Sansterre's Top 100 Teams, #5. The 2018 Golden State Warriors

[sansterre]

Glossary:

Spoiler:

Overall SRS: My combo-SRS from the regular season and playoffs as discussed in the master thread
Standard Deviations: Standard Deviations of Overall SRS from the league mean.

When I post the roster makeup of the team, I try and do it by playoff minutes. The numbers are age, regular season BPM and Playoff BPM (basketball-reference's BPM is being used here).

So if I say: "C: Vlade Divac (22), +2.3 / +4.3" I mean that Vlade Divac was their center, he was 22, he had a BPM of +2.3 in the regular season and a +4.3 in the playoffs. Yes, BPM misses out on a lot of subtle stuff but I thought it a good quick-hits indicator of the skills of the players.

I also list the playoff players (20+ MPG) in order of OLoad (which is usage that integrates assists) and it has everyone's per game average for minutes, points, rebounds, assists and stocks (steals plus blocks), but all of those (including minutes) are adjusted for pace.

I then cover the three highest players in scoring per 100 (with their true shooting relative to league average) and the three highest players in Assists per 100. I realize that these are arbitrary, but I wanted a quick-hits reference for how these teams' offenses ran.

I then talk about Heliocentrism, Wingmen and Depth. Basically I add up all of the team's VORP (again, basketball-reference) and then figure out what percentage of that VORP comes from the #1 player (Heliocentrism), from the #2 and 3 players combined (Wingmen) and Depth (everyone else). I include the ranking among the top 100 for reference. There are only 82 of these rankings, because 18 teams pre-date BPM/VORP, so I only have 82 to work with. I'm not saying that these are particularly meaningful, I just thought they were cool.

Playoff Offensive Rating: Amount by which your playoff offensive rating exceeds the offensive rating you'd expect given the regular season defensive rating of your playoff opponents. If you would be expected to post a 99 given your opponents but you post a 104, that's graded as +5. This way we can compare across eras.
Playoff Defensive Rating is the same as Offensive Rating, just the opposite.
Playoff SRS: Is SRS measured *only* in the playoffs. Overall SRS is a mix of both playoffs and regular season.
Total SRS Increase Through Playoffs: Basically their Overall SRS minus their Regular Season SRS. This is basically how much better a team did in the playoffs than you'd guess, relative to their regular season performance.
Average Playoff Opponent Offense: The average regular season offensive rating of your playoff opponents.
Average Playoff Opponent Defense: The average regular season defensive rating of your playoff opponents.

Rankings of any kind are out of my list. So if I say that the '91 Lakers had the 42nd best regular season offense, I don't mean "42nd best of All-Time", I mean "42nd best of my Top 100 Teams of All_Time". Which will be pretty comparable, but I want to be clear about this.

I also walk through the playoffs at each round, covering their opponent their SRS (at that time), how many games the series was, the margin of victory (and a "+" is always in the favor of the discussed team; losing a series by +2.0 means that you outscored the other team by two points a game on average despite losing) and for reference I put in an SRS equivalency (beat a +5 SRS team by 5 points a game, that's an equivalent +10 SRS series).

In later entries I also add the Offensive and Defensive Ratings for each playoff series. This is just how well the team did, adjusted by the opponent's regular season average (if you play a team with an average Defensive Rating of 102, and you play them with an offensive rating of 106, you get credited with a +4). Pace for teams below 1973 or so is estimated based on regular season numbers, so it could easily be wrong by some.

In writeups, if I ever say a player shot at "-8%" or something, that means "his true shooting was 8% lower than the league average that year". Any time I say "a player shot" and follow it by a percent, I am *always* using true shooting percentage unless otherwise indicated.

I also have a modern comps section for any teams pre-2011. It's basically me weighting each statistical characteristic and feeding each player's stats into the BackPicks database and choosing the best-rated comp from the list. I might list something like this:

PG: 2017 LeBron James (worse rebounding, better passing, way fewer shots)

What I mean is, "This team's point guard was basically 2017 LeBron James, but make his passing better, make his rebounding worse and make him take way fewer shots).

Anyhow. I don't know how clear any of this will be, so please let me know what does and doesn't work from these writeups. And thanks for reading!

#5. The 2018 Golden State Warriors

Spoiler:

Overall SRS: +12.17, Standard Deviations: +2.69, Won NBA Finals (Preseason 1st)

PG: Stephen Curry, +7.7 / +7.1
SG: Klay Thompson, -0.4 / +0.1
SF: Andre Iguodala, -0.3 / +2.9
PF: Kevin Durant, +7.3 / +7.6
C: Draymond Green, +2.8 / +5.0

Regular Season Metrics:

Regular Season Record: 58-24, Regular Season SRS: +5.79 (74th), Earned the 1 Seed
Regular Season Offensive Rating: +5.0 (31st), Regular Season Defensive Rating: -1.0 (90th)
Shooting Advantage: +6.4%, Possession Advantage: -5.6 shooting possessions per game

Stephen Curry (PG, 29): 32 MPPG, 31% OLoad, 27 / 5 / 6 / 2 on +11.9%
Kevin Durant (PF, 29): 34 MPPG, 30% OLoad, 27 / 7 / 5 / 3 on +8.4%
Klay Thompson (SG, 27): 35 MPPG, 22% OLoad, 20 / 4 / 3 / 1 on +4.2%
Draymond Green (C, 27): 33 MPPG, 20% OLoad, 11 / 8 / 7 / 3 on +0.0%
Andre Iguodala (SF, 34): 25 MPPG, 13% OLoad, 6 / 4 / 3 / 1 on -2.0%

Scoring/100: Stephen Curry (39.8 / +11.9%), Kevin Durant (37.1 / +8.4%), Klay Thompson (28.1 / +4.2%)
Assists/100: Draymond Green (10.7), Stephen Curry (9.2), Kevin Durant (7.6)

Heliocentrism: 33.1% (43rd of 84 teams) - Durant
Wingmen: 41.0% (28th) - Curry & Draymond
Depth: 25.9% (43rd)

Playoff Metrics:

Playoff Offensive Rating: +6.48 (36th), Playoff Defensive Rating: -7.92 (18th)
Playoff SRS: +15.73 (6th), Total SRS Increase through Playoffs: +6.38 (3rd)
Shooting Advantage: +6.2%, Possession Advantage: -1.7 shooting possessions per game
Average Playoff Opponent Offense: +2.92 (28th), Average Playoff Opponent Defense: -1.18 (69th)

Kevin Durant (PF, 29): 39 MPPG, 30% OLoad, 30 / 8 / 5 / 2 on +5.0%
Stephen Curry (PG, 29): 38 MPPG, 30% OLoad, 26 / 6 / 6 / 3 on +3.4%
Klay Thompson (SG, 27): 39 MPPG, 20% OLoad, 20 / 4 / 2 / 1 on +2.3%
Draymond Green (C, 27): 40 MPPG, 18% OLoad, 11 / 11 / 8 / 4 on -2.7%
Andre Iguodala (SF, 34): 27 MPPG, 14% OLoad, 8 / 5 / 3 / 2 on +5.4%

Scoring/100: Kevin Durant (37.5 / +5.0%), Stephen Curry (34.3 / +3.4%), Klay Thompson (25.8 / +2.3%)
Assists/100: Draymond Green (10.4), Stephen Curry (7.3), Kevin Durant (6.1)

Playoff Heliocentrism: 32.2% (54th of 84 teams) - Durant
Playoff Wingmen: 43.5% (27th) - Green & Curry
Playoff Depth: 24.3% (46th)

Round 1: San Antonio Spurs (+2.9), won 4-1, by +8.8 points per game (+11.7 SRS eq)
Round 2: New Orleans Pelicans (+4.1), won 4-1, by +8.6 points per game (+12.7 SRS eq)
Round 3: Houston Rockets (+10.6), won 4-3, by +9.0 points per game (+19.6 SRS eq)
Round 4: Cleveland Cavaliers (+2.9), won 4-0, by +15.0 points per game (+17.9 SRS eq)

Offensive / Defensive Ratings from Opposition Regular Season Average:

San Antonio Spurs: +8.9 / -3.7
New Orleans Pelicans: +0.5 / -8.9
Houston Rockets: +8.4 / -9.8
Cleveland Cavaliers: +12.7 / -4.4

Shooting Advantage / Possession Advantage per game (unadjusted):

San Antonio Spurs: +6.0% / -2.1
New Orleans Pelicans: +3.0% / +2.3
Houston Rockets: +5.0% / -0.2
Cleveland Cavaliers: +12.7% / -8.8

Postseason Usage/Efficiency Change adjusted for Opposition:

Stephen Curry: -1.4% / -8.0%
Klay Thomposon: -1.9% / -1.4%
Andre Iguodala: +1.2% / +7.9%
Kevin Durant: +0.2% / -2.9%
Draymond Green: -2.1% / -2.2%


Do you know what really grinds my gears? When I get two teams really close to each other in the Top 10 who are in back-to-back years. All the stuff I want to talk about and break down for the 2018 really ought to be saved for the 2017 Warriors. At least the ‘91 Bulls and ‘96 Bulls have very different stories to tell (or at least, I have two different ideas for those articles). And the 2018 Warriors were basically the same, but worse in every meaningful way. So I’ve gone back and forth and basically decided that I’m simply going to deal with the ‘18 edition directly, and save my primary analysis for the ‘17 team. Which means this article is going to be unusually short compared to the last dozen or so. Apologies.

