possible explanations that your data can't fully account for.

Doctor MJ wrote:So, this is the first I've heard of this on PBP and I'm exploring, but your post makes it sound like you've got like a table of numbers you're looking that we can't see. If you have something you've constructed, I'd love to see it ltj!

I do have numbers compiled, but they’re a bit of a mess and in a file I’ve got a bunch of other things in too. But I can reproduce below the data I have compiled (note: for some players I bothered to compile more years than I talked about above, while for others I didn’t—the below is just what I have for everyone):

Plus-Minus at Starter Disadvantage

Manu Ginobili

2003: +65 in 1102 possessions
2004: +137 in 1282 possessions
2005: +76 in 773 possessions
2006: +177 in 901 possessions
2007: +210 in 1645 possessions
2008: +149 in 2053 possessions
2009: +36 in 1179 possessions
2010: +100 in 2111 possessions
2011: +73 in 1228 possessions
2012: +98 in 879 possessions
2013: +152 in 1429 possessions
2014: +238 in 1807 possessions
2015: +183 in 1876 possessions
2016: +126 in 1187 possessions
2017: +165 in 1659 possessions
2018: +51 in 1690 possessions

Manu Ginobili - Always Treating Ginobili as a Starter

2003: -5 in 460 possessions
2004: +32 in 687 possessions
2005: +76 in 773 possessions
2006: +139 in 774 possessions
2007: +151 in 1030 possessions
2008: +41 in 1146 possessions
2009: +10 in 580 possessions
2010: +81 in 1117 possessions
2011: +70 in 1223 possessions
2012: +56 in 547 possessions
2013: +28 in 545 possessions
2014: +127 in 853 possessions
2015: +83 in 955 possessions
2016: +44 in 363 possessions
2017: +53 in 876 possessions
2018: +53 in 929 possessions

Tim Duncan

2001: +103 in 1552 possessions
2002: +185 in 1804 possessions
2003: +154 in 1413 possessions
2004: +133 in 1233 possessions
2005: +83 in 1173 possessions
2006: +181 in 1448 possessions
2007: +183 in 1591 possessions
2008: +121 in 1602 possessions
2009: +45 in 1774 possessions
2010: +94 in 1769 possessions
2011: +70 in 1252 possessions
2012: +101 in 1017 possessions
2013: +141 in 1145 possessions
2014: +71 in 1533 possessions
2015: +65 in 1496 possessions
2016: +68 in 415 possessions

LeBron James

2004: -75 in 1380 possessions
2005: -21 in 1421 possessions
2006: +59 in 1776 possessions
2007: +132 in 1597 possessions
2008: +39 in 1980 possessions
2009: +125 in 1298 possessions
2010: +342 in 2115 possessions
2011: +109 in 1792 possessions
2012: +84 in 1215 possessions
2013: +231 in 1409 possessions
2014: +85 in 1709 possessions
2015: +58 in 894 possessions
2016: +58 in 1238 possessions
2017: +55 in 1100 possessions
2018: +36 in 1811 possessions
2019: +24 in 879 possessions
2020: +190 in 1466 possessions
2021: +50 in 912 possessions
2022: -23 in 1401 possessions
2023: +38 in 1181 possessions
2024: +36 in 1073 possessions

Steph Curry

2014: +43 in 805 possessions
2015: +132 in 1130 possessions
2016: +181 in 956 possessions
2017: +124 in 1141 possessions
2018: +88 in 845 possessions
2019: +192 in 1008 possessions
2020: -33 in 98 possessions
2021: +120 in 853 possessions
2022: +148 in 1408 possessions
2023: -11 in 977 possessions
2024: +122 in 1717 possessions

Nikola Jokic

2017: +20 in 909 possessions
2018: +65 in 484 possessions
2019: -31 in 841 possessions
2020: +76 in 548 possessions
2021: +25 in 760 possessions
2022: +16 in 658 possessions
2023: +31 in 540 possessions
2024: +71 in 734 possessions

Giannis Antetokounmpo

2017: +14 in 1713 possessions
2018: -46 in 1148 possessions
2019: +189 in 1103 possessions
2020: +108 in 973 possessions
2021: +22 in 899 possessions
2022: +96 in 1186 possessions
2023: +135 in 1219 possessions
2024: -11 in 1216 possessions

Plus-Minus at Starter Disadvantage - Filtering out minutes with star teammates

Manu Ginobili - Without Duncan, Parker, Robinson, and 2014-onwards Kawhi (and always treating Ginobili as a starter)

