The Celtics Changed The Way Robert Williams Was Defending. Now, They’ll Have To Defend Without Him.


Before Monday night, no team in the NBA was hotter than the Boston Celtics. Boston had won six games in a row, as well as 11 of its previous 12 and 24 of its previous 28, outscoring opponents by 15.8 points per game along the way as it charged into a tie for first place in the Eastern Conference.

Boston’s overtime loss to the Toronto Raptors on Monday night may not have been the harbinger of problems to come for the team as the playoffs approach, but it did provide a window into at least one issue the team will face the rest of the way. Boston was playing without four of its five starters, as Jayson Tatum and Jaylen Brown rested sore knees on the second night of a back to back and Al Horford missed his second consecutive game for personal reasons. The Celtics should have those three starters in the lineup once they get to the postseason, but they may not have Robert Williams III, who on Monday sat out the first game of what will be an extended absence after he suffered a meniscus tear in his left knee during a Sunday afternoon victory against the Minnesota Timberwolves.

Boston’s spectacular run was fueled by its league-best defense, and specifically by a tactical change from coach Ime Udoka involving how he used the Time Lord within his scheme. For the team’s Jan. 15 game against the Bulls, Chicago was without Zach LaVine, Lonzo Ball, Patrick Williams and Alex Caruso because of health and safety protocols and various injuries, and Bulls coach Billy Donovan started Alfonzo McKinnie in the frontcourt alongside DeMar DeRozan and Nikola Vučević. Udoka responded by flipping the responsibilities of Williams and Horford. Rather than having Williams defend Vučević, Udoka gave Horford that assignment and had Williams defend McKinnie.

As Udoka explained a few weeks later: “We changed some things defensively, whether we’re having [Williams] off-ball more, not as much on the ball and so he can be a help-side defender instead of switching as much and keeping him closer to the basket. We were taking him away from the basket early in the year, and that was obviously taking away a rim protector and his comfort zone.”

Before that game against the Bulls, Williams had spent the majority of his time defending opposing big men, which often meant venturing out to the perimeter when his mark went to set a ball screen. Because of how frequently Boston switches pick and rolls, that meant Williams spent far too much time away from the basket. Since that game, Williams has largely defended whichever of the opponent’s two forwards is the weaker shooter, allowing him to roam more freely along the back line and provide help near the basket.

Using Krishna Narsu’s Defensive Position Estimate (DPE) concept, we can see the dramatic effect this tactical tweak really had. DPE is a positional average of a player’s halfcourt matchups (from Second Spectrum data) in which each opposing player is assigned his standard positional value,47 and those values are weighted by the number of possessions on which the player defended those opponents. (So, a player who defends centers (5) for 50 possessions and power forwards (4) for 50 possessions would have a DPE of 4.50, for example.)

We calculated Williams’s DPE for every game this season, as well as his seasonlong and five-game rolling averages, with matchup data from Second Spectrum. Prior to that January game against the Bulls, Williams had a DPE of 4.02; in the games since then, it’s 3.18. In the 34 games he played before Udoka’s change in strategy, Williams recorded just four games with a DPE south of 3.50; in the 27 games he played after the change, he had only eight games with a DPE above 3.50.

Allowing Williams to roam free as a help defender paid enormous dividends, given that he is one of the NBA’s best rim protectors. He has allowed just 50.7 percent of shots to be converted when he was within 5 feet of both the shooter and the rim, per NBA Advanced Stats — the sixth-best mark among 108 players who have played in at least 20 games and challenged at least three such shots per game.48

With Williams sidelined indefinitely, the Celtics will have to figure out a whole new way to protect the paint. None of the team’s other frontcourt options are nearly as strong defending the rim: Horford has allowed a 55.7 percent conversion rate on shots near the basket; Daniel Theis has allowed 60.5 percent; and Grant Williams has yielded 65.7 percent. (One interesting twist might be using Brown in the backline helper role. He doesn’t have great rim-protection numbers this season, but he’s the only Celtics regular with the same type of explosive athleticism as the Time Lord.)