So. In 2014 the Warriors had a talented team, but their comparably low levels of ball-movement failed to set up their historically great shooters well. Even without assistance on that front they were a strong team, but dependent on Andrew Bogut being healthy (which was very much a hit or miss proposition). In 2015 they brought in Steve Kerr to coach them and suddenly their offense was unlocked. They proceeded to rip through the league and win a championship, overcoming the injury-riddled Cleveland Cavaliers. In 2016 they set the single-season record for wins with 73, but fell to LeBron and the Cavs in perhaps the best NBA Finals ever. In 2017 they replaced Harrison Barnes with Kevin Durant and the results were . . . considerable. That year they obliterated everyone and won the championship. They’d been favored at -128 to win a championship, considerably more likely to win than any Jordan time (highest odds in 1997, at +100).

How good were the 2017 Warriors? So good that the 2018 Warriors were favored at -187. That’s basically giving the 2018 Warriors a 2 in 3 chance of winning the Finals, way, way, way better odds than any team had ever received. It was like . . .

In the Return of the King there’s a fairly climactic set piece where the human city of Minas Tirith is besieged by the enormous army of Mordor, its numbers so vast that they blacken the plains. But the day is saved by the Riders of Rohan arriving, a massive cavalry formation that sweeps through the orcs and disperses them easily. It’s a cool moment. As their charge is spent, and they’re reforming, seeing the orcs run, they’re feeling pretty good about themselves.

And then the Mumakil arrive.

The Mumakil are basically giant battle elephants that have been kept in reserve by the bad guys. So the Riders of Rohan look up and see these massive super-Elephants slowly ride in and realize, “Oh balls, this is going to suck.” Or as Bernard Hill (the actor playing King Theoden) quipped in the audio commentary, “Right, one - nil.” The Riders of Rohan were great; the Mumakil were simply on a different level.

That’s pretty much how the entire NBA looked after 2017. When everybody heard that the 73-win Warriors had acquired Kevin Durant, they reasonably inferred that the ‘17 Warriors would probably win it all. After they had ripped through the NBA like Mike Tyson ripping through 37 straight heavyweights from ‘85 to ‘89 . . . Honestly, nobody expected the 2017 Warriors to be thaaaat good. Let’s be real, the 2017 Cavs were extremely good. They are certainly one of the best teams to lose an NBA Finals in the last 50 years. And the 2017 Warriors took them easily.

So yeah, going into 2018, everyone pretty much assumed that the Warriors would obliterate everyone again.

And in the regular season they’d be really good . . . but not great. They won 58 games (3rd in the league) and posted a +5.79 RSRS, again 3rd in the league. Both marks are good . . . but way, way, way short of what was expected. Was it that the Warriors were mailing in the regular season?

Not exactly.

One of the really weird sub-plots with the Warriors was that the team’s offense never seemed playoff-resilient. In both 2015 and 2016 their regular season offense had been incredible, but in both postseasons it had regressed considerably (their postseason leaps were always on defense). And a lot of that was tied to Stephen Curry’s numbers dropping in the postseason. It wasn’t just that; the Warriors’ motion offense (not unlike that of the 90s Utah Jazz) could be disrupted in the playoffs.

When Kevin Durant arrived in 2017 everything changed. The Warriors suddenly became considerably more ISO-capable, so even when the flow of the offense was compromised they could still generate points efficiently. In 2017 their playoff offense went from outstanding in the regular season to one of the best ever in the postseason. Durant was the key to that change.

So within that framework, it shouldn’t be surprising that Durant got a lot of credit for how good the 2017 Warriors were - his addition was the obvious catalyst for the team’s transformation (much like how Jeff Hornacek was the catalyst for the Jazz going from also-rans to the best team in the West; he didn’t get much credit for it but I digress).

But all of the credit Durant was getting was within the framework of an offense that Curry powered. In 2018 Curry missed 31 games. Here’s how that WOWY breakdown looks:

Curry out: 17-14, -0.16 MoV
Curry in: 41-10, +9.71 MoV

Without Curry, the team was league average. With Curry, the team was outstanding. Don’t read too much into this; in 2016 Curry missed much of the series against the Blazers (this is pre-Durant) and the Warriors still won reasonably easily. So even without Curry the playoff Warriors could win a series against a good team. But these WOWY splits emphasize just how dependent the team was on Curry. Every single regressed +/- stat lists Curry as considerably more valuable than Durant for the team, at least in the regular season.

So the Warriors finished the regular season as the 2nd best team in the West. They finished behind the formidable Rockets, who combined James Harden and Chris Paul to formidable effect, with 65 wins and +8.21 RSRS to their name. It was almost certain that the Conference Finals would be between these two teams.

In the first round they’d face the +2.9 San Antonio Spurs, but they would do it without Stephen Curry, who was still recovering from injury. It scarcely mattered. The Warriors started JaVale McGee at center and won by 8.8 points per game in five. The Spurs were held to -5.3% shooting, while the Warriors shot well as a group (+1.9%). Durant averaged a 28/9/5 on +5.0% and Klay Thompson a 23/3/3 on +7.7%. An 8.8 point per game margin of victory over a decent team. That it was without Curry was fairly impressive.

In the Semis they faced Anthony Davis and the +4.1 New Orleans Pelicans. Curry would sit a game, come off the bench for a game and play for the rest. The Pelicans proved a challenging matchup; they slowed the Warriors’ offense considerably. As a team the Warriors shot only -0.2%; Curry came back at 25/5/4 on +9.4%, Durant averaged a 28/7/5 on +2.9% but Klay struggled with a 20/6/2 on -7.9%. The Warriors’ defense, however, played superbly. Anthony Davis averaged an impressive 28/15/2 (2.2 steals, 2 blocks) on -2.3% but the Pelicans overall were held to -3.2% shooting. In the end the Warriors prevailed in five by 8.6 points a game. Again, not a blow-your-mind win, but an 8.6 point per game win over a good team is still very good.

And in the Conference Finals, of course, were the +10.6 Houston Rockets, by any reasonable standard the best team in the league. Except, of course, that the Warriors had been missing Curry for much of the year. Everyone wondered. Could the Rockets really pull this off? They would have home court advantage after all.

In Game 1, James Harden came out shooting, posting a 41/4/7 on +1.7%. But the Warriors blew by them easily, shooting +9.8% as a team and winning by 13. Durant had a vintage 2012 performance, with a 37/3/1 on +6.8%, and Curry had a 28/4/2 on +15.3%.

Game 2 was the complete opposite. The Warriors struggled to shoot (-0.2% as a team) with every Warrior starter but Durant posting a negative BPM. While the Warriors were able to slow Harden (27/10/3 on -4.9%), the Rockets’ role-players played excellently and the team combined to shoot +8.0%. The game ended as a slaughter, with the Rockets winning by 22.

So, two games in the teams were tied, but the Rockets overall had looked better. But now the series was going to Golden State.

Game 3 was a brutal horse-whipping. The Rockets were held to -6.6% shooting and the Warriors shot at +7.5%. Curry had a 35/6/1 on +15.1%, Durant a 25/6/6 on +4.6% and Iguodala a 10/3/3 on +17.1% with 3 steals. The Warriors led by 21 going into the 4th, and in garbage time the Warriors ran up another 20 points.

Game 4 would be neck and neck, and Iguodala would have to sit out. The Warriors went into the fourth quarter with a ten point lead. And in the fourth quarter the Rockets’ defense would completely clamp down, holding the Warriors to 3 made shots in the entire quarter. Durant and Curry combined to shoot 2 of 13 from the field that quarter, while the Rockets scored 25 points, enough to come from behind and win by 3.

At this point the series was 2-2, but the Warriors had won two blowouts while the Rockets had only won one. But the series had three games potentially left, and two would be in Houston.

Game 5 would be the same as Game 4, with Iguodala sitting out and the Warriors holding a lead (though only by one point this time) going into the fourth. The Rockets closed it out, managing to eek out a four point win. The Warriors outshot the Rockets (+1.6% to -2.4%) but lost the possession battle by 10 shooting possessions. Curry had a strong game with a 22/6/4 on +1.7% with 4 steals, but Durant struggled with a 29/4/0 on -0.7%. So the Warriors were down 2-3, but Chris Paul had been injured going into the final minute. So the Warriors would need to win out, but Houston would be missing one of their best players.

The Warriors (still missing Iguodala) ripped through the Rockets in Game 6. Durant struggled with a 23/7/4 on -5.9%, but Curry posted a 29/5/6 on +7.4% with 3 blocks, Draymond had a 4/10/9 on +11.1% with 4 steals and 5 blocks (what a freaking statline) and Klay posted a 35/6/2 on +20.5% with 4 steals. James Harden turned the ball over 9 times and the Warriors won by 29, 22 of them in the final quarter. So it would all come down to Game 7.