2003: -14 in 100 possessions
2004: -19 in 189 possessions
2005: -7 in 215 possessions
2006: +24 in 256 possessions
2007: +28 in 385 possessions
2008: -33 in 420 possessions
2009: -12 in 116 possessions
2010: +19 in 368 possessions
2011: +35 in 499 possessions
2012: +23 in 244 possessions
2013: +8 in 264 possessions
2014: +85 in 299 possessions
2015: +54 in 413 possessions
2016: +24 in 191 possessions
2017: +48 in 424 possessions
2018: +45 in 567 possessions

Tim Duncan - Without Ginobili, Parker, Robinson, and 2014-onwards Kawhi

2001: +42 in 1056 possessions
2002: +148 in 841 possessions
2003: +6 in 220 possessions
2004: -2 in 184 possessions
2005: +3 in 180 possessions
2006: +18 in 298 possessions
2007: +22 in 226 possessions
2008: -14 in 156 possessions
2009: +0 in 397 possessions
2010: -25 in 206 possessions
2011: +12 in 73 possessions
2012: +29 in 162 possessions
2013: +13 in 150 possessions
2014: +0 in 105 possessions
2015: +7 in 68 possessions
2016: -4 in 36 possessions

LeBron James - Without Wade, Bosh, Kyrie, Love, and Davis (for 2010 and earlier, as well as 2019, no star player filtering out was done, so see the above entries for LeBron in those years)

2011: +43 in 573 possessions
2012: -6 in 548 possessions
2013: +64 in 630 possessions
2014: -20 in 635 possessions
2015: +2 in 207 possessions
2016: -2 in 424 possessions
2017: -11 in 563 possessions
2018: -23 in 1273 possessions
2019:
2020: +125 in 877 possessions
2021: +43 in 756 possessions
2022: -45 in 1102 possessions
2023: -29 in 737 possessions
2024: +1 in 487 possessions

Stephen Curry - Without Draymond and Durant

2014: +3 in 307 possessions
2015: +22 in 366 possessions
2016: -11 in 136 possessions
2017: +1 in 211 possessions
2018: +40 in 107 possessions
2019: -8 in 112 possessions
2020: +13 in 50 possessions
2021: -35 in 216 possessions
2022: +94 in 1082 possessions
2023: -37 in 446 possessions
2024: +93 in 1077 possessions

Nikola Jokic - Without Murray

2017: +38 in 368 possessions
2018: +27 in 251 possessions
2019: -84 in 383 possessions
2020: +56 in 294 possessions
2021: +10 in 417 possessions
2022: +16 in 658 possessions
2023: +8 in 315 possessions
2024: +71 in 580 possessions

Giannis Antetokounmpo - Without Middleton, Holiday, or Lillard

2017: +8 in 1324 possessions
2018: -16 in 535 possessions
2019: +121 in 406 possessions
2020: +91 in 573 possessions
2021: -22 in 215 possessions
2022: +51 in 294 possessions
2023: +3 in 527 possessions
2024: -20 in 583 possessions

___________________________

I would caution again that the sample sizes on these can be low. They’re definitely low sample size on a season-by-season basis, and for the stuff that filters out other star players they’re arguably just really low even in multi-year spans, especially for the guys that there’s not a ton of years for.

__________________________

EDIT: I forgot to add here the data I have on just Duncan and Ginobili with the other off the court. Here’s what I have for that:

Plus Minus at Starter Disadvantage - Duncan & Ginobili without the other

Manu Ginobili - Without Duncan

2003: -30 in 317 possessions
2004: -21 in 556 possessions
2005: +10 in 328 possessions
2006: +54 in 456 possessions
2007: +62 in 733 possessions
2008: +8 in 874 possessions
2009: -49 in 434 possessions
2010: -9 in 827 possessions
2011: +35 in 619 possessions
2012: +68 in 536 possessions
2013: +74 in 740 possessions
2014: +180 in 879 possessions
2015: +143 in 1001 possessions
2016: +68 in 982 possessions
2017: +165 in 1659 possessions
2018: +51 in 1690 possessions

Manu Ginobili - Without Duncan (and always treating Ginobili as a starter)

