RealGM

DraymondGold wrote:The Question of Curry's Superior Impact, and the importance of individual Offense > Defense

After LeBron (assuming he's rightly going in first), I think most people have a general group of players: 01 Shaq, 03 Duncan, 04 Garnett, 17 Curry, 23 Jokic. Assuming LeBron will go in first, I’d like to focus on the next five players. Of these five players, I think most people favor the big men. However, I’d argue Curry is still underrated after all these years.

Curry’s rather singular in how he gets his impact: there’s never been another player with his gravity, with his unique combination of on and off-ball shooting, scoring, and playmaking. It may be harder for people to see how he dominates the league compared to an outlier athletic phenom like Shaq, or players in the classic two-way big mold like Duncan and Garett. But like the Barkley saying “Jump shooting teams can’t win championships”, people may intuitively favoring a traditional mold, while underrating how revolutionary Curry was. The data on the whole suggests Curry really did have the more valuable full-season peak than Shaq, Duncan, Garnett, and Jokic, and I think that’s something people should take seriously: I don’t think it can be simply explained away just by being ‘ahead of his time’ or in a better situation or something like that.

A. Is Curry really the most valuable of these players according to the data?

Team Overall SRS
- 17 Curry’s Warriors 16.15 >> 01 Shaq’s Lakers 12.20 > 03 Duncan’s Spurs 9.01 > 23 Jokic’s Nuggets +8.50 > 04 Garnett’s Timberwolves +5 (decimal values weren’t given)
So while Curry obviously had better teammates, there is significantly more value to go around for the peak-Curry Warriors.

Peak WOWY (change in Margin of victory per game with/without a player in game, in 5-year stretch surrounding the peak year):
- 15-19 Curry +12.47 (52 missed games) > 21–25 Jokic +8.91 (36 missed games) > 01–05 Duncan +6.95 (30 missed games) > 99–03 Shaq +6.68 (42 missed games) > 2004–2008 Garnett +5.66 (23 missed games; no usable sample of missed games in 02–06)

Full Season Plus Minus (relative plus minus for playoffs):
- -17 Curry 19.29 > 23 Jokic 12.48 > 01 Shaq 9.73 > 03 Duncan 9.11 > 04 Garnett 7.81
Full Season Plus Minus (relative plus minus for playoffs, 7x playoff weighting):
- 17 Curry 21.2 > 01 Shaq 14.36 > 23 Jokic 11.15 > 03 Duncan 10.55 > 04 Garnett 4.92
Curry stays significantly ahead as we go to longer peak samples (~5 years).

Full Season On/off:
- 23 Jokic 22.3 > 17 Curry 21.1 > 04 Garnett 20.1 > 03 Duncan 17.0 > 01 Shaq 16.1
Curry sneaks ahead of Jokic and gains a much larger advantage over Garnett/Duncan/Shaq as we go to long peak samples (~5 year).

Full Season On/off (7x playoff weighting):
- 17 Curry 22.0 > 03 Duncan 21.7 > 04 Garnett 21.4 > 23 Jokic 17.7 > 01 Shaq 13.1
Curry gains a much larger advantage over Duncan/Garnett/Jokic as we go to long peak samples (~5 years). Shaq sneaks out in first in longer samples, but only if we include the year 2000 which is before the era.

So the raw impact metrics tend to suggest Curry was the most impactful, and often by a large margin. Now raw impact metrics are noisy with high uncertainty, so this is far from conclusive. What about adjusted metrics?

Full Season Augmented Plus Minus (a combination of the most accurate box stat on the market, and plus minus data)
- 23 Jokic 7.3 > 17 Curry 6.8 > 04 Garnett 6.6 > 03 Duncan 6.2 > 01 Shaq 5.9
Full Season Augmented Plus Minus (7x playoff weighting, averaged game-wise)
- 23 Jokic 7.4 > 17 Curry 7.1 > 03 Duncan 6.9 > 01 Shaq 6.2 > 04 Garnett 6.0

APM (regular season only)
- 17 Curry 8.5 > 03 Duncan 6.8 > 04 Garnett 6.4 > 23 Jokic 6.2 > 01 Shaq 5.6