The Warriors took Game 7 by 9 points. Curry had a 27/9/10 on +5.8% with 4 steals and Durant posted a 34/5/5 on +11.3% with 3 blocks. The Rockets’ offense struggled, combining to shoot -9.5%. The Warriors prevailed, winning the series and advancing to the Finals.

A word about that series.

History seems to remember this series in the following way: “The Warriors went down 2-3, and then got lucky when Chris Paul got injured and won out as a direct result.” Given that narrative, it would be pretty hard to put the ‘18 Warriors this high, right? Granted, the ‘18 Rockets were really, really good (one of the better regular season teams ever), but you don’t ever want an ATG team to only win a series out of sheer luck.

But I don’t think they did. First off, don’t forget that while the Rockets may have lost Chris Paul for two games, the Warriors lost Iguodala for four. I’m not saying the two cancel out, but let’s not pretend that the Rockets won games 4 and 5 against a full-strength Warriors team. Furthermore, the Warriors won the series by a fairly nuts 9 points a game. How is that possible, when we all know the series was super-tight? The Rockets had one big win, of 22 points, but their other two were tight, by only 3 and 4 points (both games Iguodala missed). The Warriors’ wins were by 13, 41, 29 and 9. Did the Warriors run up the score in the fourth quarter of their two big blowouts? Yes. But in Game 3, when the Rockets pulled Harden the Warriors were already up by 31. In Game 6 the Warriors were up by 25 when the Rockets pulled Harden. Even if we only count the score from when the Rockets threw in the towel, the Warriors still had an average MoV of 7 points a game. Let’s consider the series breakdown with the injuries and adjusted for blowouts (the Rockets were up by 27 in Game 2 when the Warriors’ starters were pulled):

All Healthy: Warriors 2-1, +5.7
No Iguodala: Warriors 0-2, -3.5
No Iguodala, No Paul: Warriors 2-0, +17.0

When everybody was healthy, the ‘18 Warriors were beating a +10 OSRS team by 5.7 points per game, which is pretty nuts (and with 2 of those 3 games in Houston). From a pythag point of view, the Warriors seem pretty clearly to have been the better team. And by a fair margin. Let’s adjust the series with Iguodala and Paul’s unadjusted postseason +/- data (Iguodala at +6.0 per 100 (+3.1 per game given mpg and pace), Paul at +11.4 per 100 (+7.7 per game):

Game 1: Warriors by 13, 1-0
Game 2: Rockets by 27, 1-1
Game 3: Warriors by 31, 2-1
Game 4: Warriors by 0.1, 3-1 (adding 3.1 for Iggy)
Game 5: Rockets by 0.9, 3-2 (adding 3.1 for Iggy)
Game 6: Warriors by 20.4, 4-2 (adding 3.1 for Iggy, subtracting 7.7 for Paul)
Game 7: Warriors by 4.4, 5-2 (adding 3.1 for Iggy, subtracting 7.7 for Paul)

Obviously this is a fairly ridiculous exercise. But it’s very reasonable to think that with a healthy Iguodala Games 4 and 5 become coin flips, and it’s hard to imagine that Paul is more than 9 points more valuable than Iguodala for Game 7. It’s totally true that the series went to seven, and the Warriors had to win two games that Paul seemed to miss. I’m just saying that the Warriors were hurt by injuries almost as much and most objective indicators suggest that the Warriors were the better team that series by a pretty respectable amount. And that victory over the ‘18 Rockets is actually one of the better wins against really good teams of anyone on this list.

Unrelated, check out this weird list of Durant games from the series (points, assists and MoV)

Game 1: 37 points, 1 assist (+13)
Game 2: 38 points, 0 assists (-27)
Game 3: 25 points, 6 assists (+31)
Game 4: 27 points, 3 assists (-3)
Game 5: 29 points, 0 assists (-4)
Game 6: 23 points, 4 assists (+25)
Game 7: 34 points, 5 assists (+9)

Games where Durant scored more than 30: 2-1 with a +0.0 MoV
Games where Durant scored more than 25: 2-3 with a -2.4 MoV
Games where Durant actually recorded an assist: 4-1 with a +15 MoV

Obviously these cutoffs are arbitrary and from a small sample size. But there’s a pretty conspicuous pattern where Durant would seem to forget about passing and the Warriors would struggle, whether he was scoring well or not. And in games where worked as more of a playmaker, the Warriors were dominant whether Durant was scoring well or not. Interesting.

In the Finals were LeBron and the Cavs. You know how the ‘17 Cavs were one of the better teams to lose an NBA Finals ever? The ‘18 Cavs were emphatically *not* on that level. Here was their season leading up to the Finals:

Regular Season: +0.59 SRS
Round 1: Indiana Pacers (+1.2), won 4-3 by -5.7 points per game (-4.5 SRS eq)
Round 2: Toronto Raptors (+5.7), won 4-0 by +14.0 points per game (+19.7 SRS eq)
Round 3: Boston Celtics (+3.6), won 4-3 by +0.9 points per game (+4.5 SRS eq)

The series against Toronto? Hot fire. Everything else? Hot . . . something else. They barely made it past the Pacers (and were outscored badly in the process) and barely made it past the Celtics. Sure the LeBron Cavs didn’t take the regular season seriously, but a +0.59 RSRS is a joke. Even all-playoff no-regular-season teams like the Hakeem Rockets and ‘01 Lakers had better SRSs than that. The ‘18 Cavs were flat-out one of the worst teams to be in the NBA Finals in a while.

LeBron still played amazingly for the series. He averaged a 34/9/10 on +6.4% shooting. And the Cavs got swept by 15 points a game. It wasn’t close. At all. The Warriors never won by less than 8 points. I’m not even going to dwell on the stats. It was a straight up murder. On one hand, a 15 point-per-game win in the Finals is super impressive. On the other hand, the ‘18 Cavs were really not legitimate opposition.

12 | Warriors
11 |
10 |
9 |
8 |
7 | Rockets
6 |
5 |
4 | Sixers, Jazz
3 | Celtics, Pacers
2 | Cavs, Wizards, Thunder, Pelicans
1 | Raptors, Spurs, TWolves, Nuggets
0 | Bucks, Blazers, Hornets, Clippers
-0 | Pistons
-1 | Heat, Lakers
-2 | Mavs
-3 | Knicks, Nets
-4 | Magic
-5 | Hawks, Grizzlies
-6 | Bulls, Kings
-7 |
-8 | Suns
-9 |
-10|

So, where should the 2018 Warriors place all-time? Well, they played in an extremely competitive year; very few tanking/awful teams compared to some. There are so many wacky mechanics here.

Personally, I’d drop them several places. They get a lot of credit for their blowouts of the Rockets (and if I had a cap, as I expect to for v2) it would adjust the value of that win down considerably. And their regular season was fairly weak. If I ran v2 and it came back with the ‘18 Warriors at . . . I don’t know, #8, or #11 or something, I’d be completely fine with it.

And don’t get me wrong. Healthy, this team probably was Top Five level. They just weren’t for much of the year. Their regular season was torched by losing Curry for 30+ games, and their playoffs were compromised by losing Curry for a series and Iguodala for six games. I totally get why version 1 likes the ‘18 Warriors so much. But I think that a more realistic rating for them is a few places down.

Sorry for the short article; super excited for the #4 article though!

Back to the Main Thread

[Colbinii]

Two Curry-era Warrior teams in the top 5.

[Vladimir777]

I agree with your idea to save the main analysis for the 2017 Warriors post. I remember being so frustrated with the league during these years, because, as you said, it was such a given that the Durant Warriors would win that other teams might as well not have played the seasons.

I love that you're not only quoting Lord of the Rings, but the glorious Lord of the Rings DVDs (assuming it's the Extended Edition, by the way), which were by far my favorite DVDs of my childhoods. Good God, the content on those suckers!

I remember that wonderful Western Conference Finals. Everyone knew that the victor would go on to be the NBA champion. 2018 was also notable since LeBron straight up BALLED OUT during the postseason. I also remember the Rockets missing something like 27 straight 3-pointers in that game 7. As someone who routed against the Warriors, it was a very frustrating game to watch live.

[Odinn21]

I'm not sure if I raised my concerns about this before but I think your formula need an agent for number of games played within a series.

Some examples;
The Spurs beat the Suns in 5 games in 2005 playoffs, that was +14.1 SRS eq for the Spurs.
The Warriors beat the Rockets in 7 games in 2018 playoffs, that is +19.6 SRS eq for the Warriors.

My point is not about the teams. Why that +19.6 is more valuable than that +14.1 even though +14.1 had the more decisive victory.
Playoffs basketball is not regular season basketball, MOV/SRS based approaches are wonky. What if the Warriors won the series in 6 but with +6.0 MOV instead of +9.0, is that worse really?

Also even though I haven't thought about a solution, it made me think about 2018 Rockets being #94 on the list. They went toe to toe with a top 5 ever team after a league's best 8.21 SRS, 65W season and they are only 94th?