2003: -13 in 151 possessions
2004: -14 in 305 possessions
2005: +10 in 328 possessions
2006: +40 in 422 possessions
2007: +54 in 482 possessions
2008: -29 in 535 possessions
2009: -39 in 212 possessions
2010: +6 in 463 possessions
2011: +32 in 614 possessions
2012: +55 in 358 possessions
2013: +2 in 320 possessions
2014: +84 in 406 possessions
2015: +49 in 563 possessions
2016: +38 in 333 possessions
2017: +53 in 876 possessions
2018: +53 in 929 possessions

Tim Duncan - Without Ginobili

2001: +103 in 1552 possessions
2002: +185 in 1804 possessions
2003: +59 in 628 possessions
2004: -25 in 507 possessions
2005: +17 in 728 possessions
2006: +58 in 1003 possessions
2007: +35 in 679 possessions
2008: -20 in 423 possessions
2009: -40 in 1029 possessions
2010: -15 in 485 possessions
2011: +32 in 643 possessions
2012: +71 in 674 possessions
2013: +63 in 456 possessions
2014: +13 in 605 possessions
2015: +25 in 621 possessions
2016: +10 in 210 possessions

Texas Chuck wrote:So Manu is a known great player kinda kept down by his lower minutes totals. None of this should be too surprising.

One factor to look at him vs Duncan is that Duncan was always a starter where as Manu was frequently a bench player. The bench units being both something Manu more familiar with playing with, but also you structure your starting lineup to maximize its effectiveness whereas Duncan with bench guys is going to be less optimized than it is for Manu.

To take nothing away from Manu, just some possible explanations that your data can't fully account for.

I think this is a fair point, and if you look at the raw data I reported in my post above, we see that it is actually the case that Duncan tends to have relatively few minutes at a starter disadvantage without any of the Spurs’ other stars with him. He was definitely skewed away from bench lineups.

But yeah, I’ll note that it really is amazing how much Ginobili + bench units destroyed teams in his later years even at a starter disadvantage. For instance, from 2014-2018, the Spurs with Ginobili at a starter disadvantage (counting Ginobili as a starter) and with none of the other star Spurs on the court went +256 in 1894 possessions, which is +13.52 per 100 possessions. Even if we expand that out to 2010-2018, we’re looking at +341 in 3269 possessions, which is +10.43 per 100 possessions. Pretty breathtaking stuff. The team didn’t do nearly as well with Ginobili at a starter disadvantage and no other Spurs on during the earlier years of Ginobili’s career. And that’s despite the fact that we’d generally think that, if anything, Ginobili was at his best in some of those earlier years. Some of that is probably just a result of random variance. But perhaps this was partly a result of the Spurs building their bench around Ginobili as the years went by, resulting in him being even more effective with them.

How can you consider Kevin Love a star from 2016-2018 yet not consider Brook Lopez, Klay Thompson or Andre Iguodala stars?

These numbers also greatly benefit the teams who are super deep 8-12 in the rotation, while hurting teams who have more higher-salaried players near the top since those teams will inevitably have less depth and more minimum-level players.

It also benefits teams who went deep into the luxury tax to retain lesser bench pieces [See Golden State] and hurt teams who avoided the Tax or 2nd Apron.

Colbinii wrote:How can you consider Kevin Love a star from 2016-2018 yet not consider Brook Lopez, Klay Thompson or Andre Iguodala stars?

These numbers also greatly benefit the teams who are super deep 8-12 in the rotation, while hurting teams who have more higher-salaried players near the top since those teams will inevitably have less depth and more minimum-level players.

It also benefits teams who went deep into the luxury tax to retain lesser bench pieces [See Golden State] and hurt teams who avoided the Tax or 2nd Apron.

You know why

Reminds me of another poster who came up with WOWY “adjustments” for LeBron that included 2011 Shaq but of course not 2015 Varejao who got injured as well as expansion caveats for early ‘70s Kareem but not ‘90s Jordan.

Colbinii wrote:How can you consider Kevin Love a star from 2016-2018 yet not consider Brook Lopez, Klay Thompson or Andre Iguodala stars?

I think that including Kevin Love and not those guys is amply justified by the fact that he’s a superior basketball player, and has been recognized as such. The only arguable one is Klay—who actually is a multi-time all-star, who made all-NBA, etc. I didn’t add Klay to the mix because in my experience from running lineup numbers for the Warriors, the team tends to actually get better if you take Klay out. So adding Klay seemed to me like it’d just lower an already-somewhat-small sample size to take out a guy who isn’t actually a positive. For instance, leaving aside the starter disadvantage stuff, from 2014-2024, in minutes with Steph on and Draymond and Durant off, the Warriors did better if Klay was off too than they did with Klay on.

