With a 2-0 series lead in the NBA Finals heading to Cleveland, the FiveThirtyEight NBA Elo model gives the Golden State Warriors an 89 percent chance of repeating as champions, which — admit it — feels low. I’m personally on record as being more bullish about the Warriors’ chances than the models, mostly because the playoffs are typically super-deterministic, and the Warriors — as defending champs and the winningest team in history — seem super-determined.
Moreover, they’re now 4-0 against the Cavaliers this season, winning by an average of 22 points per contest. Kevin Love will reportedly be out in Game 3 — though he was also out the last time Cavs put up much resistance against the champs, in last year’s finals. To take this series, the Warriors need only win two of their five remaining games — a feat they have accomplished in every five-game stretch they’ve played this season (they have only lost three of five once, against the Thunder in the Western Conference Finals).
But there’s one other thing about the Warriors that has flown under the radar and may also speak well of their chances: their complete lack of regard for their opponent.
Simply put, the Warriors haven’t been losing much more against stronger teams than against weaker ones. In the regular season, there was literally zero relationship between the strength of the Warriors’ opponents and how well they performed. Against teams that won 50 games or more, they went 14-1. Of their eight losses suffered with Stephen Curry playing, four came against teams with losing records — including a schlubbing at the hands of the 17-win Lakers. Their “trend” actually has a slightly downward slope at either location:
Now, it’s a crazy result, but does it make any sense?1
Listen to the latest episode of our sports podcast Hot Takedown.
Here’s the good news for the Cavs: Three of the five remaining games are in Cleveland, and the Warriors have been significantly weaker on the road than at home. Their average margin was 7.4 points lower on the road (relative to expectation based on SRS difference) than at home –- good for the league’s sixth-largest such differential. That’s pretty remarkable for a team whose road win percentage was higher than the vaunted Spurs’ overall win percentage (the Road Warriors were the equivalent of a 68-win team).
But better news for the Cavs is that the Warriors haven’t really kept this phenomenon going in the playoffs. There, they’ve looked mortal and lost to stronger teams more often than they have to weaker ones:
Of course, with such small samples, already shaky trends are even less meaningful. The two huge losses on the road against Oklahoma City are enough to drag down the Warriors’ playoff road results. The Warriors have lost four of seven road games in these playoffs, with two of those losses coming with Curry out of the line-up. Further, it’s not really clear whether Curry was back to 100 percent during the Warriors’ worst stretch of the season against the Thunder — or if he is now.
For comparison, let’s take a look at Cleveland:
The Cavs — who made the second-most 3-pointers during the regular season — also had a weakish relationship between opponent strength and performance. However, they were certainly closer to standard than Golden State was, particularly at home (including a 34-point loss to the Warriors in January). But they have yet to lose at home in these playoffs.
Now, this may seem like a trivial thing, but it matters immensely for these teams’ prospects. The Cavaliers need their strength to matter, or they’re toast. The Warriors need to win two of five, with three on the road. With their regular season home/away splits — that is, against an average regular season opponent — their chances of doing this would be somewhere along the lines of 99.7 percent, around a 1-in-300 proposition for the Cavs. The model, on the other hand — which gives the Cavaliers’ strength full credit — assigns the Cavaliers a relatively high 11 percent chance of winning four of five. Either way, the Cavs’ chances don’t look great, but that’s a 30-fold difference.