[sansterre]

I'm not sure if I raised my concerns about this before but I think your formula need an agent for number of games played within a series.

Some examples;
The Spurs beat the Suns in 5 games in 2005 playoffs, that was +14.1 SRS eq for the Spurs.
The Warriors beat the Rockets in 7 games in 2018 playoffs, that is +19.6 SRS eq for the Warriors.

My point is not about the teams. Why that +19.6 is more valuable than that +14.1 even though +14.1 had the more decisive victory.
Playoffs basketball is not regular season basketball, MOV/SRS based approaches are wonky. What if the Warriors won the series in 6 but +6.0 MOV instead of +9.0, is that worse really?

Also even though I haven't thought about a solution, it made me think about 2018 Rockets being #94 on the list. They went toe to toe with a top 5 ever team after a league's best 8.21 SRS, 65W season and they are only 94th?

All great points.

1) I agree in principle with the number of games needing to be included. That said, v2 is going to be mostly regression-driven, so I expect games to weighted as much as is predictive. So if winning in 5 is more correlated with future success than winning in 7, it will be counted.

2) The problem with the Rockets is a few things. First, their regular season was awesome but my formula is weighted heavily toward the postseason. The second is that the current weighting of results (MoV but no games played count and no blowout cap) really gouges them for that Western Conference Finals. And lastly, right now my SRS is ELO-esque, which is to say, the ratings go up through time. And between the '18 Warriors regular season being only pretty good *and* their first round with the Spurs being without Curry, their OSRS going into the series was far lower than it really should have been. So v1 of the formula looks at Houston losing by 9 points a game to a +8.7 OSRS team and penalizes them *hard*. I expect the '18 Rockets to move up a lot in v2; in this list they're way too low. But they're always going to be screwed by the fact that no formula will objectively rate the '18 Warriors as an ATG team based on their resume before that series. So to some extent, any purely objective formula will probably underrate the '18 Rockets, specifically because it will underestimate the '18 Warriors going into their series. Unless I figure out how to run SRS repeatedly until everything cancels out or something.

[Odinn21]

1) I agree in principle with the number of games needing to be included. That said, v2 is going to be mostly regression-driven, so I expect games to weighted as much as is predictive. So if winning in 5 is more correlated with future success than winning in 7, it will be counted.

Regression methods might find the optimal point between number of games and MOV. Which version do you have in mind?

2) The problem with the Rockets is a few things. First, their regular season was awesome but my formula is weighted heavily toward the postseason. The second is that the current weighting of results (MoV but no games played count and no blowout cap) really gouges them for that Western Conference Finals. And lastly, right now my SRS is ELO-esque, which is to say, the ratings go up through time. And between the '18 Warriors regular season being only pretty good *and* their first round with the Spurs being without Curry, their OSRS going into the series was far lower than it really should have been. So v1 of the formula looks at Houston losing by 9 points a game to a +8.7 OSRS team and penalizes them *hard*. I expect the '18 Rockets to move up a lot in v2; in this list they're way too low. But they're always going to be screwed by the fact that no formula will objectively rate the '18 Warriors as an ATG team based on their resume before that series. So to some extent, any purely objective formula will probably underrate the '18 Rockets, specifically because it will underestimate the '18 Warriors going into their series. Unless I figure out how to run SRS repeatedly until everything cancels out or something.

- Your version of ELO-esque SRS only considers possibility of winning? Like in chess? Or do you have some other things to factor in such as # of total games, or # of close games, etc?

But they're always going to be screwed by the fact that no formula will objectively rate the '18 Warriors as an ATG team based on their resume before that series.

This is the tricky part, yeah. This is mostly on injuries. Curry played in only 41.3% of total minutes in that season (the usual range for a healthy superstar starts around 65% and it's mostly around 75~%).

[sansterre]

1) I agree in principle with the number of games needing to be included. That said, v2 is going to be mostly regression-driven, so I expect games to weighted as much as is predictive. So if winning in 5 is more correlated with future success than winning in 7, it will be counted.

Regression methods might find the optimal point between number of games and MOV. Which version do you have in mind?

2) The problem with the Rockets is a few things. First, their regular season was awesome but my formula is weighted heavily toward the postseason. The second is that the current weighting of results (MoV but no games played count and no blowout cap) really gouges them for that Western Conference Finals. And lastly, right now my SRS is ELO-esque, which is to say, the ratings go up through time. And between the '18 Warriors regular season being only pretty good *and* their first round with the Spurs being without Curry, their OSRS going into the series was far lower than it really should have been. So v1 of the formula looks at Houston losing by 9 points a game to a +8.7 OSRS team and penalizes them *hard*. I expect the '18 Rockets to move up a lot in v2; in this list they're way too low. But they're always going to be screwed by the fact that no formula will objectively rate the '18 Warriors as an ATG team based on their resume before that series. So to some extent, any purely objective formula will probably underrate the '18 Rockets, specifically because it will underestimate the '18 Warriors going into their series. Unless I figure out how to run SRS repeatedly until everything cancels out or something.

- Your version of ELO-esque SRS only considers possibility of winning? Like in chess? Or do you have some other things to factor in such as # of total games, or # of close games, etc?

But they're always going to be screwed by the fact that no formula will objectively rate the '18 Warriors as an ATG team based on their resume before that series.

This is the tricky part, yeah. This is mostly on injuries. Curry played in only 41.3% of total minutes in that season (the usual range for a healthy superstar starts around 65% and it's mostly around 75~%).

1) Honestly, I probably shouldn't have used a term of art. Really, I was just going to build a giant sheet of every second round matchup and have, as components, all the data available before that round (RS Wins, RS SRS, prior series MoV, prior series MoV (blowout adjusted), prior series w-l), compare those components to the results of that second round series and see what is the most predictive just using basic CORREL, SLOPE and INTERCEPT functions. I'd love to say that I'm using a sophisticated multivariate regression tool, but I'll just be fumbling through by manually trying a bunch of combinations that yield the combo of best fit.
2) By ELO I mean less that it estimates the chance to win prior to the match, and more that it's time-linear. If Anand loses to Caruana, his ELO is docked by an amount based on the data available at that time. If Caruana goes on to demolish Carlsen, Anand doesn't get a retroactive bonus because new data suggests that Caruana is better than we thought. I'm saying that my current formula doesn't retroactively adjust SoS based on subsequent series.
3) Yeah. I don't think there's any way around injuries.

[Odinn21]

1) Honestly, I probably shouldn't have used a term of art. Really, I was just going to build a giant sheet of every second round matchup and have, as components, all the data available before that round (RS Wins, RS SRS, prior series MoV, prior series MoV (blowout adjusted), prior series w-l), compare those components to the results of that second round series and see what is the most predictive just using basic CORREL, SLOPE and INTERCEPT functions. I'd love to say that I'm using a sophisticated multivariate regression tool, but I'll just be fumbling through by manually trying a bunch of combinations that yield the combo of best fit.

I thought you were running a specific regression like Ridge, Bayesian, Generalized Linear or Huber.

With the commands you refer to, I'd assume that you're using MS Excel and the easiest ways to determine best fits without going that deep are;
- You can create histograms or distribution charts and overlap them with sample size to see the better fit.
- Or you can use coefficients that can be found in (Chart Tools > Trendline), then use those coefficients to check residual squares, then it's just Data Analysis under Data tab. The program will give you regression statistics as R^2, adj. R^2 and standard error, also ANOVA test values and some more. This is Ordinary Least Squares regression.
Basically, you'll be seeing the values which your commands use and check their reliability directly in numbers.

2) By ELO I mean less that it estimates the chance to win prior to the match, and more that it's time-linear. If Anand loses to Caruana, his ELO is docked by an amount based on the data available at that time. If Caruana goes on to demolish Carlsen, Anand doesn't get a retroactive bonus because new data suggests that Caruana is better than we thought. I'm saying that my current formula doesn't retroactively adjust SoS based on subsequent series.

Maybe you can use 538's ELO numbers?
https://projects.fivethirtyeight.com/complete-history-of-the-nba/#hawks
Though it would require recording that db basically.

3) Yeah. I don't think there's any way around injuries.

Yeah, sadly...

[Warriors Analyst]

Some thoughts I have on this team. I think this might actually be the most well balanced roster the Warriors had in their KD years. Was it the best team? Probably not. OP makes a convincing argument with numbers to back it that KD's desire to freelance hurt the Warriors against Houston AND that the series would have been over in 5-6 games with both rosters fully healthy. The obvious argument that the 2016-17 Warriors have over the 2017-18 roster is better on-court chemistry. That team probably beats Houston at full health too, but without Andre, I don't think that the the 2016-17 Warriors would have been able to pull it off against Houston.