And, indeed, since you brought it up, I just ran the numbers, and if you filter Steph’s starter disadvantage numbers down to just minutes without Durant, Draymond, AND Klay, then his 2014-2024 net rating at a starter disadvantage is +6.80 instead of +4.26. So taking out Klay actually catapults Steph up towards the top, rather than making his numbers look worse. But it also cuts the sample size almost in half. I prefer the Durant + Draymond numbers, because the sample size is bigger and because I personally think Klay has always just been a way-over-hyped role player anyways. But if you like Klay more than I do (and most people do!), you’re welcome to prefer that +6.80 number for Steph instead.

These numbers also greatly benefit the teams who are super deep 8-12 in the rotation, while hurting teams who have more higher-salaried players near the top since those teams will inevitably have less depth and more minimum-level players.

Maybe, though I suspect that a lot of these minutes are with role-player starters and top of the bench guys, rather than with the 8-12 guys. Remember, this isn’t just minutes with 4 bench guys. It’s any minutes at a starter disadvantage. A starter disadvantage can be something like your team having 3 or 4 starters on the court and the other team having all 5 starters on. So nothing here really requires these minutes to be with the 8-12 guys, and I imagine we’re largely talking about minutes taken up by other guys (and obviously that’s even more the case in the versions of the numbers that don’t filter out other star players).

But yes, this generally goes to the second caveat I specifically mentioned in my OP—which is that “none of this analysis corrects for the precise quality of the role players or bench guys on each team.”

It also benefits teams who went deep into the luxury tax to retain lesser bench pieces [See Golden State] and hurt teams who avoided the Tax or 2nd Apron.

Yeah, that again goes to that same second caveat I mentioned in my OP. Not sure the Warriors are the best example of this, since they actually got a lot of lesser bench guys on veteran minimums, but I’m not an expert on NBA teams’ salary decisions with bench guys. The bigger thing with bench players is probably more about getting lucky with guys that actually work out—which is probably only loosely related to salary but does also go to my second caveat.

homecourtloss wrote:

Colbinii wrote:How can you consider Kevin Love a star from 2016-2018 yet not consider Brook Lopez, Klay Thompson or Andre Iguodala stars?

These numbers also greatly benefit the teams who are super deep 8-12 in the rotation, while hurting teams who have more higher-salaried players near the top since those teams will inevitably have less depth and more minimum-level players.

It also benefits teams who went deep into the luxury tax to retain lesser bench pieces [See Golden State] and hurt teams who avoided the Tax or 2nd Apron.

You know why

Reminds me of another poster who came up with WOWY “adjustments” for LeBron that included 2011 Shaq but of course not 2015 Varejao who got injured as well as expansion caveats for early ‘70s Kareem but not ‘90s Jordan.

Given this sarcastic comment, I hope you read my post I made just below yours, see that me not including Klay Thompson actually *hurt* Steph’s number a good deal, and then rethink your completely unfair and baseless assumption that any data I’ve posted is optimally formulated to look as good as possible for the players I like the most.

There’s a reason why RAPM variants love Manu

Very cool data, haven't seen this sort of thing analyzed before so I appreciate you sharing.

This sort of thing is definitely tricky because of all the potential adjustments you can make; the Klay thing, as has been pointed out, and the fact that removing him makes Steph's numbers look far better (and decreases the sample size further to like, half a season's worth of possessions I think?). Lot of different ways to try to adjust things and get very different values. Of course it seems like no matter how you slice it, Manu's numbers are ridiculously impressive.

I calculated Steph's numbers with Klay off the court, but those have already been posted. Also LeBron 2009-13: net rating with a starter disadvantage and sans Wade/Bosh is +10.99.

lessthanjake wrote:

Colbinii wrote:How can you consider Kevin Love a star from 2016-2018 yet not consider Brook Lopez, Klay Thompson or Andre Iguodala stars?

I think that including Kevin Love and not those guys is amply justified by the fact that he’s a superior basketball player, and has been recognized as such. The only arguable one is Klay—who actually is a multi-time all-star, who made all-NBA, etc. I didn’t add Klay to the mix because in my experience from running lineup numbers for the Warriors, the team tends to actually get better if you take Klay out. So adding Klay seemed to me like it’d just lower an already-somewhat-small sample size to take out a guy who isn’t actually a positive. For instance, leaving aside the starter disadvantage stuff, from 2014-2024, in minutes with Steph on and Draymond and Durant off, the Warriors did better if Klay was off too than they did with Klay on.

