One thing I'm watching for on each end as the most anticipated series of the playoffs resumes:
Stephen Curry's defense
We know the Rockets are hunting Stephen Curry. That is not novel, and not something to do just because Curry is recovering from another ill-timed knee injury. The Cavaliers mastered this. It is a way to exhaust Curry -- and the Rockets want to bump and grind both Curry and Klay Thompson -- and poke at the only player in the Hamptons 5 who is not at least borderline All-Defense level when engaged.
(By the way: I don't love the Hamptons 5 nickname, but it appears to have won the day.)
When James Harden calls Curry's man up for a screen, he foists an untenable choice onto the Warriors: Switch Curry onto Harden, or have him lunge to cut Harden off, temporarily placing two defenders on Harden and leaving the back line of the defense vulnerable.
Curry has no shot defending Harden one-on-one. Harden overpowers him and bolts to the rim:
Subtly brilliant thing there (and on many similar plays): Clint Capela, the lone non-3-point shooter who saw any minutes for Houston in Game 2, plants himself along the right side of the rim. Harden slices down the left side, opposite Capela. That setup creates more space between the two -- making it tougher for Draymond Green to effectively guard both of them at once. It also opens a cleaner angle for a lob pass.
It's easy to say the Warriors should try harder to avoid the switch, but the Rockets are not passive recipients of Golden State's choices. They are forcing the Warriors into that switch. Trevor Ariza, forever underappreciated glue guy and Curry's most frequent hiding spot, will screen for Harden two and even three times -- as often as it takes until he pries enough separation for Golden State to surrender. Eric Gordon, another Curry hiding place, is a canny and physical screener.
Even so, there are possessions when Golden State gives up too easily. I'd like to see a bit more of this:
The execution isn't airtight, but the principle is interesting: Let Ariza roll into open space and force him to make plays. He can do that some, but the Warriors have the defensive personnel -- long, mean, anticipatory -- to fluster and delay him, so they can reset into 5-on-5 mode.
They stack help defenders off of Houston's weaker shooters -- Capela and PJ Tucker -- leaving a tricky crosscourt pass as Ariza's only option.
And the defenders involved -- Green and Andre Iguodala -- rotate back even as that pass is still on Ariza's fingertips.
Of course, Houston has some answers for this too. The Rockets can add stress by bumping up the shooting quotient and replacing either Tucker or Capela with Gordon. After experimenting with some wacky stuff in Game 1 -- Ryan Anderson and Gerald Green together, notably -- Houston landed on a nice balance of defense and shooting in Game 2. (For the record: I didn't think it was insane to try Anderson and Green when Curry and Kevin Durant sat together -- even though I didn't expect Anderson to play non-garbage minutes. That is the time to tilt the shooting-versus-defense equation more toward shooting. It became clear right way that Anderson-Green was a step too far, and Mike D'Antoni adjusted.)
The Rockets already have played Harden, Chris Paul, and Gordon 23 minutes together in two games, and they might dare to send that trio out longer in Game 3. They have limited the time any three of Ariza, Tucker, Luc Mbah a Moute and Capela/Nene Hilario play together almost entirely to the stints of the starting lineup -- which includes Tucker, Ariza, and Capela. They benched Mbah a Moute in the second half of Game 2 and played him and Tucker together just six minutes in Game 1.
More shooting makes extreme help rotations riskier.
The Warriors have experimented with a third counter to Houston's Curry hunting: switching Curry off of his man the moment that player starts moving toward Harden.
Houston can puncture that gambit, too. They can call up new screeners until Curry has no choice but to participate, though that drains the shot clock. They can set up the floor so that no Rocket clutters the pathway between Curry and Harden.
More promisingly: They can bait Golden State into that switch and exploit the holes that pop open as it is in motion. Gerald Green almost did that randomly in the clip above. His intent is to screen for Harden. The Warriors respond by having Kevon Looney slide away from Capela and onto Green. When Green and Capela crisscross, Green flashes open; Capela transforms into an accidental pindown screener. Harden appears to be the only one who sees the opportunity.
Prep for this, and Houston can be one step ahead.
Golden State's missing mojo
There are four general ways for Golden State to score:
1. Their fancy motion offense.
2. Transition after Houston turnovers or misses at the rim.
3. Transition that preys on Houston mistakes -- bad floor balance, failing to match up, etc.
4. Isolations.
Nos. 1 and 3 will determine the series. There isn't much for Houston to do about No. 2. Harden and Paul are going to miss some bunnies, and every team will commit some turnovers. You can suggest Houston should miss and cough it up less often. Duh. You can say that for any team, in any game. Stuff happens. Less of that stuff happening is good for Houston. Some games, they'll manage. Some games, they won't. I'm not sure what sort of planning they can really do, beyond having one of their corner shooters loop back on defense earlier than usual -- something that has been a clear point of emphasis.
They can exert some control over No. 3 -- and they did so in Game 2. In Game 1, Houston worried too much about finding the "right" matchups getting back on defense. The Warriors strolled into at least five open 3-pointers in Game 1 as Houston players crisscrossed the court, confused, scanning for appropriate matchups. More than almost anything else -- role players missing shots, their reliance on isolations reaching historic levels, losing the minutes Curry and Durant sat together -- those 3s cost Houston Game 1.
Amid chaos, Houston players just have to guard whoever is closest to them. That sort of flexibility is the whole point of playing five like-sized guys (when Tucker is at center) and switching everything. They did it in Game 2.
No. 4 -- isolations -- is happening a lot. The Warriors averaged 11.5 isolations per 100 possessions in the regular season, one of the lowest marks in the league, per Second Spectrum tracking data. That number is up to 27.8 so far in this series -- more than even Houston averaged in the regular season.