The contributions of Kevon Looney and Jordan Bell go deeply under appreciated by outside observers. Houston made a gamble that they would be able to attack Steph, Looney, and Bell on switches over and over again and that they'd break. They didn't. Bell wasn't great, but Looney in particular, was a godsend by virtue of just holding up and providing about neutral value on a huge volume of switches. Looney averaged 19.3 minutes a game in the WCF. He defended 10.4 FGA a game in those 19.3 minutes and only gave up a difference of 0.3 DFG vs expected FG% and Bell defended 8.5 FGA a game in 14.2 minutes for a +3.6 FG% . So pretty much every other minute, Looney and Bell were getting dragged into plays and asked to defend on an island against CP3 or Harden. Houston's bet didn't really pay off and the Warriors were able to hold on defensively by relying on a third year C who'd struggled with health issues since he came into the league and an undersized C they found in the second round.

Maybe those numbers don't really seem all that impressive, after all, Looney was basically neutral and Bell slightly negative on differential DFG/FG, but consider the alternatives once Andre Iguodala went down. The Warriors got 42 minutes combined from Javale (3 minutes in one game), Zaza (12 minutes in 3 games), and West (27 minutes in 4 games) for that entire series. McCaw, who had been useful in the previous year's playoffs, was recovering from a terrifying concussion and played 5 minutes in two games. Kerr effectively the put the kibosh on all of his vet C's and did his best to cobble together playable lineups between Dray at C with wings or Looney/Bell at the C.

Only five lineups played more than 10 minutes in that entire series. Curry/Thompson/Durant/Green/Looney played the most, 74 minutes and had a +20.5 net rating... despite only having a 101.3 ORTG, so that lineup absolutely crushed it on defense with an 80.8 DRTG. Death Lineup 2.0 played 56 minutes and only had a +2.5 net rating and a 114 DRTG. The starters with Bell had a +18.2 net rating in 39 minutes, but their defense was merely good at 103.8 DRTG, compared to the godly DRTG of the stares with Looney. An odd lineup that worked really well: Steph/Klay/Young/Livingston/Dray struggled to score and had a 102.4 ORTG, but... a 68.3 DRTG in 19 minutes. The Death Lineup with Livingston in place of Andre was fine with a +7.8 net rating, but was pretty eh on defense and gave up a 111.1 DRTG. It's hard to draw much from lineups that played less than 10 minutes but more than 5 minutes, but the Looney or Bell at C lineups all held up really well on defense, whereas the West or vet C lineups got torched on D.

So bringing it back to the question of the 2016-17 Warriors vs the 2017-18 Warriors: yes, a healthy 2016-17 team probably wins against Houston, assuming a happy and well-integrated KD, but I do believe that the vet C's still would have gotten destroyed and then kiboshed leaving us with... a rookie Damion Jones, a second year Looney who isn't ready for rotation minutes, and James Michael McAdoo (11 minutes in 4 games) as your remaining big men. Matt Barnes helps for sure as a weird mix of Andre and Dray's skillsets, but he was hurt for most of the playoffs, so we never really got to see how he held up in as a Warrior in playoff minutes. McCaw was solid for us, but if you're going to play McCaw lots of minutes, you need Barnes, Andre, and Livingston to play lots of minutes at the PF next to Draymond. Ian Clark was useful and had good chemistry with David West on offense, but he was a minus defender and Nick Young proved a totally passable defender when he got ISO'd against Houston.

So I do think that the 2016-17 Warriors vs 2017-18 Houston would be closer than a lot of people think. I suspect the Warrior still win by virtue of KD being happy and compliant within the offensive system, more on that later, but if Andre goes down, the Warriors absolutely lose, even if CP3 goes down. There's just no way you can cobble together 48 minutes of a passable frontcourt between Dray/Matt Barnes/Livingston/McAdoo against that version of the Rockets.

But back to the KD thing. In his All the Smoke interview, Draymond talked at length about the KD situation. One of the most illuminating things he said was that he actually thought there was a considerable chance KD would leave at the end of the 2018 season. Draymond made it sound like he had to convince KD to just suck up one more season for the chance at a threepeat and he laid out a few things that painted a picture of a disgruntled KD in 2017-18. Draymond said that KD was very hurt by not getting crowned the best player in the world after his Finals MVP performance against the Cavs. Dray pretty much insinuated that the rejection KD felt leaked onto his oncourt behavior and KD got paranoid and defensive. One prime example Dray made was that KD would complain to him that Steph and Klay were intentionally looking him off to get their own shots, which is not a conclusion I think most people would make unless they got into their head that they weren't getting the recognition they received and were struggling with their own ego. So Dray says that he and Iguodala would go out of their way to call plays for KD and to push Kerr to do so in order to keep KD happy.

So what happens in the WCF? KD falls right into Houston's trap when they switch and gets ISO happy. As OP pointed out, when KD passed, the Warriors all but one game and they had a huge margin of victory. When KD decided instead to just ISO to try and prove... I don't know... that the Warriors NEEDED him to play hero ball to win... Houston won games. I do suspect that a happier KD in 2016-17 who isn't on a revenge tour against his detractors makes more sound decisions against Houston and that probably helps us, but yeah, KD was a weird variable for us against Houston.

Part of that I think is on Kerr. Something I noticed over the years is that it seemed like a lot of the ISO's KD got under Kerr occurred in the between the elbow and the baseline in the mid post. There is no disputing that KD is an incredible ISO scorer, but I think that there's a somewhat limited set of outcomes from a midpost ISO, whereas a top of the key ISO probably has a more broad set of outcomes and makes the defense work harder. When you drive from the mid post I just think it's harder to truly collapse the defense and bring multiple defenders to you, which in turn limits your passing options. You have a more limited view of the court and it's not as easy to get deep in the paint and kick the ball out to ANY place on the court, whereas from the top of the key, you have a more clear path to kicking it to EITHER corner/wing at any point in the drive and if you get deep enough, you might be able to whip a pass back to the top of they key if you have that type of vision.

I would never yearn for Mike Brown over Kerr, but one thing I appreciated about Brown's brief stint as the Warriors coach in the 2016-17 playoffs when Kerr missed games was that Brown had 0 qualms about just... letting KD ISO or PNR at the top of the key. It was as simple as dragging Rudy Gobert or one of the hapless Portland defenders to the top of the key and then KD would shoot over them or drive by them, at which point KD could dunk it, kick it out to shooters, etc. On a similar note, I felt like the 2019 WCSF against Houston was effectively a coin flip until KD went down, at which point I felt pretty confident that the Warriors would win the series. It's not like the Warriors were more talented without KD, but I expected that the offense wouldn't stagnate into midpost ISO's as often and that in crunch time, the Warriors could run the Steph/Dray PNR without concerning themselves with KD's feelings. Lo and behold in Game 6, that's pretty much all they did in the fourth quarter and in a nice twist, the ball moved around! One of my principal complaints with Kerr is his ideological opposition to PNR, which seems to be driven by a misguided belief that the PNR is predictable and doesn't stimulate ball movement. I really don't think that's true at all, especially with the Steph/Dray PNR, which would end so often in Dray lobbing it to the dunker on the baseline or kicking it out to the wing for Klay or Andre to shoot or pass it and force defenses to rotate in a panic for fear of the ball coming back to Steph, but I digress...

I do think it's pretty clear that the 2016-17 Warriors were a happier, and probably better team as a result than the 2017-18 Warriors. It's remarkable how much can get weird when one of your key players starts to break the offense to satisfy their own urges. But I also think there's a compelling argument that the 2017-18 Warriors have a better and more balanced roster than the 2016-17 team. I stand by my prediction that the 17-18 team was better equipped for Houston and that the 16-17 team would absolutely not be able to withstand the Andre injury.

[parsnips33]

Some thoughts I have on this team. I think this might actually be the most well balanced roster the Warriors had in their KD years. Was it the best team? Probably not. OP makes a convincing argument with numbers to back it that KD's desire to freelance hurt the Warriors against Houston AND that the series would have been over in 5-6 games with both rosters fully healthy. The obvious argument that the 2016-17 Warriors have over the 2017-18 roster is better on-court chemistry. That team probably beats Houston at full health too, but without Andre, I don't think that the the 2016-17 Warriors would have been able to pull it off against Houston.

The contributions of Kevon Looney and Jordan Bell go deeply under appreciated by outside observers. Houston made a gamble that they would be able to attack Steph, Looney, and Bell on switches over and over again and that they'd break. They didn't. Bell wasn't great, but Looney in particular, was a godsend by virtue of just holding up and providing about neutral value on a huge volume of switches. Looney averaged 19.3 minutes a game in the WCF. He defended 10.4 FGA a game in those 19.3 minutes and only gave up a difference of 0.3 DFG vs expected FG% and Bell defended 8.5 FGA a game in 14.2 minutes for a +3.6 FG% . So pretty much every other minute, Looney and Bell were getting dragged into plays and asked to defend on an island against CP3 or Harden. Houston's bet didn't really pay off and the Warriors were able to hold on defensively by relying on a third year C who'd struggled with health issues since he came into the league and an undersized C they found in the second round.