And, indeed, since you brought it up, I just ran the numbers, and if you filter Steph’s starter disadvantage numbers down to just minutes without Durant, Draymond, AND Klay, then his 2014-2024 net rating at a starter disadvantage is +6.80 instead of +4.26. So taking out Klay actually catapults Steph up towards the top, rather than making his numbers look worse. But it also cuts the sample size almost in half. I prefer the Durant + Draymond numbers, because the sample size is bigger and because I personally think Klay has always just been a way-over-hyped role player anyways. But if you like Klay more than I do (and most people do!), you’re welcome to prefer that +6.80 number for Steph instead.

These numbers also greatly benefit the teams who are super deep 8-12 in the rotation, while hurting teams who have more higher-salaried players near the top since those teams will inevitably have less depth and more minimum-level players.

Maybe, though I suspect that a lot of these minutes are with role-player starters and top of the bench guys, rather than with the 8-12 guys. Remember, this isn’t just minutes with 4 bench guys. It’s any minutes at a starter disadvantage. A starter disadvantage can be something like your team having 3 or 4 starters on the court and the other team having all 5 starters on. So nothing here really requires these minutes to be with the 8-12 guys, and I imagine we’re largely talking about minutes taken up by other guys.

But yes, this generally goes to the second caveat I specifically mentioned in my OP—which is that “none of this analysis corrects for the precise quality of the role players or bench guys on each team.”

It also benefits teams who went deep into the luxury tax to retain lesser bench pieces [See Golden State] and hurt teams who avoided the Tax or 2nd Apron.

Yeah, that again goes to that same second caveat I mentioned in my OP. Not sure the Warriors are the best example of this, since they actually got a lot of lesser bench guys on veteran minimums, but I’m not an expert on NBA teams’ salary decisions with bench guys. The bigger thing with bench players is probably more about getting lucky with guys that actually work out—which is probably only loosely related to salary but does also go to my second caveat.

I guess I still don't understand why you wouldn't include other "stars".

Like, is 2014 Wade or Bosh really that much better than 2021 Lopez or 2016 Klay? Probably not.

It seems like you should just include everyone--stars included--and provide those numbers as well.

lessthanjake wrote:

homecourtloss wrote:

Colbinii wrote:How can you consider Kevin Love a star from 2016-2018 yet not consider Brook Lopez, Klay Thompson or Andre Iguodala stars?

These numbers also greatly benefit the teams who are super deep 8-12 in the rotation, while hurting teams who have more higher-salaried players near the top since those teams will inevitably have less depth and more minimum-level players.

It also benefits teams who went deep into the luxury tax to retain lesser bench pieces [See Golden State] and hurt teams who avoided the Tax or 2nd Apron.

You know why

Reminds me of another poster who came up with WOWY “adjustments” for LeBron that included 2011 Shaq but of course not 2015 Varejao who got injured as well as expansion caveats for early ‘70s Kareem but not ‘90s Jordan.

Given this sarcastic comment, I hope you read my post I made just below yours, see that me not including Klay Thompson actually *hurt* Steph’s number a good deal, and then rethink your completely unfair and baseless assumption that anything I’ve posted is optimally formulated to look as good as possible for the players I like the most.

The continuous baseless assumptions and accusations of intentional bias is so corrosive to this forum. I genuinely have no idea why people do this

For context, that poster's referencing my team change WOWY database -- a database which I spent many hours of my free time putting together for free for the benefit of people on this forum. In that thread
.. .I explicitly stated in the post in question that it's a work in progress, and that feedback/advice/volunteering to help are welcome.
... I explicitly state that the contextual adjustments, including the ones they called out such as the expansion caveats and the multi-year 2011 LeBron team change, are optional, done for a first-pass attempt to try to correct for the obvious and crippling limitations of raw WOWY,
... *and I explicitly include the raw data and encourage people who prefer no contextual adjustments to refer to the raw data*,
... *and I explicitly explained this to them at the time of the time of the post

Like... basically everything they accused me of was explicitly addressed already.

And yet for some reason, they apparently hold such a grudge about this that they chose to randomly **** on me, calling me out for supposed intentional bias, bringing it up at a time when it's completely irrelevant, a full year later. It's mind boggling.

There's just such a basic lack of reading comprehension (we both explicitly addressed their complaint in prior posts!) and such a basic lack of respect (why do we need to resort to sarcasm and accusations of bad intentions for situations that pretty blatantly don't deserve this??).