There are different versions of RAPM, but to start with ‘Vanilla RAPM’, Curry again looks the best:
Vanilla RAPM (regular season only)
- 17 Curry 6.51 > 04 Garnett 4.70 > 23 Jokic 4.51 > 03 Duncan 3.79 > 01 Shaq 3.50
Curry remains ahead as you extend the RAPM to long peak samples (~5 year). Playoff RAPM is flawed, given the smaller samples and uneven lineups/matchups, but 15-19 Curry stays ahead of all but Duncan in 5-year playoff RAPM, and 13–17 Curry is ahead of Duncan too (note we lose playoff data in 23 from this source, so don’t have full Jokic stats).

EPM (regular season)
17 Curry 8.2 > 23 Jokic 7.9 > 04 Garnett 6.4 > 03 Duncan (no data before 2002)
EPM (playoffs)
- 17 Curry 8.9 > 03 Duncan 7.8 > 23 Jokic 7.2 > Garnett (no data before 2002)
EPM is commonly considered the most accurate descriptive stat on the market. I don’t have a subscription, but from posts from other people, available Shaq data in regular season. In the playoffs, 17 Curry is ranked 3rd on record over every Duncan, Garnett, Jokic, or Shaq playoffs on record.

So the adjust impact metrics are more mixed, but also frequently portray Curry at either best, and always in the top two.

On the whole, the data’s noisy, and one can always find stats they favor one player over another. But in a brief survey, looking across raw impact stats (WOWY and plus minus based), adjusted impact stats, and our best available hybrid stat, in regular season and full season samples across sample sizes, Curry tends to come out on top.

B. How is it that a one-way offensive player could be more impactful than two-way players?
In the recent DoctorMJ thread, we discuss the idea that individual offensive impact supersedes individual defensive impact in the play-by-play era. I suspect people are underrating how much more important individual offense is. Which is not to say that defense isn’t important (it’s equally as important at a team level) or that having great individual defenders aren’t a key foundation to building great team defenses… but still, individual offensive impact seems to significantly outweigh individual defensive impact.

What’s the evidence? Take the offensive and defensive splits in Goldstein RAPM, one of the classic sources for single-year RAPM (1997–2019)
-Offensive RAPM samples over +7: 4
Defensive RAPM samples over +7: 0
-Offensive RAPM samples over +6: 19
Defensive RAPM samples over +6: 4
-Offensive RAPM samples over +5: 50
Defensive RAPM samples over +5: 21

What about nbarapm’s 3-year and 5-year RAPM samples?
-Offensive RAPM samples over +8: 7 (peak Curry comes first)
Defensive RAPM samples over +8: 0
-Offensive RAPM samples over +7: 27
Defensive RAPM samples over +7: 2
-Offensive RAPM samples over +6: 55
Defensive RAPM samples over +6: 11

One can also see that individual offense trumps individual defense in WOWY data, and in team data (e.g. in the Top 20 teams all time in overall SRS, ~75% of those teams’ best players get more value on offense while 25% are get comparable or more value on defense).

How can this be? Volume and efficiency.

Volume: Even the best defenders have significantly less volume than the best offensive players.
In a Thinking basketball study of the 2011 season, the highest defensive usage tops out at 22.4%; the highest offensive usage is 34% during that time. Now usage is a flawed stat, and the study only examined one season (defensive usage may be higher in the dead-ball era), but the available studies suggest defenders have much less volume than offensive players. Which makes sense! Offenses are proactive — they can design the offenses around using the skills of their best player. Defenses are reactive — the opposing offenses can scheme to take the best defenders out of the action, putting them on the weak side or attacking weaker defenders.

Efficiency: The available data also suggests defenders aren’t more impactful on opponent efficiency than the best offensive players. The best offensive players lift their team’s offensive efficiency by +5 to +8 according to nbarapm, while the best defenders hurt opponent’s offensive efficiency by -4.5 to -6 (see the DoctorMJ thread).

