For a half, game two of the NBA Finals had all the hallmarks of an epic showdown. Then, the Golden State Warriors went on a 35-14 tear in the third quarter and turned a tight contest into a laugher. The Boston Celtics fell behind by as much as 29 and trailed by so much that Golden State’s Stephen Curry didn’t have to play in the fourth quarter. The Warriors won, 107-88.

In the early going, Golden State looked flustered by Boston’s defensive pressure. The Celtics were intent on turning Curry into a driver instead of letting him launch threes, and while Curry made effective passes to set up teammates, his teammates kept missing layups and open shots.

Boston’s inability to control the ball gave the Warriors life. The Celtics had seven turnovers in the first 12 minutes and finished with 18 for the game. Boston also showcased some questionable shot selection, especially from Jayson Tatum.

In one of the weirder moments of the broadcast, ESPN’s Mark Jackson praised Tatum when he missed a midrange jumper over defensive liability Jordan Poole. This was classic “letting the defense off the hook” decision-making.

Still, Tatum wasn’t the problem for Boston. That “honor” goes to his fellow starters, Marcus Smart (2 points, 2 rebounds, 5 assists, 5 turnovers, 4 fouls), Jaylen Brown (5-17 shooting), and Al Horford (2 points, 8 rebounds and step-slow defense). Robert Williams III was similarly unproductive (just 2 points, 2 rebounds and 2 blocks in 14 minutes), though he was also playing on a gimpy knee.

As would be expected in a blowout win, the Warriors got contributions from nearly everyone in the rotation. Curry did the heavy lifting on offense — as a passer, decoy, and shot-maker. The Warriors’ offense found its greatest success in the third quarter when they did something somewhat atypical for them: high pick-and-roll sets with Curry as the ball handler.

Perhaps the only sour note for Golden State was Klay Thompson’s 4-19 brickfest. Head coach Steve Kerr left Thompson in the game through much of the fourth quarter, seemingly to get him going. While Thompson did hit a long two off a pindown, he was just 1-5 in the period.

Game three on Wednesday night could yield some interesting tactical adjustments. Boston has to figure out how to corral Curry. The challenge is that Golden State can run their standard read-and-react offensive sets, and if those aren’t working, give Curry a ball screen and let him go to work.

My guess is that Boston will try trapping and blitzing Curry on pick-and-rolls and hope that he’ll make some bad passes or that his teammates will miss shots. One difficulty with that is that the Warriors can start the action from near midcourt because of Curry’s preposterous shooting range. That gives Celtics defenders with long runs to get back into the action, and the Warriors excel in those kinds of power play situations.

Golden State’s defensive game plan for game two seemed very similar to game one. They’re conceding some threes in an effort to keep Tatum and Brown away from the paint. In game one, Boston made those long balls. In game two, Tatum was 6-9 from deep while the rest of the team shot just 9-28 — 32.1%.

Boston has had success throughout the playoffs hunting weaker defenders. They’ll likely persist, though Golden State’s superb rotations and help defenders create some challenges. Perhaps in game three, the Celtics need to make the Seth Partnow “play better” adjustment.

The Four Factors

Below are the four factors that decide who wins and loses in basketball — shooting (efg), rebounding (offensive rebounds), ball handling (turnovers), fouling (free throws made).

I’ve simplified them a bit. While the factors are usually presented as percentages, that’s more useful over a full season. In a single game, the raw numbers in each category are easier to understand.

Key Stats

Below are a few performance metrics, including the Player Production Average (PPA) Game Score (very similar to the one I used to call Scoreboard Impact Rating). PPA is my overall production metric, which credits players for things they do that help a team win (scoring, rebounding, playmaking, defending) and dings them for things that hurt (missed shots, turnovers, bad defense, fouls).

Game Score (GmSC) converts individual production into points on the scoreboard. The scale is the same as points and reflects each player’s total contributions for the game. The lowest possible GmSC is zero.

PPA is a per possession metric designed for larger data sets. In small sample sizes, the numbers can get weird. But some readers prefer it, so I’m including PPA scores as well. Reminder: in PPA, 100 is average, higher is better and replacement level is 45. For a single game, replacement level isn’t much use, and I reiterate the caution about small samples producing weird results.

POSS is the number of possessions each player was on the floor in this game.

PTS = points scored

ORTG = offensive rating, which is points produced per individual possessions x 100. League average this season is 111.7. Points produced is not the same as points scored. It includes the value of assists and offensive rebounds, as well as sharing credit when receiving an assist.

USG = offensive usage rate. Average is 20%.

ORTG and USG are versions of stats created by Wizards assistant coach Dean Oliver and modified slightly by me. ORTG is an efficiency measure that accounts for the value of shooting, offensive rebounds, assists and turnovers. USG includes shooting from the floor and free throw line, offensive rebounds, assists and turnovers.