I mean, if I chose not to read or comprehend blatant statements in someone else's posts, a post that someone clearly spent immense amounts of time and effort working on purely for the benefit of the community, and simultaneously had the gall to think that their post was so blatantly wrong that it was at all reasonable to think it was intentionally misleading people (while again simultaneously ignoring the statements in the post that actually addressed my concerns)... maybe my first response wouldn't be to publicly mock them and accuse them of intentional bias ? I would like to hope I would have the humility to just ask what their intentions are first. "Hey, it seems like you're missing something, am I misinterpreting?" Like how hard is that to post?

And it has genuinely bad consequences for the quality of this forum! It offends well-intentioned people, as they have just offended me. It discourages research, encourages a lack of effort, and a decreases the quality of the forum. It encourages people to leave the forum -- as it did Squared2020, once one of the best historical researchers on the internet, who quit the forum and potentially his research after a buildup forum toxicity and a lack of human empathy, not too dissimilar to this post above. And it totally misleads people who are new to the forum into thinking the OP actually is genuinely biased, which given the context of the situation, is pretty unfounded and totally ridiculous.

Perhaps my post here is a slight overreaction given the scope of the post above. But it's not that hard to just treat people with decency. I'm not asking too much. And there's a history of this kind of behavior on this forum, and it's just unacceptable. This stuff is genuinely harmful for the forum!

Apologies Jake you were victim to this too. I'd hope an apology is in order, and like you say jake, a genuine honest second attempt at engaging with new content. If the toxic behavior continues, I think it's best that we just ignore the absurdity of the claim, perhaps ignore the poster if the behavior continues, and move on engaging with the more productive posts and the well-intentioned people on the forum.

Can this be done with duos? Would be interesting to see Steph and Draymond together at a starter disadvantage

Colbinii wrote:

lessthanjake wrote:

Colbinii wrote:How can you consider Kevin Love a star from 2016-2018 yet not consider Brook Lopez, Klay Thompson or Andre Iguodala stars?

I think that including Kevin Love and not those guys is amply justified by the fact that he’s a superior basketball player, and has been recognized as such. The only arguable one is Klay—who actually is a multi-time all-star, who made all-NBA, etc. I didn’t add Klay to the mix because in my experience from running lineup numbers for the Warriors, the team tends to actually get better if you take Klay out. So adding Klay seemed to me like it’d just lower an already-somewhat-small sample size to take out a guy who isn’t actually a positive. For instance, leaving aside the starter disadvantage stuff, from 2014-2024, in minutes with Steph on and Draymond and Durant off, the Warriors did better if Klay was off too than they did with Klay on.

And, indeed, since you brought it up, I just ran the numbers, and if you filter Steph’s starter disadvantage numbers down to just minutes without Durant, Draymond, AND Klay, then his 2014-2024 net rating at a starter disadvantage is +6.80 instead of +4.26. So taking out Klay actually catapults Steph up towards the top, rather than making his numbers look worse. But it also cuts the sample size almost in half. I prefer the Durant + Draymond numbers, because the sample size is bigger and because I personally think Klay has always just been a way-over-hyped role player anyways. But if you like Klay more than I do (and most people do!), you’re welcome to prefer that +6.80 number for Steph instead.

These numbers also greatly benefit the teams who are super deep 8-12 in the rotation, while hurting teams who have more higher-salaried players near the top since those teams will inevitably have less depth and more minimum-level players.

Maybe, though I suspect that a lot of these minutes are with role-player starters and top of the bench guys, rather than with the 8-12 guys. Remember, this isn’t just minutes with 4 bench guys. It’s any minutes at a starter disadvantage. A starter disadvantage can be something like your team having 3 or 4 starters on the court and the other team having all 5 starters on. So nothing here really requires these minutes to be with the 8-12 guys, and I imagine we’re largely talking about minutes taken up by other guys.

But yes, this generally goes to the second caveat I specifically mentioned in my OP—which is that “none of this analysis corrects for the precise quality of the role players or bench guys on each team.”

It also benefits teams who went deep into the luxury tax to retain lesser bench pieces [See Golden State] and hurt teams who avoided the Tax or 2nd Apron.

Yeah, that again goes to that same second caveat I mentioned in my OP. Not sure the Warriors are the best example of this, since they actually got a lot of lesser bench guys on veteran minimums, but I’m not an expert on NBA teams’ salary decisions with bench guys. The bigger thing with bench players is probably more about getting lucky with guys that actually work out—which is probably only loosely related to salary but does also go to my second caveat.

