The Blazers are a tough team to project forward because so many things were unexpected last year. Multiple players shot career best percentages and the team played very differently during different time periods. Without concrete reasons to explain improvements, their success could be random. If that’s true then Portland should be bracing for a regression (as Kevin Pelton argued the other day).
However, many of those concrete explanations exist. One of the more common ones is the late-season change to the starting lineup. Maurice Harkless saw his minute nearly double, Noah Vonleh’s minutes were cut by a third, and Meyers Leonard’s playing time disappeared completely because of his injury four games prior. The argument goes that these new lineups played exceptionally well and, assuming coach Terry Stotts plays these lineups more often, they’ll drive Portland’s improvement next year.
There’s many facets to this argument but let’s focus on just the defense for now. After the starting lineup changed, the defense improved from bad to almost average. They had been giving up 108.7 points per 100 possessions and improved to 107.4 for the final 11 games, according to basketball-reference.com stats. Can these lineup changes explain the improvements in the Blazers’ defense and is there potential for more?
Vonleh’s statistics are surprisingly neutral on the defensive end. Portland gave up 0.2 fewer points per 100 possessions with Vonleh on the floor. However, this speaks more to the nature of lineup data and on/off statistics than to his influence on the court. These numbers are noisy. They don’t take into account who the player is playing with or against. As a result, we have to take all the data with a grain of salt and combine it with our observations. No conclusions derived solely from lineup data should be taken as gospel.
Even real plus minus, a stat that takes into account who is on the court at the same time, is role specific. It can change substantially from year to year and only comments on what a player’s impact is in a given role. Vonleh rated about neutral in this statistic as well so perhaps we should give Vonleh a little credit. When surrounded by starters he won’t single-handedly undermine a team’s defense. That’s about as far as I’ll go. Vonleh looked so lost last year that it seems excessive to devote two full paragraphs arguing that less Vonleh equals better defense.
Leonard is a bit trickier. His on/off numbers show him as a small negative on defense. His defensive real plus minus is more incriminating. His presence give up an extra 1.26 points per 100 possessions, according to ESPN.com stats. Although that may not sound like much, that number ranked Leonard as the 82nd worst defensive power forward out of 92. That’s obviously not good but his overall numbers don’t tell the entire story.
Leonard performed much differently depending on which frontcourt player he was paired with. The combination Meyers and Mason Plumlee was a disaster, giving up 115.9 points per 100 possessions, per NBAwowy.com. That’s substantially worse than the league-worst Los Angeles Lakers performed as a team last year. However, two-player lineup data is exceptionally noisy as it averages out a lot variance. For example, the Davis-Lillard duo had a defensive rating of 107.8 but that two-man combination was a part of both of the following lineups:
Lillard/McCollum/Henderson/Aminu/Davis - Allowed 88.7 points per 100 possessions in 24.7 minutes
Lillard/McCollum/Crabbe/Davis/Leonard - Allowed 129.6 points per 100 possessions in 36.1 minutes
Is the Davis-Lillard combination a good defensive combination or a bad defensive combination? Obviously, it depends on who they’re playing with but the drastic differences make it difficult to say anything about the pair in isolation. That’s not as much of a problem when evaluating Leonard and Plumlee though.
Their lineup data is much more clustered. Their overall defensive rating was 115.9 and the best lineup they were a part of (minimum 10 minutes played) gave up 113 points per 100 possessions in 12 minutes played. That still would have been worse than the Lakers. Every single lineup with both Leonard and Plumlee got absolutely torched. Every single one. Lineup data is noisy but that’s an incredibly strong signal indicating that the pair was a disaster.
This seems to match the eye test as well. Leonard’s biggest weakness is defending in space. He lacks lateral mobility and struggles to get himself positioned in the right place. Keep Leonard near the rim and he does alright; Ask him to chase power forwards around and it starts to get ugly.
That’s part of why the Leonard-Davis combination was so much more effective. They had a defensive rating of 105.7 in 817 minutes together and the typical noise in different lineups. The stingiest lineup they were a part of gave up less than 80 points per 100 possessions, while the worst allowed more than 125. It’s hard to draw strong conclusions when the data is that varied but it’s clear that the Davis-Leonard frontcourt has the potential to be a part of a good defense.
This, again, matches the eye test. Davis was perhaps Portland’s best defender at the power forward position and he pushed Leonard back to his comfort zone at center. Davis’ defensive real plus minus is also the best on the team at 2.62. It’s not that Leonard can or can’t play defense in a vacuum — it’s with whom he plays defense and what he’s asked to do on that end.
The implications for Portland’s rotation are clear: (1) don’t play Vonleh, (2) don’t play Leonard at power forward, and (3) if you’re going to play Leonard, try and pair him with Davis.
Last year, the Blazers had 42 lineups that played more than 10 minutes and didn’t include Vonleh or the Meyers-Mason pairing. Those lineups played 1964.4 minutes (about half the season) and gave up 106.5 points per 100 possessions, almost exactly the league average. When the Blazers avoided their bad lineups they had an average defense. That’s not over a cherry-picked month or a 10-game hot streak. That’s over half a season and it offers a pretty compelling reason for the Blazers’ improvement at the end of the year. It wasn’t random — whenever the Blazers played this way their defense improved. They just happened to play this way a lot more often at the end of the year.
The counter argument is that those 1964.4 minutes were systematically different than the other 1996.6. Vonleh and the Meyers-Mason combo started most games, so by removing those minutes perhaps you’re cutting out the moments when the Blazers played the opponent's best offensive lineups. There’s some truth to that but Stotts closed games with a variety of players meaning a portion of these lineups played against the other team’s best. Plus, the Blazers don’t have to stop at avoiding bad lineups. The new pieces they added this summer allow them to create new lineups as well.
Turner might not be a substantial defensive upgrade over Henderson in isolation but he’s bigger and rebounds better. The trio of Turner, Harkless, and Aminu should allow Portland to keep two big, rebounding wings on the floor at all times. That’s important because Portland performed much better when they went small with both Aminu and Harkless (DRTG of 105.6) rather than just Harkless (113.5) or just Aminu (108.8). Turner won’t be interchangeable with Harkless and Aminu on the defensive end but he’s a much better facsimile than anything they had last year. His addition should help Portland stay big enough even when they go small.
Keeping two big forwards on the floor at all times implies that the Lillard-CJ McCollum-Allen Crabbe combination would be played sparingly. That might not be a bad thing, at least from a defensive perspective. They did OK when paired with Aminu and Davis or Plumlee, but struggled otherwise. Given how much size that perimeter trio is giving up, it makes sense that you would need to load up on defensive talent at the other two positions. Festus Ezeli should help immensely in these situations and his presence could make this potent offensive lineup structurally sound on the defensive end. With all the depth the Blazer have now, Stotts should have no problem creating these types of synergistic combinations.
At the end of last season, Portland employed the use of addition by subtraction. By playing Vonleh and Leonard less they were able to improve their defense during the last 11 games. Vonleh won’t see the court much next year and the Blazers’ depth will allow Stotts to play Leonard only in specific situations. This, in combination with the additions of Turner and Ezeli, gives Portland a pretty clear and easy path to an average defense. A top-10 offense and an average defense is the resume of a playoff team but not a contender, supporting the current perception of the team. It’s not likely they take a step back but the way forward remains murky at best.