Maybe those numbers don't really seem all that impressive, after all, Looney was basically neutral and Bell slightly negative on differential DFG/FG, but consider the alternatives once Andre Iguodala went down. The Warriors got 42 minutes combined from Javale (3 minutes in one game), Zaza (12 minutes in 3 games), and West (27 minutes in 4 games) for that entire series. McCaw, who had been useful in the previous year's playoffs, was recovering from a terrifying concussion and played 5 minutes in two games. Kerr effectively the put the kibosh on all of his vet C's and did his best to cobble together playable lineups between Dray at C with wings or Looney/Bell at the C.

Only five lineups played more than 10 minutes in that entire series. Curry/Thompson/Durant/Green/Looney played the most, 74 minutes and had a +20.5 net rating... despite only having a 101.3 ORTG, so that lineup absolutely crushed it on defense with an 80.8 DRTG. Death Lineup 2.0 played 56 minutes and only had a +2.5 net rating and a 114 DRTG. The starters with Bell had a +18.2 net rating in 39 minutes, but their defense was merely good at 103.8 DRTG, compared to the godly DRTG of the stares with Looney. An odd lineup that worked really well: Steph/Klay/Young/Livingston/Dray struggled to score and had a 102.4 ORTG, but... a 68.3 DRTG in 19 minutes. The Death Lineup with Livingston in place of Andre was fine with a +7.8 net rating, but was pretty eh on defense and gave up a 111.1 DRTG. It's hard to draw much from lineups that played less than 10 minutes but more than 5 minutes, but the Looney or Bell at C lineups all held up really well on defense, whereas the West or vet C lineups got torched on D.

So bringing it back to the question of the 2016-17 Warriors vs the 2017-18 Warriors: yes, a healthy 2016-17 team probably wins against Houston, assuming a happy and well-integrated KD, but I do believe that the vet C's still would have gotten destroyed and then kiboshed leaving us with... a rookie Damion Jones, a second year Looney who isn't ready for rotation minutes, and James Michael McAdoo (11 minutes in 4 games) as your remaining big men. Matt Barnes helps for sure as a weird mix of Andre and Dray's skillsets, but he was hurt for most of the playoffs, so we never really got to see how he held up in as a Warrior in playoff minutes. McCaw was solid for us, but if you're going to play McCaw lots of minutes, you need Barnes, Andre, and Livingston to play lots of minutes at the PF next to Draymond. Ian Clark was useful and had good chemistry with David West on offense, but he was a minus defender and Nick Young proved a totally passable defender when he got ISO'd against Houston.

So I do think that the 2016-17 Warriors vs 2017-18 Houston would be closer than a lot of people think. I suspect the Warrior still win by virtue of KD being happy and compliant within the offensive system, more on that later, but if Andre goes down, the Warriors absolutely lose, even if CP3 goes down. There's just no way you can cobble together 48 minutes of a passable frontcourt between Dray/Matt Barnes/Livingston/McAdoo against that version of the Rockets.

But back to the KD thing. In his All the Smoke interview, Draymond talked at length about the KD situation. One of the most illuminating things he said was that he actually thought there was a considerable chance KD would leave at the end of the 2018 season. Draymond made it sound like he had to convince KD to just suck up one more season for the chance at a threepeat and he laid out a few things that painted a picture of a disgruntled KD in 2017-18. Draymond said that KD was very hurt by not getting crowned the best player in the world after his Finals MVP performance against the Cavs. Dray pretty much insinuated that the rejection KD felt leaked onto his oncourt behavior and KD got paranoid and defensive. One prime example Dray made was that KD would complain to him that Steph and Klay were intentionally looking him off to get their own shots, which is not a conclusion I think most people would make unless they got into their head that they weren't getting the recognition they received and were struggling with their own ego. So Dray says that he and Iguodala would go out of their way to call plays for KD and to push Kerr to do so in order to keep KD happy.

So what happens in the WCF? KD falls right into Houston's trap when they switch and gets ISO happy. As OP pointed out, when KD passed, the Warriors all but one game and they had a huge margin of victory. When KD decided instead to just ISO to try and prove... I don't know... that the Warriors NEEDED him to play hero ball to win... Houston won games. I do suspect that a happier KD in 2016-17 who isn't on a revenge tour against his detractors makes more sound decisions against Houston and that probably helps us, but yeah, KD was a weird variable for us against Houston.

Part of that I think is on Kerr. Something I noticed over the years is that it seemed like a lot of the ISO's KD got under Kerr occurred in the between the elbow and the baseline in the mid post. There is no disputing that KD is an incredible ISO scorer, but I think that there's a somewhat limited set of outcomes from a midpost ISO, whereas a top of the key ISO probably has a more broad set of outcomes and makes the defense work harder. When you drive from the mid post I just think it's harder to truly collapse the defense and bring multiple defenders to you, which in turn limits your passing options. You have a more limited view of the court and it's not as easy to get deep in the paint and kick the ball out to ANY place on the court, whereas from the top of the key, you have a more clear path to kicking it to EITHER corner/wing at any point in the drive and if you get deep enough, you might be able to whip a pass back to the top of they key if you have that type of vision.

I would never yearn for Mike Brown over Kerr, but one thing I appreciated about Brown's brief stint as the Warriors coach in the 2016-17 playoffs when Kerr missed games was that Brown had 0 qualms about just... letting KD ISO or PNR at the top of the key. It was as simple as dragging Rudy Gobert or one of the hapless Portland defenders to the top of the key and then KD would shoot over them or drive by them, at which point KD could dunk it, kick it out to shooters, etc. On a similar note, I felt like the 2019 WCSF against Houston was effectively a coin flip until KD went down, at which point I felt pretty confident that the Warriors would win the series. It's not like the Warriors were more talented without KD, but I expected that the offense wouldn't stagnate into midpost ISO's as often and that in crunch time, the Warriors could run the Steph/Dray PNR without concerning themselves with KD's feelings. Lo and behold in Game 6, that's pretty much all they did in the fourth quarter and in a nice twist, the ball moved around! One of my principal complaints with Kerr is his ideological opposition to PNR, which seems to be driven by a misguided belief that the PNR is predictable and doesn't stimulate ball movement. I really don't think that's true at all, especially with the Steph/Dray PNR, which would end so often in Dray lobbing it to the dunker on the baseline or kicking it out to the wing for Klay or Andre to shoot or pass it and force defenses to rotate in a panic for fear of the ball coming back to Steph, but I digress...

I do think it's pretty clear that the 2016-17 Warriors were a happier, and probably better team as a result than the 2017-18 Warriors. It's remarkable how much can get weird when one of your key players starts to break the offense to satisfy their own urges. But I also think there's a compelling argument that the 2017-18 Warriors have a better and more balanced roster than the 2016-17 team. I stand by my prediction that the 17-18 team was better equipped for Houston and that the 16-17 team would absolutely not be able to withstand the Andre injury.

Totally nailed it on this. I don't care if Looney can't jump over a single sheet of paper now, he'll always be good in my book based off how he guarded switches in that Rockets series. He's a really smart defensive player and it's a shame injuries have capped his ceiling.

Also getting bogged in to the high post iso with KD, especially with how well Tucker was guarding it, annoyed me to no end in 2019. Had the same feeling after KD went down (obviously I thought it screwed us longer term, but in that series I knew we would take it by going back to what they had no hope of guarding) If Kerr wasn't such a trianglehead, we would have likely seen a ton more KD running PNR or even just isoing from the wing and I don't think Tucker would hold up nearly as well

[Warriors Analyst]

Some thoughts I have on this team. I think this might actually be the most well balanced roster the Warriors had in their KD years. Was it the best team? Probably not. OP makes a convincing argument with numbers to back it that KD's desire to freelance hurt the Warriors against Houston AND that the series would have been over in 5-6 games with both rosters fully healthy. The obvious argument that the 2016-17 Warriors have over the 2017-18 roster is better on-court chemistry. That team probably beats Houston at full health too, but without Andre, I don't think that the the 2016-17 Warriors would have been able to pull it off against Houston.

The contributions of Kevon Looney and Jordan Bell go deeply under appreciated by outside observers. Houston made a gamble that they would be able to attack Steph, Looney, and Bell on switches over and over again and that they'd break. They didn't. Bell wasn't great, but Looney in particular, was a godsend by virtue of just holding up and providing about neutral value on a huge volume of switches. Looney averaged 19.3 minutes a game in the WCF. He defended 10.4 FGA a game in those 19.3 minutes and only gave up a difference of 0.3 DFG vs expected FG% and Bell defended 8.5 FGA a game in 14.2 minutes for a +3.6 FG% . So pretty much every other minute, Looney and Bell were getting dragged into plays and asked to defend on an island against CP3 or Harden. Houston's bet didn't really pay off and the Warriors were able to hold on defensively by relying on a third year C who'd struggled with health issues since he came into the league and an undersized C they found in the second round.

Maybe those numbers don't really seem all that impressive, after all, Looney was basically neutral and Bell slightly negative on differential DFG/FG, but consider the alternatives once Andre Iguodala went down. The Warriors got 42 minutes combined from Javale (3 minutes in one game), Zaza (12 minutes in 3 games), and West (27 minutes in 4 games) for that entire series. McCaw, who had been useful in the previous year's playoffs, was recovering from a terrifying concussion and played 5 minutes in two games. Kerr effectively the put the kibosh on all of his vet C's and did his best to cobble together playable lineups between Dray at C with wings or Looney/Bell at the C.