I guess I still don't understand why you wouldn't include other "stars".

Like, is 2014 Wade or Bosh really that much better than 2021 Lopez or 2016 Klay? Probably not.

It seems like you should just include everyone--stars included--and provide those numbers as well.

We have to draw the line somewhere. I think I drew a sensible line that most would intuitively agree with (especially now that I’ve provided numbers filtering out Klay too). And the more you filter out players, the smaller the sample gets. The sample sizes are already small enough that it makes me a bit uneasy, so filtering out more stuff would make the data worse IMO. That said, if you prefer further filtering, you’re welcome to run the numbers yourself on PBPstats and report out the results here!

parsnips33 wrote:Can this be done with duos? Would be interesting to see Steph and Draymond together at a starter disadvantage

Yes, it can be done with duos as well! Here is data for Steph and Draymond together at a starter disadvantage:

Steph Curry and Draymond Green at a Starter Disadvantage

2014: +40 in 498 possessions
2015: +110 in 764 possessions
2016: +192 in 820 possessions
2017: +113 in 697 possessions
2018: +10 in 375 possessions
2019: +131 in 475 possessions
2020: -46 in 48 possessions
2021: +155 in 637 possessions
2022: +54 in 326 possessions
2023: +25 in 531 possessions
2024: +29 in 640 possessions

Total 2014-2024: +813 in 5811 possessions (+13.99 per 100 possessions)

lessthanjake wrote:

parsnips33 wrote:Can this be done with duos? Would be interesting to see Steph and Draymond together at a starter disadvantage

Yes, it can be done with duos as well! Here is data for Steph and Draymond together at a starter disadvantage:

Steph Curry and Draymond Green at a Starter Disadvantage

2014: +40 in 498 possessions
2015: +110 in 764 possessions
2016: +192 in 820 possessions
2017: +113 in 697 possessions
2018: +10 in 375 possessions
2019: +131 in 475 possessions
2020: -46 in 48 possessions
2021: +155 in 637 possessions
2022: +54 in 326 possessions
2023: +25 in 531 possessions
2024: +29 in 640 possessions

Total 2014-2024: +813 in 5811 possessions (+13.99 per 100 possessions)

Looks about right. Man I still think that '21 team could have done something if a few breaks had gone their way

jalengreen wrote:Very cool data, haven't seen this sort of thing analyzed before so I appreciate you sharing.

This sort of thing is definitely tricky because of all the potential adjustments you can make; the Klay thing, as has been pointed out, and the fact that removing him makes Steph's numbers look far better (and decreases the sample size further to like, half a season's worth of possessions I think?). Lot of different ways to try to adjust things and get very different values. Of course it seems like no matter how you slice it, Manu's numbers are ridiculously impressive.

I calculated Steph's numbers with Klay off the court, but those have already been posted. Also LeBron 2009-13: net rating with a starter disadvantage and sans Wade/Bosh is +10.99.

Yeah, agreed. One can make lots of different filtering decisions and it can swing the numbers (as we saw with the Klay thing). It takes some time to run these, so the only numbers I ran for any of these guys are exactly what I’ve reported out. So I don’t really know what further filtering would do in specific cases, but I suspect it could change things (even if only just due to randomness). What I ran was just the filtering I thought made the most sense, without knowing beforehand how those filtering decisions would affect the data.

And yeah, I actually think that LeBron’s 2009-2013 numbers being really great here in a sense tends to validate that the method is measuring something meaningful, since it does show LeBron’s consensus best years as being the ones where he looks best in this analysis. I take that as an encouraging sign about the analysis. (EDIT: Unfortunately, this is only true to a lesser extent now that I corrected the Miami player-filtered data).

That said, I’d note again that lumping in 2009 and 2010 with the players-filtered-out version is a bit iffy, since filtering out stars inherently means more of these minutes are with bench or deep bench guys, and not filtering them out is very likely going to leave a lot larger a proportion of the minutes being with the other starters (but just in starter states of 4v5, 3v5, and 3v4). And, without filtering out anyone, that’d include often having the next best guys on the team on (which obviously isn’t the case with years where star teammates are filtered out). So it’s a bit comparing apples and oranges to compare filtered and non-filtered years. But, as I said in my OP, there’s not an obvious person to filter out in those years, so I don’t really have a better solution. My inclination is to think that any filtering in 2009 and 2010 probably wouldn’t change the bottom-line number for 2009-2013 much anyways, since it’d probably make the 2009 and 2010 numbers less good but also make those years a smaller part of the sample, leaving the super high 2012 and 2013 numbers as a larger part of the data. For instance, if we filtered out Mo Williams and Varejao, we get +24 in 316 possessions in 2009, and +23 in 233 possessions in 2010. Those unsurprisingly look less good than the non-filtered numbers for those years do, but the overall 2009-2013 time period ends up being +10.34 anyways, which is barely lower than what you listed above and is still outrageous.

