So, we’re over the magical 20 game mark where stats start to mean something, and the Raptors are, for the most part, rolling. League-leading 19-4 record, second best net rating in the league, and even with their schedule being a bit easy, they are currently riding a seven-game win streak that includes wins over the Grizzlies and (admittedly shorthanded) Warriors.
We spent a bunch of time on what was working so well last time out, so this time we’ll look at one of the more concerning trends instead.
The Raptors just cannot maintain big leads to save their lives. In last night’s Warriors game, the game had to go to overtime after Golden State came back from being down by as much as 18, including being down by 13 late in the 3rd Q and down by eight entering the 4th Q. In two of the Raptors’ four losses, the Raptors held leads of 19 and 13 points. Their two point win over Orlando — the lead was 18 at one point. All this in the past ten games.
Now, it’s unfair to say they can’t maintain leads. They maintain some of their leads. But there is a worrying trend that a big lead is not safe for the Raptors. So let’s look at why that is.
First off, to blow a big lead, you need to have a big lead. So, what’s working? Well, that’s easy.
Look at this list of net ratings (point differential per 100 possessions) for the big minutes players on the team.
Danny Green +17.9 RTG (1)
Pascal Siakam +14.9 RTG (4)
Kyle Lowry +13.5 RTG (11)
Serge Ibaka +13.3 RTG (12)
Kawhi Leonard +10.2 RTG (21)
Jonas Valanciunas +6.0 RTG (54)
Those numbers in brackets, by the way, are each players’ rank league-wide in net rating (among the 279 players with 15+ MPG, 10+ GP). That’s... impressive.
So, the starting lineup is working. Both versions, as we noted last time out.
Lineup | MP | ORTG | DRTG | RTG
Lowry-Green-Leonard-Siakam-Ibaka: 234 MP, 117.7, 101.6, +16.1
Lowry-Green-Leonard-Siakam-Valanciunas: 109 MP, 122.9, 98.3, +24.6
Lowry-Leonard-Ibaka: 346 MP, 119.6, 105.9, +13.7
Lowry-Leonard-Valanciunas: 166 MP, 116.1, 100.8, +15.3
Yep. Can’t go wrong.
So, we know where the big leads are coming from. That leaves an obvious suspect to pin the blown leads on.
First off, we don’t really have an established bench unit. See the lineup minutes above for the starters? Even the second version of the starters has played over 100 minutes together. With five-man units, a good rule of thumb is that 50 minutes played is the sample you want to be even a little confident in the results.
The most used lineups where at least one of Lowry or Leonard are resting have about 30 minutes on the season. The most used lineup has appeared in only ten of the team’s 23 games.
So, right away, that’s bad. There’s no way for the bench unit to develop any sort of chemistry if they don’t play together. For reference, the bench unit had reached 60 minutes played together at this time last year, and that was the version with OG Anunoby there before he was inserted into the starting lineup two weeks into November.
So, to evaluate the various bench looks we’ve seen, we’ll use two-man to four-man lineup combinations instead of full five-man units. First, let’s just take a quick look at how individual players have fared away from both Lowry and Leonard.
Player without KL-KL | MP | RTG
VanVleet: 197 MP, +1.4
Wright: 183 MP, -6.1
Miles: 168 MP, -12.8
Anunoby: 155 MP, -0.3
Valanciunas: 135 MP, +0.3
Ibaka: 91 MP, +1.9
Siakam: 63 MP, +10.8
Green: 41 MP, +16.8
I’ve left off guys like Brown and Powell who have played minutes with the bench, but probably shouldn’t. Powell might yet take Miles’ minutes (he’s played to a -3.6 RTG compared to Miles’ awful -12.8), but with him still hurt, we’ll focus on what’s been under the team’s control of late.
They’ve started dabbling of late with getting Siakam reps with the bench, and that’s gone very well, as you can see above. So let’s take a look at what pairings he’s having success with (again, away from Lowry and Leonard). Green has had even more success, but with him being a bit of a security blanket for Kawhi in that starting lineup, I suspect they’ll want to leave it like that for now.
Pairing without KL-KL | MP | RTG
VanVleet-Siakam: 55 MP, +13.3
Wright-Siakam: 43 MP, +11.0
Miles-Siakam: 19 MP, +0.6
Anunoby-Siakam: 26 MP, +25.4
Valanciunas-Siakam: 33 MP, +15.8
Ibaka-Siakam: 28 MP, +6.1
Green-Siakam: 10 MP, +6.3
Look how much better that looks! Small samples, but throw in how successful the bench was last year and it makes sense to re-connect the Delon-FVV-Siakam set. In 38 minutes together (still a small sample), the three have posted a +13.8 net rating away from the two stars.
