Death to all-bench units

With 1:35 to play in the third quarter of Tuesday’s game against the New York Knicks, head coach Dwane Casey called Norman Powell’s number.

The high-energy second-round pick with a shade over 10 minutes of NBA experience to that point was checking in as the Raptors held an 80-76 lead, opposite a Knicks lineup that contained no starters. The tight score and late juncture made it a somewhat high-leverage situation but the relative lack of competition quality made it a reasonable time to see what Powell could provide.

After all, Casey is going to need to see what Powell can do in meaningful minutes some time over the next two-plus weeks. Terrence Ross is out at least a fortnight after suffering a ligament injury in his left thumb, DeMarre Carroll continues to deal with plantar fasciitis in his right foot, and Powell, with his defensive acumen, appears the best bet of the team’s four de facto rookies to be called into action on the wing as a result.

Powell would close out the quarter and head to the bench for the remainder of the game, the Knicks having gone on a 9-5 (really, 9-4 since Cory Joseph was already at the line when Powell entered) to end the frame. That’s no real fault of Powell, who missed a corner three in his 95 seconds of action but otherwise did nothing to earn his -5.

Instead, the blame for the poor end to the quarter could be assigned to Casey, who put his rookie in no position to succeed. Powell played his minutes as part of an all-bench unit, so his quality of teammates matched the quality of competition (in relative terms – I’m borrowing terminology from hockey here that doesn’t marry perfectly). Powell was slotted at the two, Cory Joseph at the one, with a laughable Anthony Bennett-Patrick Patterson-Bismack Biyombo trio making up the frontcourt.

All coaches go to all-bench units at times. Casey seems a proponent of the “hockey change” in non-blowout situations, and the Raptors averaged 3.42 minutes per game with no starters on the floor last year, per data from the excellent Seth Partnow of Nylon Calculus. While that may seem like a lot or a little depending on your basketball worldview, it’s almost exactly the league average (3.41). Partnow also found that the start of the second quarter and the end of the third-start of the fourth are the times where teams average the fewest starters, and even in games classified as “close,” teams only averaged 1.6 starters in the final minute of the third.

But 1.6 is a fary cry from zero, and the Raptors may have punted an opportunity to grow or at least hold their lead. Lineups with zero starters are outscored by 3.07 points per-48 minutes on average, more than 1.5 points per-48 worse than lineups with even a single starter. To get really specific using Partnow’s data, when a home team has one starter and a visiting team has none, the home team outscores the visiting team by 2.7 points per-48 minutes. That effect is tiny over 95 seconds but it speaks to the general strategy of matching an all-bench unit with one that contains some of your better players.

The Raptors turned a potential edge – Derek Fisher using a full bench mob – into a neutral situation, one that may have even skewed New York’s way given how thin the Raptors were Tuesday. Again, it was a lineup with Joseph, a struggling Patterson who hasn’t hit an open triple in Bruno Caboclo’s lifetime, and three players who don’t offer a ton on the offensive end. It’s surprising that unit even scored. The quality of the second unit doesn’t fall on Casey, especially given the injuries and the unseasonable cold spell that’s befallen Patterson (and previously, Ross), but an over-reliance on Joseph making chicken salad out of chicken shit does.

In Casey’s defense, he was already facing a situation in which Kyle Lowry and DeMar DeRozan were being used heavily. They wound up playing 37 and 39 minutes, respectively, and are averaging 34.9 and 36.3. Keeping those players under 40 minutes is a worthwhile goal, even in a tight game, but Casey has to find a way to better stagger their minutes, as the team simply can’t afford to have them both off the floor in a game that’s remotely close right now.

When Lowry and DeRozan share the floor, the Raptors are outscoring teams by an estimated 7.5 points per-100 possessions (PPC). With only one of them on the floor, the Raptors are outscoring teams by an estimated 6.6 PPC. With neither, the Raptors are being outscored by 30.5 PPC, and their offense is only managing 70.4 PPC.

Again: When Lowry and DeRozan hit the bench together, the Raptors score at the rate that me, William Lou, and any three commenters would score at, and the net outcome is sub-Sixersian. Thirty-and-a-half points per-100 possessions is an insane rate, and the Raptors are averaging 4.9 minutes per game with their starting backcourt on the pine together. Considering they’ve only played one blowout to inflate those numbers – the Raptors only played two minutes with an all-bench unit against Miami – that’s simply too high.

This isn’t the reason the Raptors lost on Tuesday. There are a number of strange, small things that the game ended up hinging on, and the last 95 seconds of the third quarter was a potential marginal gain that the Raptors eschewed. That’s an important lesson for tight games in the future – Lowry and DeRozan can’t play 48 minutes, but their minutes need to be staggered such that one of them is always on the floor.