So the best offensive players have significantly more volume, and a clear efficiency advantage

Given this asymmetry, Curry comes out as one of the clear offensive GOAT candidates, often with the best offensive RAPM on record, while also being a neutral to positive defender. This may help contextualize why so many of the overall impact metrics are higher on Curry than some of the two-way candidates, e.g. Duncan or KG. Shaq of course gets much of his impact from offense, as dose Jokic, so this could still be complementary to them. But the data we have seems to still favor Curry over those two.

Now It's not so compelling to end the debate -- far from it. No metric is perfect, and each can have wide uncertainty bars. And indeed, people may have contextual reasons to prefer other players (e.g. health/durability). But there do seem to be a trend in the data we have that Curry is the most impactful, and is being underrated in votes so far. Food for thought as people discuss!

DraymondGold wrote:

lessthanjake wrote:

Joao Saraiva wrote:The thing about Curry... well in 16 it didn't end well and his impact in the playoffs is not at the same level, making it hard to choose him over 01 Shaq, 09, 12, 16 or 18 Bron, 03 Duncan...

Then KD is added. And then the Warriors have it all - ultimate spacing with Klay, Curry and KD. Two elite defenders in Iguodala and Draymond. They were deep enough. And when Steph wasn't playing, the Warriors were still unstopable. When Curry had subpar performances in the finals it didn't matter, cause the fire power was still there.

So despite the team being incredible together it's hard to value Steph as much as other guys cause when they weren't there the team would go absolutely nowhere. In 17 the Warriors might even miss Steph for 50% or more of the time and still end up as champions. You can't say that about other guys.

I'm looking at Giannis, Kobe 06 whom I think is underrated peak wise and then Kevin Garnett and Steph, but still not decided the order I want them in. My guess is that Steph will go ahead on this list, but while he is a great ceilling raiser I don't see him putting up Kobe's impact in 06 for example.

Just some food for thought.

I don’t think the bolded is borne out by the data at all. Steph missed 53 games in the regular season + playoffs in the years Durant was on the Warriors. The Warriors were only 29-24 in those games, with a +0.28 net rating per 100 possessions.. Similarly, in the 6156 minutes when Steph was off the floor in those Durant years, the Warriors had a net rating of -0.50 per 100 possessions. The Warriors were actually demonstrably a pretty mediocre team without Steph. Which is not a bad place for a major star’s team to be without them, but they were definitely not “unstoppable” without him, nor do we have any indication that they could’ve won the title without him.

I think the argument people make otherwise is basically to say that they did fine without him *in the playoffs*. But that argument amounts to looking at 6 games in the 2018 playoffs, against a 47–win team and a 48-win team, with four of those games being at home. They did well in those games, but it doesn’t make a lot of sense to me to ignore way larger samples in favor of a playoff sample of a few games against early-round minnows anyways. Nor does beating 47-48 win teams actually give any indication whatsoever that a team could win a title.

I think it’s essentially undeniable that Steph was the guy that made the Warriors tick, and that they were really not an elite team without him. They were certainly a *talented* team without him, but the results really don’t bear out that they were actually all that good without him. And, of course, with him they were a top contender for GOAT team. That’s a huge feather in Steph’s cap. As I’ve noted, I’m not inclined to vote Steph higher than #4 here, but looking at what the Durant Warriors did with and without him is actually a data point that suggests he should be ranked higher than that IMO.

Just to add on to this:

17-19 Warriors net rating with all 4 all-stars: +15.9
17-19 Warriors net rating with 3 all stars, no Durant: +11.1
17-19 Warriors net rating with Dray+Curry, no Durant no Klay: +11.6
17-19 Warriors net rating with only Curry, no Durant no Klay no Dray: +9.4
17-19 Warriors net rating with Durant + Klay + Dray, no Curry: +3.7
17-19 Warriors net rating with none of the 4 all-stars: -10.3
Per pbpstats.com, full season data.

The data we have make it pretty clear — without Curry, this team was a far cry from the dominance that earns strong championship odds, and Curry is clearly the only player to have the effect on the Warriors.

People may argue that the team was built around Steph (although isn’t that the case for basically every candidate here?), or that his health is a concern, etc., but it’s really not clear at all how many championships they’d win if they had KD the whole time without Curry. Still a great team, sure. But multiple championships? That’s no so clear!