It even looks like Siakam can make the bench work with Valanciunas there. Anunoby and Valanciunas are the more common sense fit — Valanciunas needs his wings/forwards to fight over screens, one area of the game Siakam is not quite up to speed on yet, and shooting from his forwards (which Anunoby in theory provides more of than Siakam, early season shooting numbers aside). While Siakam needs space to wreak havoc in the paint, which Ibaka provides better than Valanciunas. But he’s so dynamic this year that it doesn’t seem to matter.
Looking at the performance of the frontcourt pairs with the two stars, you get this:
Pairing with KL-KL | MP | RTG
Ibaka-Siakam: 272 MP, +13.6
Valanciunas-Siakam: 131 MP, +15.9
Ibaka-Anunoby: 45 MP, +1.9
Valanciunas-Anunoby: 31 MP, +22.2
So, everything works except OG and Serge, which is not shocking — Serge needs the rebounding help at PF, and having two floor spacers and no screener or extra creator makes for an awkward offensive look.
Long story short, if Siakam is playing minutes with the bench, you’d expect OG to play some minutes with the starters, and that so far seems to only make sense beside Valanciunas.
One solution could be to keep the pairings relatively tied together — so start off the logic with Siakam playing with the bench. And working backwards, that means that Serge should play with the bench too so he’s not getting minutes beside OG. And therefore JV and OG should be playing minutes together with the starters.
Now this has nothing to do with who starts on any given night. If JV starts, start OG with him. You can still give Pascal and Serge a few minutes with the starters, and if need be close with them, leaving the minutes split probably closer to 24 minutes for each pair. And on the more common nights (based on the season thus far) where you start Serge and Pascal, you sub JV and OG in early (to play, say, five minutes in the middle of each quarter with the stars), and bring the smaller pair in with the bench.
That’s just one way to do it, but one big takeaway here: the bench needs Pascal. They need to find a way to play him there.
One More Concern
There are also some other stylistic concerns with the bench unit that many have noticed. The most glaring one has been the relative usage of Wright and VanVleet. VanVleet has been handling the ball a lot this season. His usage is way off what it was last year, and it seems to be messing with the success of that bench unit, which was pretty tightly tied to VanVleet’s value as a high IQ off-ball mover and floor spreader in a catch and shoot role. Keeping him off ball also lets him dedicate more energy to the defensive end of the floor, where it’s really his energy and tenaciousness that helps him drive results on that end, unlike other players who can rely on length defensively.
VanVleet is taking 32% of his shots as pull-up jumpers outside of ten feet. That’s up from 20% last year. At the same time, his catch and shoot jumpers have plummeted from 38% of his attempts last year down to 32% this year, and his shots at the rim are down as well (36% vs 42%). This is a completely different role for him from last season, and it’s not going well. His effective FG% on pull-up jumpers is about the same (43%, down a bit from 46%). His catch and shoot efficiency is still very good (60% eFG%). He’s seen a drop in efficiency at the rim, because his only attempts there now are when he drives in there himself against a set defence.
In the meantime, the team has almost cut Delon Wright’s playmaking opportunities in half. His assist rate is down from 20% last season to 12% this season, a dramatic drop off, and one that VanVleet’s marginal increase in assist rate from 23% to 24% doesn’t nearly make up for. Wright’s usage rate is also down, so he’s not finishing plays as much either.
This is one other way in which that bench unit is being managed very differently from last year, when we saw a bunch of what were essentially rookies in terms of playing experience, with one vet, become a dominant unit early in the season and hold up as a slight net positive even as scouting reports started to get filled in.
The Raptors are a very good team with a lot of margin for error. They are currently using up a lot of that margin (though obviously not all of it on most nights). Which is fine, for now. The team might find another way to make the bench units viable. But there is a relatively easy path to a strong bench unit that is largely just finding a way to recreate as much of last year’s bench unit as possible.
But what do you think? This is a stance I’ve held on this bench look since the start of the season, so maybe I’m seeing what I want to see (not that I really want any facet of this team to struggle). Is there another fix for the bench that doesn’t require overloading the star players? Or maybe there’s no problem here at all, and we should just wait it out?
All stats per NBA.com.