Where I disagree with you is that the "second unit" doesn't actually exist like this. No one makes 5 in/5 out substitutions. The second unit isn't actually a thing in real games the way we list it out on paper. In actual rotations it's always going to be a mix of starters and bench players. So even without having Hayward off the bench you're still going to accomplish your goal of having one of Hayward/Tatum/Kemba running alongside the bench (left Brown off because I'm not sure he's that type of player even on 2nd unit to really run an offense).
Since it does't really matter in the "second unit" I say start Hayward to maximize the shared time on the court of all our best players.
this is all assuming health of course.
I guess i wasn't really thinking about it in terms of an all or nothing (i.e. 5on5 substitutions) approach. Generally, the more minutes Gordo plays in the starting lineup, the less minutes he's playing in other lineups with less offensive firepower. I guess if he starts and is quickly subbed out it's the same thing.
ConstableGeneva wrote:Kemba/Brown/Tatum/Theis with Gordo: 116.6 ORtg (92nd percentile), 104.7 DRtg (93rd percentile), +11.9 NRtg (97th percentile)
Kemba/Brown/Tatum/Theis with Smart: 105.0 ORtg (13th percentile), 100.0 DRtg (99th percentile), +5.0 NRtg (80th percentile)
That's too much of a dropoff in offense to justify the improvement in defense.
And to hugepatsfan's point, we don't use hockey 5-for-5 substitutions. Brad usually pulls out Hayward or Tatum early to bring them back before the 1Q/3Q ends so they can lead the "second unit." So not only do you maximize your starting group this way by getting them more minutes together, you also have someone to make your bench group's offense functional/more potent.
It's the same counter-argument to fans who have suggested that Kemba should come off the bench in favor of Smart.
EDIT: Smaller sample but the current starting lineup w/ Smart is scoring about the same (105.1 pts per 100 possessions) in the playoffs.
This is only one side of the equation. The offense lost by not having Gordo in the starting lineup doesn't just disappear, in theory it would be re-allocated to a different lineup. Is there a way to compare this data set against a different "bench" lineup to see the differences? For example, what's the data on Wanamaker/Brown/Kanter/GrantW/Smart vs. Wanamaker/Brown/Kanter/GrantW/Hayward? (it doesn't have to be that exact lineup, it can really be any common lineup that isn't the starting lineup)
Gordo w/o Kemba/Brown/Tatum: 110.2 ORtg (43rd percentile), 91.4 DRtg (100th percentile), +18.8 NRtg (100th percentile)
Smart w/o Kemba/Brown/Tatum: 109.2 ORtg (35th percentile), 95.5 DRtg (100th percentile), +13.7 NRtg (99th percentile)
[Remember that these are small sample sizes. Like if you put Tatum w/ the bench group (more than twice as many possessions as the other two combos above), the numbers don't look as good.]
Anyway, putting Gordo into the second unit didn't do much for our offensive numbers compared to Smart-led second unit. I think the issue there is the guys surrounding Gordo/Smart. You need to have better shooters/scorers and more skilled/experienced players than Semi/Grant/Wanamaker. It's still a team game. Gordo isn't a Lou Williams type who will just get you buckets on his own regardless of whoever he's playing with. He's better served and more effective when there's ball/player movement. He's a system player like Horford but a better scorer.
Ultimately for me, it's more important to win the minutes where you're up against the opponent's best/starting lineups. So why sacrifice the effectiveness of your best/starting group just to win more possessions with the bench group? By the way, our bench had the 6th best net rating in the league. We just did it with our defense (3rd best DRtg). We were no longer losing leads the way we did in past seasons when our bench comes in. Considering all the injuries to our key guys, it's not as bad as people made it out to be.