I will note that Manu has timeframes that are pretty outrageous too. For instance, his 2012-2016 span with stars filtered is a +13.75 net rating.

_____________

Important Edit: It turns out I’d made a calculation error for LeBron’s player-filtered Miami years, and they’re actually not as good as I listed. I have changed the numbers in the OP and the post with the raw data, to account for this. But, it also makes this post exchange be based on a faulty premise. LeBron’s actual net rating at a starter disadvantage without Wade or Bosh from 2009-2013 is still +10.99 (the above poster who used that number must’ve actually ran the data themselves, rather than using my faulty data). But the 2009-2013 starter-disadvantage net rating if we used the William-and-Varejao-filtered numbers for 2009 and 2010 and filtered out Wade and Bosh for Miami would be just +6.43. So that is a big decrease, though it’s largely just because player-filtering in 2009 and 2010 makes the 2009-2013 sample not be swamped by the way larger size of the unfiltered 2009 and 2010 numbers.

Be interested to see Harden's numbers here, especially during his OKC days. 2011-2012 especially, those Harden bench lineups were pretty absurd. Curious to see if that translated to his time on the Rockets or Nets.

parsnips33 wrote:

lessthanjake wrote:

parsnips33 wrote:Can this be done with duos? Would be interesting to see Steph and Draymond together at a starter disadvantage

Yes, it can be done with duos as well! Here is data for Steph and Draymond together at a starter disadvantage:

Steph Curry and Draymond Green at a Starter Disadvantage

2014: +40 in 498 possessions
2015: +110 in 764 possessions
2016: +192 in 820 possessions
2017: +113 in 697 possessions
2018: +10 in 375 possessions
2019: +131 in 475 possessions
2020: -46 in 48 possessions
2021: +155 in 637 possessions
2022: +54 in 326 possessions
2023: +25 in 531 possessions
2024: +29 in 640 possessions

Total 2014-2024: +813 in 5811 possessions (+13.99 per 100 possessions)

Looks about right. Man I still think that '21 team could have done something if a few breaks had gone their way

Yeah, I was pretty surprised by that 2021 number when it popped up. I’m sure variance comes into play a good deal, but Steph and Draymond were very good that year!

Just for reference on this, here’s how the numbers look for Duncan and Ginobili together at a starter disadvantage (always treating Ginobili as a starter):

Manu Ginobili and Tim Duncan at a Starter Disadvantage (always treating Ginobili as a starter)

2003: +8 in 309 possessions
2004: +46 in 382 possessions
2005: +66 in 445 possessions
2006: +99 in 352 possessions
2007: +97 in 548 possessions
2008: +70 in 611 possessions
2009: +49 in 368 possessions
2010: +75 in 654 possessions
2011: +38 in 609 possessions
2012: +1 in 189 possessions
2013: +26 in 225 possessions
2014: +43 in 447 possessions
2015: +34 in 392 possessions
2016: +6 in 30 possessions

The total there for 2003-2016 comes out to +11.83 per 100 possessions. If we take the same length timeframe we did for Steph and Draymond, we get to +12.63 for Duncan and Ginobili at a starter disadvantage from 2004-2014. Steph and Draymond in that peak 2015-2019 five-year period were +17.76 together at a starter disadvantage. Meanwhile, Duncan and Ginobili were +16.39 together at a starter disadvantage from 2005-2009. It’s all not quite as high as Steph and Draymond, but it’s close!

And that’s another interesting thing about Ginobili to me. I’m getting outside the scope of my own thread, but I’ll note generally that, especially in the mid-2000s, Ginobili’s minutes with Duncan were the only lineup combination I’ve seen that competes with Steph + Draymond minutes over multi-year spans, in terms of how high the net rating is. So I see Ginobili as someone who can be an integral part of GOAT-level lineups with other great players *and* be incredibly good at elevating role-player lineups that are at a disadvantage.