Last night the Boston Celtics beat the Denver Nuggets in triple overtime, 118-114, stretching their win streak to seven games since Rajon Rondo went down with a season-ending knee injury. At the time, removing the Celtics’ All-Star point guard seemed like a fatal blow to a team that was already struggling to keep their heads above water. Despite a slew of off-season additions, efficient offense has been just as elusive as ever in Boston. The Celtics have averaged just 100.4 points per 100 possessions, tying Memphis for 22nd in the league, and without their offensive engine they appeared ready for a major stall.
But over this seven game win streak the Celtics’ offensive efficiency has jumped over three points to 103.7 points per 100 possessions. The defense in Boston hasn’t missed a beat, so if the Celtics have somehow stumbled onto a sustainable and repeatable recipe for efficient offense they quickly become a major player in the Eastern Conference playoff picture. If this is just a mirage, an extended adrenalin boost born of desperation, then we can expect the Celtics to fade back into the pack. So the question becomes, what is responsible for this bump in offensive efficiency, and is it something that can be sustained?
The first piece of the puzzle is that Boston lucked into a lineup of less than impressive defensive teams. Here are the total defense rankings of the seven teams they’ve beaten on their current streak – 12th, 29th, 25th, 6th, 17th, 23rd, 13th. Other than the Clippers there is not a potent defense to be found on that list. However, it’s possible that in that soft schedule the Celtics have been able to find some things that work, and that might work against a more stout opponent. One place to begin the search is looking at how their shot-selection has changed, using Expected Points Per Shot (XPPS). This metric uses the expected values of shots from different locations to evaluate the quality of a team or player’s shot selection.
For the sake of creating a full picture I divided these XPPS numbers into three categories. I looked at the Celtics’ XPPS from their recent seven game win streak. I also looked at their team numbers from before the Rondo injury, when he was on the floor and when he was off the floor. The values from when Rondo was on the floor have his own shot attempts removed, so that we are only looking at the quality of his teammates’ shots. I also looked at Actual Points Per Shot for each category and calculated the Shot-Making Difference between the two. For context, the league average for XPPS and Actual Points Per Shot is 1.047.
|XPPS||Actual Points Per Shot||Shot-Making Differential|
|With Rondo (Pre-Injury)||1.023||1.070||+0.047|
|Without Rondo (Pre-Injury)||1.031||1.067||+0.036|
|Without Rondo (Post-Injury)||1.049||1.143||+0.094|
When Rondo is on the floor the Celtics have had one of the least efficient shot-selections in the league. A vast majority of the shots Rondo creates for his teammates are mid-range jumpshots for players like Brandon Bass, Jason Terry, Jeff Green, Kevin Garnett and Paul Pierce. Those players are all reasonably solid shooters from that distance and despite the inefficient appearance of their shot-selection, they have actually managed fairly well, scoring 1.070 points per shot. When Rondo was off the floor, before his injury, things looked nearly identical.
However, since Rondo’s injury the Celtics’ have made some clear changes in their shot selection. Their XPPS has pushed above average for the first stretch this season and they’ve also been outperforming their XPPS by nearly twice the margin they were earlier in the season. To see exactly what’s causing this numeric shift we can look at what percentage of the Celtics’ shot attempts are coming from the six different areas that make up XPPS.
Celtics Shot Selection
|Restricted Area||In the Paint (Non-RA)||Mid-Range||Corner 3||Above The Break 3||Free Throws|
|With Rondo (Pre-Injury)||25.8%||10.8%||33.0%||7.3%||12.6%||10.3%|
|Without Rondo (Pre-Injury)||29.4%||12.3%||29.0%||6.0%||11.4%||11.1%|
|Without Rondo (Post Injury)||31.3%||10.7%||27.4%||6.7%||11.9%||11.7%|
There are two big shifts here. Comparing their shot selection with Rondo to their distribution now, they’ve moved about 6% of their shots from the mid-range to the rim. They’ve also taken about 1.5% of their shots and moved them behind the three-point line to the free throw line. Those are significant changes, and ones that bode well for the Celtics’ offense over the rest of the season. There is no reason that those changes in shot distribution couldn’t be sustained over the rest of the season and on their own, separate from any questions of shooting accuracy, they make the Celtics’ offense better by a definitive, albeit small, margin.