Boston’s defense against opponent drives has actually fared similarly well with Williams off the floor (1.07 points per possession including a drive, according to Second Spectrum) as with him in the game (1.06), but his presence served as a deterrent, with opponents venturing into the lane about five fewer times per 100 possessions. Without Williams behind them, players like Tatum, Brown, Marcus Smart, Derrick White and Payton Pritchard will have to be even stronger at the point of attack to prevent ball-handlers from getting downhill, where they can either do more damage near the basket or force more (and more-panicked) help, allowing them to spray the ball outside to open shooters.

Williams’s absence will also have a knock-on effect on both Horford and Grant Williams.

Horford began the season playing quite well but saw his effectiveness wane during the long stretch between late December and the All-Star break. He has seemed refreshed after getting a week-plus to rest his body; but with the regular season heading into the home stretch, the Celtics may have to count on him for even more minutes. Horford has exceeded 32 minutes of burn only 19 times in his 64 games this season, and at age 35, he is likely more susceptible to burnout than the 24-year-old Williams. How Horford holds up under increased strain is suddenly one of the most important questions for Boston’s season, when a few days ago, it wasn’t clear how often he’d be part of Udoka’s closing lineups.

Similarly, the Celtics will now need major contributions from Grant Williams on a nightly basis. His third NBA season has been his best to date as he has dramatically improved his 3-point accuracy, and he provides tremendous versatility on defense. Among 253 players who have played 1,000 or more minutes this season, Grant Williams ranks seventh in Bball-Index’s Positional Versatility metric. Playoff opponents will surely test that improved outside shooting, though, daring him (along with Smart, White and Horford) to beat them with jumpers while devoting extra attention to Tatum and Brown. He’s also not quite as strong a passer as Robert Williams, which could throw a small wrench in Boston’s much-improved offensive machine.

Without the vertical spacing Robert Williams provides as a pick-and-roll threat, the Celtics’ supplementary perimeter players will be under more pressure to both knock down their threes — which could be more highly contested, as opponents may not be as tempted to help so deep into the paint before recovering to shooters — and create off the bounce when Tatum and Brown draw defensive attention away from them. Smart has once again settled in as a low-30s 3-point shooter on decent volume, while White has connected on only 25.8 percent of his treys since he was acquired at the trade deadline. White shot 34.4 percent from deep during his time in San Antonio, so he’s a better shooter than what he’s shown so far in Boston, but he’s not necessarily a plus in that department.

It’s worth noting that Smart has shown improvement as a playmaker this season, while being counted on to play as a point guard more often than ever before. He has played 100 percent of his minutes at the point this year, according to Basketball-Reference.com — the first time since his rookie season that he’s played primarily as a point guard.

Smart has done well with the added responsibility (and the leadership burden that comes with it),49 as he’s on pace for career-best marks in assists per game, per 36 minutes and per 100 possessions, as well as assist rate. He has five double-digit assist games this season after recording only 11 such games in his career before this year. He seems to be making high-level reads, and manipulating the defense as opposed to reacting to it, more often than ever before.

Horford is also quite a capable playmaker, particularly when picking out cutters from the high post area but also when working as a short-roller out of ball-screen action. He’s a much different type of pick-and-roll threat than Robert Williams, though, as Williams rolls hard to the basket far more often. Theis provides a bit more of a dive threat but in a far less-athletic package, which makes the action much less threatening to opposing defenses. (Theis, though, is arguably the league’s best executor of the “Gortat screen,” wherein he seals off the help defender on a pick and roll while he’s rolling to the rim.)

After having surgery this week, Williams is expected to miss the rest of the regular season. But as ESPN's Adrian Wojnarowski reported Wednesday morning, the procedure Williams underwent — likely a meniscus removal (which carries a shorter recovery time), based on a four- to six-week timeline for return — makes it possible he could get back on the floor if Boston makes an extended postseason run. Even if he does, though, he’s unlikely to immediately be the same force he had been throughout this season. Williams emerged this year as an All-Defense-worthy player, one who was firmly in the mix for Defensive Player of the Year. How the Celtics account for his absence, individually and collectively, will perhaps be the most important determinant of their playoff fate.

Check out our latest NBA predictions.