Only five lineups played more than 10 minutes in that entire series. Curry/Thompson/Durant/Green/Looney played the most, 74 minutes and had a +20.5 net rating... despite only having a 101.3 ORTG, so that lineup absolutely crushed it on defense with an 80.8 DRTG. Death Lineup 2.0 played 56 minutes and only had a +2.5 net rating and a 114 DRTG. The starters with Bell had a +18.2 net rating in 39 minutes, but their defense was merely good at 103.8 DRTG, compared to the godly DRTG of the stares with Looney. An odd lineup that worked really well: Steph/Klay/Young/Livingston/Dray struggled to score and had a 102.4 ORTG, but... a 68.3 DRTG in 19 minutes. The Death Lineup with Livingston in place of Andre was fine with a +7.8 net rating, but was pretty eh on defense and gave up a 111.1 DRTG. It's hard to draw much from lineups that played less than 10 minutes but more than 5 minutes, but the Looney or Bell at C lineups all held up really well on defense, whereas the West or vet C lineups got torched on D.

So bringing it back to the question of the 2016-17 Warriors vs the 2017-18 Warriors: yes, a healthy 2016-17 team probably wins against Houston, assuming a happy and well-integrated KD, but I do believe that the vet C's still would have gotten destroyed and then kiboshed leaving us with... a rookie Damion Jones, a second year Looney who isn't ready for rotation minutes, and James Michael McAdoo (11 minutes in 4 games) as your remaining big men. Matt Barnes helps for sure as a weird mix of Andre and Dray's skillsets, but he was hurt for most of the playoffs, so we never really got to see how he held up in as a Warrior in playoff minutes. McCaw was solid for us, but if you're going to play McCaw lots of minutes, you need Barnes, Andre, and Livingston to play lots of minutes at the PF next to Draymond. Ian Clark was useful and had good chemistry with David West on offense, but he was a minus defender and Nick Young proved a totally passable defender when he got ISO'd against Houston.

So I do think that the 2016-17 Warriors vs 2017-18 Houston would be closer than a lot of people think. I suspect the Warrior still win by virtue of KD being happy and compliant within the offensive system, more on that later, but if Andre goes down, the Warriors absolutely lose, even if CP3 goes down. There's just no way you can cobble together 48 minutes of a passable frontcourt between Dray/Matt Barnes/Livingston/McAdoo against that version of the Rockets.

But back to the KD thing. In his All the Smoke interview, Draymond talked at length about the KD situation. One of the most illuminating things he said was that he actually thought there was a considerable chance KD would leave at the end of the 2018 season. Draymond made it sound like he had to convince KD to just suck up one more season for the chance at a threepeat and he laid out a few things that painted a picture of a disgruntled KD in 2017-18. Draymond said that KD was very hurt by not getting crowned the best player in the world after his Finals MVP performance against the Cavs. Dray pretty much insinuated that the rejection KD felt leaked onto his oncourt behavior and KD got paranoid and defensive. One prime example Dray made was that KD would complain to him that Steph and Klay were intentionally looking him off to get their own shots, which is not a conclusion I think most people would make unless they got into their head that they weren't getting the recognition they received and were struggling with their own ego. So Dray says that he and Iguodala would go out of their way to call plays for KD and to push Kerr to do so in order to keep KD happy.

So what happens in the WCF? KD falls right into Houston's trap when they switch and gets ISO happy. As OP pointed out, when KD passed, the Warriors all but one game and they had a huge margin of victory. When KD decided instead to just ISO to try and prove... I don't know... that the Warriors NEEDED him to play hero ball to win... Houston won games. I do suspect that a happier KD in 2016-17 who isn't on a revenge tour against his detractors makes more sound decisions against Houston and that probably helps us, but yeah, KD was a weird variable for us against Houston.

Part of that I think is on Kerr. Something I noticed over the years is that it seemed like a lot of the ISO's KD got under Kerr occurred in the between the elbow and the baseline in the mid post. There is no disputing that KD is an incredible ISO scorer, but I think that there's a somewhat limited set of outcomes from a midpost ISO, whereas a top of the key ISO probably has a more broad set of outcomes and makes the defense work harder. When you drive from the mid post I just think it's harder to truly collapse the defense and bring multiple defenders to you, which in turn limits your passing options. You have a more limited view of the court and it's not as easy to get deep in the paint and kick the ball out to ANY place on the court, whereas from the top of the key, you have a more clear path to kicking it to EITHER corner/wing at any point in the drive and if you get deep enough, you might be able to whip a pass back to the top of they key if you have that type of vision.

I would never yearn for Mike Brown over Kerr, but one thing I appreciated about Brown's brief stint as the Warriors coach in the 2016-17 playoffs when Kerr missed games was that Brown had 0 qualms about just... letting KD ISO or PNR at the top of the key. It was as simple as dragging Rudy Gobert or one of the hapless Portland defenders to the top of the key and then KD would shoot over them or drive by them, at which point KD could dunk it, kick it out to shooters, etc. On a similar note, I felt like the 2019 WCSF against Houston was effectively a coin flip until KD went down, at which point I felt pretty confident that the Warriors would win the series. It's not like the Warriors were more talented without KD, but I expected that the offense wouldn't stagnate into midpost ISO's as often and that in crunch time, the Warriors could run the Steph/Dray PNR without concerning themselves with KD's feelings. Lo and behold in Game 6, that's pretty much all they did in the fourth quarter and in a nice twist, the ball moved around! One of my principal complaints with Kerr is his ideological opposition to PNR, which seems to be driven by a misguided belief that the PNR is predictable and doesn't stimulate ball movement. I really don't think that's true at all, especially with the Steph/Dray PNR, which would end so often in Dray lobbing it to the dunker on the baseline or kicking it out to the wing for Klay or Andre to shoot or pass it and force defenses to rotate in a panic for fear of the ball coming back to Steph, but I digress...

I do think it's pretty clear that the 2016-17 Warriors were a happier, and probably better team as a result than the 2017-18 Warriors. It's remarkable how much can get weird when one of your key players starts to break the offense to satisfy their own urges. But I also think there's a compelling argument that the 2017-18 Warriors have a better and more balanced roster than the 2016-17 team. I stand by my prediction that the 17-18 team was better equipped for Houston and that the 16-17 team would absolutely not be able to withstand the Andre injury.

Totally nailed it on this. I don't care if Looney can't jump over a single sheet of paper now, he'll always be good in my book based off how he guarded switches in that Rockets series. He's a really smart defensive player and it's a shame injuries have capped his ceiling.

Also getting bogged in to the high post iso with KD, especially with how well Tucker was guarding it, annoyed me to no end in 2019. Had the same feeling after KD went down (obviously I thought it screwed us longer term, but in that series I knew we would take it by going back to what they had no hope of guarding) If Kerr wasn't such a trianglehead, we would have likely seen a ton more KD running PNR or even just isoing from the wing and I don't think Tucker would hold up nearly as well

The KD injury was so bizarre. When he came back against Toronto I felt so supremely confident in the Warriors’ chances at the comeback in part because KD would be somewhat limited physically. That probably sounds insane in the abstract, but paradoxically a limited KD who isn’t going to demand high volumes of ISO’s and will instead just finish possessions in a super Harrison Barnes role was just what we needed to keep the offense humming. He was so good in game 5 until he went down. Alas.

Glad you liked my write up and thoughts on Looney. I really adore his game. He’s such a brilliant player despite glaring physical limitations. I’ve wondered if those limitations helped turn him into the player he is now — in high school, he was the KD of Wisconsin high school basketball. His highlights are littered with step backs, dribble pull ups, ball handling prowess. It’s absolutely hilarious to see now. Even this year when he’s been physically hampered, he’s been a plus minus god, as is expected. I wish casual Warriors fans better saw his value.

[parsnips33]

Spoiler:

Some thoughts I have on this team. I think this might actually be the most well balanced roster the Warriors had in their KD years. Was it the best team? Probably not. OP makes a convincing argument with numbers to back it that KD's desire to freelance hurt the Warriors against Houston AND that the series would have been over in 5-6 games with both rosters fully healthy. The obvious argument that the 2016-17 Warriors have over the 2017-18 roster is better on-court chemistry. That team probably beats Houston at full health too, but without Andre, I don't think that the the 2016-17 Warriors would have been able to pull it off against Houston.

The contributions of Kevon Looney and Jordan Bell go deeply under appreciated by outside observers. Houston made a gamble that they would be able to attack Steph, Looney, and Bell on switches over and over again and that they'd break. They didn't. Bell wasn't great, but Looney in particular, was a godsend by virtue of just holding up and providing about neutral value on a huge volume of switches. Looney averaged 19.3 minutes a game in the WCF. He defended 10.4 FGA a game in those 19.3 minutes and only gave up a difference of 0.3 DFG vs expected FG% and Bell defended 8.5 FGA a game in 14.2 minutes for a +3.6 FG% . So pretty much every other minute, Looney and Bell were getting dragged into plays and asked to defend on an island against CP3 or Harden. Houston's bet didn't really pay off and the Warriors were able to hold on defensively by relying on a third year C who'd struggled with health issues since he came into the league and an undersized C they found in the second round.