Golden State is a good isolation team! Durant, of course, is on fire, canning absolutely filthy shots. His one-on-one game is the antidote to precisely this sort of switch-everything defense. Golden State is averaging exactly one point per possession on isolations when the guy going one-on-one either shoots or passes to a teammate who shoots right away, per Second Spectrum. That would have ranked second in the regular season.
It also is worse than Golden State's overall efficiency. And that brings us to method of scoring No. 1: Houston can win this series, even with Durant raining daggers, if it forces a disproportionate share of Golden State's offense toward isolation play.
The Warriors know how to beat switches without abandoning their motion offense and resorting to one-on-one ball. They have all sorts of tricks. The most common: slipping screens before setting them.
It's really, really hard to switch a screen that never happens. You can try switching early, almost ahead of the screen, but if you telegraph that, the Warriors burn you. You can scrap the switch, but if you hesitate for even a beat, it's too late. On the above play, Harden tried to stay home on Iguodala. He couldn't do it; he had leaned toward half court for a half-second, girding himself for a switch, and that tiny bit of misdirected momentum was all Iguodala needed.
Here's Draymond Green pulling a different version of the same trick:
Watch Tucker's feet as he defends Green. Green doesn't shift up toward half court to fake a screen for Curry. He waits for Curry to creep down toward him at the 3-point arc, then moves almost exactly parallel to the baseline on his initial cut toward Curry.
Green also makes both of his cuts -- toward Curry, then toward the rim -- hard. The net effect is that Tucker ends up taking one step backward, toward the rim, as Green and Curry meet; Tucker plants his right foot in the paint. That step undoes Houston's defense. Tucker has more ground to cover switching onto Curry, and Tucker knows he has to get there immediately. Curry leverages that panic against him with a blow-by drive.
The Warriors are so good at little things that gain them a few inches. Their offense comes to life in that extra space. Watch how Shaun Livingston moves down into the paint to screen for Thompson, and even play-acts the first part of a hard cut to the rim -- a fake slip, basically -- before setting the pick:
That theatrical spasm gets Gerald Green backpedaling. Look at the gap between Green and Thompson when Livingston's pick makes contact:
Green has no chance to switch onto Thompson from there. Instead, Gordon has to chase Thompson, only Livingston's pick leaves him so far behind that the Rockets have to switch again -- with Anderson taking Thompson. That mismatch unlocks an open 3 for Nick Young.
Along the same lines, I loved this gem from Game 1:
Curry pretends he's heading into a handoff from Looney, waits for Ariza to overreact -- and potentially prepare for a switch -- and veers the other way for a quick-hitter. This is a cousin of Draymond Green's play-action keeper, where he fakes a handoff, fools defenders into switching or lunging at Green's bogus hand-off recipient, and then jaunts off for a layup.
The story of Game 2, on this end of the floor, was Houston vaporizing all that stuff. The Rockets read almost every slip and feint. They switched early, without ceding any territorial advantage. They aborted the switch when that was the best way of maintaining defensive integrity, but never when doing so would put them hopelessly behind. When a Warrior did slip open toward the rim, a help defender abandoned Green or Iguodala to block his path and force Golden State to swing the ball out of the danger zone. Paul was masterful at apparating into help position and deflecting kickout passes.
(Green and Iguodala are 0-of-5 from deep combined through two games. Tucker made as many 3s in Game 2 as Golden State's blah shooters have attempted in the entire series. The Warriors need Iguodala and Green to hit shots. At some point -- probably at home -- they will.)
There was no adjustment. Houston was just better.
A cycle of sorts played out in Game 2: Houston smothered Golden State's beautiful game and nudged them toward one-on-one play. Durant obliged. But when the Warriors tried re-engage their motion offense, they discovered the detour away from it had sapped their usual rhythm. They either didn't cut or cut at three-quarters speed. At one point in the second quarter, Steve Kerr stood up and angrily motioned for his team to get moving.
David West attempted two post-ups from midrange early in the shot clock. Iguodala jacked one. Thompson dribbled around for a while and launched an inexplicable baseline fadeaway with double digits left on the shot clock. A few minutes later, Thompson took a rare contested off-the-bounce 3 from above the break -- also with a ton of time remaining on the shot clock.
The Warriors didn't look like themselves. But the Rockets deserve a lot of the credit. They erased Golden State's easiest looks -- the ones that get the Warriors flexing and running. Houston lured Golden State into a game the Warriors don't really want to play, even if they can. They unnerved the Warriors a little.
The Rockets were even smart about double-teaming Durant when the Warriors played three non-shooters around him -- say, Livingston, Iguodala and Looney -- making it easier to trap without surrendering an open look.
Looney being such a key player is a reminder of Golden State's strange roster construction. They have six centers who have very little utility in this series. Outside the Hamptons 5 they have only two reliable wings, and one of them is Nick freaking Young. They don't have a true small-ball power forward beyond Durant. They haven't used Durant at center in part because they don't quite have the wing depth to arrive there organically while starting Iguodala. The Rockets might have more viable five-out lineups than the team that perfected the five-out lineup.
Patrick McCaw's falloff and injury loom large in that regard. Omri Casspi might have been useful. They could unearth Quinn Cook at some point, but I just got a bit queasy at the thought of Cook defending the Rockets.
But that's the battleground I'll be watching in Game 3: Can the Warriors cut and pass and screen their way into a half-dozen more buckets? They are still in good shape. They have home court. Curry will come alive. Houston needed Game 2 and played like it.
But if Houston can erase those half-dozen buckets, it has a chance.