The other piece of this shot selection equation is accuracy. As I pointed out above, since Rondo’s injury the Celtics have improved their shot selection by a healthy amount, going from way below average to slightly above average. But an even bigger factor in their improved offensive production has been the doubling of their Shot-Making Differential. Shot-Making Differential is the difference between the expected value and the actual value of their shots, so seeing that they’ve doubled their output in this category already takes into account the improvement in their shot selection. For reference, the Celtics’ Shot-Making Differential of 0.096 over the last seven games would rank second in the NBA this season, just behind the Oklahoma City Thunder. In short, this piece of their improvement is probably not sustainable, at least not at this level.
Summarizing what we’ve seen so far, using the most basic descriptors, the Celtics began taking better shots, but also began making those shots at a rate vastly superior to what they were making earlier in the season. It seems that a big piece of their improvement may be that they have stumbled into a streak of particularly accurate shooting. For us to assume that increased accuracy is in any way sustainable we’d have to assume that there have been some other changes to their offensive attack besides just the locations where their shots are coming from. One place to look is at the types of offensive possessions they have been using to create those shots.
The table below contains data from mySynergySports and shows the percent of offensive possessions the Celtics have used in different ways, before and after Rondo’s injury. It also shows the points per possession the Celtics have averaged in each of those possession types.
Celtics Possession Types
|Possession Type||% of Possessions (Pre-injury)||% of Possessions (Post-Injury)||PPP (Pre-Injury)||PPP (Post-Injury)|
|PnR Ball Handler||11.5%||14.0%||0.80||0.86|
Right away we see a few more significant shifts in the Celtics’ offensive makeup. With Pierce taking on a bigger role as offensive facilitator, a much smaller portion of their offensive possessions are being used on post-ups. Surprisingly, those post-ups have become a much more efficient option, jumping from 0.90 points per possession to 1.04. A slightly higher percentage of possessions have been moved to cutters, which have also seen an increase in efficiency. We see the same pattern with spot-up shots. But the biggest piece is what has happened with the Celtics’ pick-and-roll.
Before Rondo was injured, pick-and-roll possessions used by the ball handler made up 11.0% of the Celtics’ offense and they averaged 0.80 points per possession in those situations. With Rondo out that percentage has jumped to 14.5% and the Celtics are averaging 0.86 points per possession. Those sound like small differences, but they have huge import for the Celtics’ offense. When Rondo ran the pick-and-roll his first, second and third priorities were to find an open shooter. This single-minded focus made the job of a defense much more simple. The fact that the Celtics were still able to succeed to any degree with such a reluctant scorer handling the ball is a testament to his prowess at moving the defense and creating space. Now with players like Pierce, Terry, Barbosa and Bradley running more pick-and-rolls the defense is presented with several more dimensions to account for. The efficiency of possessions used by pick-and-roll screeners for Boston has declined slightly without Rondo at the helm, but the threat of a scoring focused ball handler has more than offset that. In the end, although the Celtics have no one who even approaches Rondo’s ability to create shots for others in the pick-and-roll, using players who are better at creating shots for themselves may be making the Celtics more difficult to defend.
It’s important to remember that we’re looking at a seven game sample, and things can rapidly change in several different directions. However, the Celtics have made some nice fundamental changes to help them plug the numerous holes created by Rondo’s absence. They are taking better shots, and creating them in some new ways. Although the incredible accuracy they’ve displayed during their win streak may not be here to stay, it does seem that the changes they’ve made have created a more full and healthy offense.