Maybe those numbers don't really seem all that impressive, after all, Looney was basically neutral and Bell slightly negative on differential DFG/FG, but consider the alternatives once Andre Iguodala went down. The Warriors got 42 minutes combined from Javale (3 minutes in one game), Zaza (12 minutes in 3 games), and West (27 minutes in 4 games) for that entire series. McCaw, who had been useful in the previous year's playoffs, was recovering from a terrifying concussion and played 5 minutes in two games. Kerr effectively the put the kibosh on all of his vet C's and did his best to cobble together playable lineups between Dray at C with wings or Looney/Bell at the C.

Only five lineups played more than 10 minutes in that entire series. Curry/Thompson/Durant/Green/Looney played the most, 74 minutes and had a +20.5 net rating... despite only having a 101.3 ORTG, so that lineup absolutely crushed it on defense with an 80.8 DRTG. Death Lineup 2.0 played 56 minutes and only had a +2.5 net rating and a 114 DRTG. The starters with Bell had a +18.2 net rating in 39 minutes, but their defense was merely good at 103.8 DRTG, compared to the godly DRTG of the stares with Looney. An odd lineup that worked really well: Steph/Klay/Young/Livingston/Dray struggled to score and had a 102.4 ORTG, but... a 68.3 DRTG in 19 minutes. The Death Lineup with Livingston in place of Andre was fine with a +7.8 net rating, but was pretty eh on defense and gave up a 111.1 DRTG. It's hard to draw much from lineups that played less than 10 minutes but more than 5 minutes, but the Looney or Bell at C lineups all held up really well on defense, whereas the West or vet C lineups got torched on D.

So bringing it back to the question of the 2016-17 Warriors vs the 2017-18 Warriors: yes, a healthy 2016-17 team probably wins against Houston, assuming a happy and well-integrated KD, but I do believe that the vet C's still would have gotten destroyed and then kiboshed leaving us with... a rookie Damion Jones, a second year Looney who isn't ready for rotation minutes, and James Michael McAdoo (11 minutes in 4 games) as your remaining big men. Matt Barnes helps for sure as a weird mix of Andre and Dray's skillsets, but he was hurt for most of the playoffs, so we never really got to see how he held up in as a Warrior in playoff minutes. McCaw was solid for us, but if you're going to play McCaw lots of minutes, you need Barnes, Andre, and Livingston to play lots of minutes at the PF next to Draymond. Ian Clark was useful and had good chemistry with David West on offense, but he was a minus defender and Nick Young proved a totally passable defender when he got ISO'd against Houston.

So I do think that the 2016-17 Warriors vs 2017-18 Houston would be closer than a lot of people think. I suspect the Warrior still win by virtue of KD being happy and compliant within the offensive system, more on that later, but if Andre goes down, the Warriors absolutely lose, even if CP3 goes down. There's just no way you can cobble together 48 minutes of a passable frontcourt between Dray/Matt Barnes/Livingston/McAdoo against that version of the Rockets.

But back to the KD thing. In his All the Smoke interview, Draymond talked at length about the KD situation. One of the most illuminating things he said was that he actually thought there was a considerable chance KD would leave at the end of the 2018 season. Draymond made it sound like he had to convince KD to just suck up one more season for the chance at a threepeat and he laid out a few things that painted a picture of a disgruntled KD in 2017-18. Draymond said that KD was very hurt by not getting crowned the best player in the world after his Finals MVP performance against the Cavs. Dray pretty much insinuated that the rejection KD felt leaked onto his oncourt behavior and KD got paranoid and defensive. One prime example Dray made was that KD would complain to him that Steph and Klay were intentionally looking him off to get their own shots, which is not a conclusion I think most people would make unless they got into their head that they weren't getting the recognition they received and were struggling with their own ego. So Dray says that he and Iguodala would go out of their way to call plays for KD and to push Kerr to do so in order to keep KD happy.

So what happens in the WCF? KD falls right into Houston's trap when they switch and gets ISO happy. As OP pointed out, when KD passed, the Warriors all but one game and they had a huge margin of victory. When KD decided instead to just ISO to try and prove... I don't know... that the Warriors NEEDED him to play hero ball to win... Houston won games. I do suspect that a happier KD in 2016-17 who isn't on a revenge tour against his detractors makes more sound decisions against Houston and that probably helps us, but yeah, KD was a weird variable for us against Houston.

Part of that I think is on Kerr. Something I noticed over the years is that it seemed like a lot of the ISO's KD got under Kerr occurred in the between the elbow and the baseline in the mid post. There is no disputing that KD is an incredible ISO scorer, but I think that there's a somewhat limited set of outcomes from a midpost ISO, whereas a top of the key ISO probably has a more broad set of outcomes and makes the defense work harder. When you drive from the mid post I just think it's harder to truly collapse the defense and bring multiple defenders to you, which in turn limits your passing options. You have a more limited view of the court and it's not as easy to get deep in the paint and kick the ball out to ANY place on the court, whereas from the top of the key, you have a more clear path to kicking it to EITHER corner/wing at any point in the drive and if you get deep enough, you might be able to whip a pass back to the top of they key if you have that type of vision.

I would never yearn for Mike Brown over Kerr, but one thing I appreciated about Brown's brief stint as the Warriors coach in the 2016-17 playoffs when Kerr missed games was that Brown had 0 qualms about just... letting KD ISO or PNR at the top of the key. It was as simple as dragging Rudy Gobert or one of the hapless Portland defenders to the top of the key and then KD would shoot over them or drive by them, at which point KD could dunk it, kick it out to shooters, etc. On a similar note, I felt like the 2019 WCSF against Houston was effectively a coin flip until KD went down, at which point I felt pretty confident that the Warriors would win the series. It's not like the Warriors were more talented without KD, but I expected that the offense wouldn't stagnate into midpost ISO's as often and that in crunch time, the Warriors could run the Steph/Dray PNR without concerning themselves with KD's feelings. Lo and behold in Game 6, that's pretty much all they did in the fourth quarter and in a nice twist, the ball moved around! One of my principal complaints with Kerr is his ideological opposition to PNR, which seems to be driven by a misguided belief that the PNR is predictable and doesn't stimulate ball movement. I really don't think that's true at all, especially with the Steph/Dray PNR, which would end so often in Dray lobbing it to the dunker on the baseline or kicking it out to the wing for Klay or Andre to shoot or pass it and force defenses to rotate in a panic for fear of the ball coming back to Steph, but I digress...

I do think it's pretty clear that the 2016-17 Warriors were a happier, and probably better team as a result than the 2017-18 Warriors. It's remarkable how much can get weird when one of your key players starts to break the offense to satisfy their own urges. But I also think there's a compelling argument that the 2017-18 Warriors have a better and more balanced roster than the 2016-17 team. I stand by my prediction that the 17-18 team was better equipped for Houston and that the 16-17 team would absolutely not be able to withstand the Andre injury.

Totally nailed it on this. I don't care if Looney can't jump over a single sheet of paper now, he'll always be good in my book based off how he guarded switches in that Rockets series. He's a really smart defensive player and it's a shame injuries have capped his ceiling.

Also getting bogged in to the high post iso with KD, especially with how well Tucker was guarding it, annoyed me to no end in 2019. Had the same feeling after KD went down (obviously I thought it screwed us longer term, but in that series I knew we would take it by going back to what they had no hope of guarding) If Kerr wasn't such a trianglehead, we would have likely seen a ton more KD running PNR or even just isoing from the wing and I don't think Tucker would hold up nearly as well

The KD injury was so bizarre. When he came back against Toronto I felt so supremely confident in the Warriors’ chances at the comeback in part because KD would be somewhat limited physically. That probably sounds insane in the abstract, but paradoxically a limited KD who isn’t going to demand high volumes of ISO’s and will instead just finish possessions in a super Harrison Barnes role was just what we needed to keep the offense humming. He was so good in game 5 until he went down. Alas.

Glad you liked my write up and thoughts on Looney. I really adore his game. He’s such a brilliant player despite glaring physical limitations. I’ve wondered if those limitations helped turn him into the player he is now — in high school, he was the KD of Wisconsin high school basketball. His highlights are littered with step backs, dribble pull ups, ball handling prowess. It’s absolutely hilarious to see now. Even this year when he’s been physically hampered, he’s been a plus minus god, as is expected. I wish casual Warriors fans better saw his value.

Looney's game now reminds me of the martial arts concept of "no wasted movement". His defense, even with how limited his mobility has become, is still so solid because what few moves he does make are almost always right. He's never gonna bite on a fake or get taken out of position by a dribble move because he conserves his movement so much, whether intentionally or not

Part of why he was so effective on Harden who is a master at faking guys off balance and out of position