Thanks to the good people at Synergy Sports, we now have free access to their amazing database of play-by-play statistics. Launched as a service to NBA teams and scouts, Synergy logs every play that ends in a shot, shooting foul, or turnover. They then categorize each "play" by its type, such as post up, isolation, P&R ball (when the ball-handler ends the play), P&R man (when the screener ends the play), off screen, transition, etc. Synergy then assigns the outcome to both the offensive player and his primary defender, allowing some fascinating insight into how players and teams fare on different types of plays, both offensively and defensively.
To add an extra dimension, Synergy has teamed with the NBA to link its stats database to video clips of every play it categorizes. So if I'm interested in Andrew Bogut's post offense, I can not only look at his 09/10 stats--0.80 points per play, 126th in the league, 43.3% shooting--but also video of all 512 plays he had in the post this year. Pretty damn cool. Anyway, enough talk--let's look at the numbers and see what we can learn.
Offense: Room for improvement...lots of it
The points per play (PPP) metric used by Synergy is slightly different from the notion of "points per possession," which I refer to all the time as the best measure of a team's offensive efficiency. The main difference is that possession-based metrics embed the effect of offensive rebounds extending possessions, meaning that there can be more than one "play" per possession. That's why the Bucks' overall PPP of 0.91 in the Synergy data is considerably less than what the points/possession stats will tell you (1.05). Still, both reinforce what we all saw just by watching the games. The Bucks were just 24th in the league in overall PPP and 23rd in points/100 possessions, though their efficiency on both ends improved after the acquisition of John Salmons. Over the final 30 games, the Bucks scored 106.5 pts/100 and allowed just 100.9 pts/100, compared to 104.9 and 103.1 overall. We don't have Synergy data splits available to mirror that, but as we discuss the Bucks' season stats it's worth keeping that in mind.
Nearly a quarter of Bucks plays came on spot-up opportunities, where they scored 0.99 PPP (10th in the league) and made just shy of 40% of their shots. Over half of those attempts were threes (38%), so the low percentage is understandable. Luke Ridnour was the Bucks' best in this category with 1.21 PPP, good for 19th in the league, while John Salmons (1.12 PPP), Jerry Stackhouse (1.12) and Carlos Delfino (1.09, 38% of his plays) were also very good. Most of Ersan Ilyasova's plays also came on spot ups (35%) but he scored just 0.94 PPP.
Pick and rolls were the next largest source of plays (21.4%). Most P&Rs resulted in plays for the ball-handler (16.1%), but those types of plays were also the least effective in terms of generating points (0.78 PPP, 23rd). Not too surprising given how many of those plays were missed shots by Brandon Jennings, who got 43.1% of his plays off P&R and averaged just 0.76 PPP on 35.5% shooting. Luke Ridnour (0.91 PPP, 48.6% fg) and John Salmons (0.83 PPP) were notably better.
As you'd expect, P&R plays ending with shots by the screener (usually shots around the cup or open jumpers) were much more effective but also harder to get. The Bucks were 13th in the league with 1.04 PPP, but they accounted for only 5.3% of plays. To give you some context, the Phoenix Suns were the best in the league on P&R plays, ranking 1st in PPP when Steve Nash or another ball-handler shot (0.95) and 2nd when Amare Stoudemire or another screener finished the play (1.17). Unfortunately you can't get good shots from P&R all the time, and even the Suns only got 19.4% of their plays that way. For the Bucks, Bogut was exceptional on P&R plays, scoring 1.36 PPP (10% of his plays, 6th in the league) while Ilyasova was also good (1.1 PPP).
The Bucks struggled the most in the post, averaging just 0.79 PPP (27th) on the 8.6% of plays that came from the block. Bogut was responsible for the majority of those plays (512 of 831 total plays), and it's somewhat interesting to note that he was barely better than the Bucks' team average in PPP terms. On the other end of the spectrum, I'm not the least bit surprised to report that Ilyasova's numbers are absolutely horrendous on the block (0.56 PPP, 26.8% shooting). Isolation plays accounted for a slightly higher percentage of plays (10.7%) and a higher PPP as well (0.84, 18th). I assumed the Bucks were less iso-heavy than other teams, and compared to a team like the Hawks (17.3% of plays, 0.88 PPP) that's certainly true.
The only other major category of plays came from transition (10.4%), which obviously tend to be easier shots (1.09 PPP). But the Bucks' lack of athletes and focus on defensive rebounding limited transition opportunities overall (4th worst in the league with around 10.4 ppg) and they also weren't particularly effective when they did get out on the break (24th in PPP). To provide a bit of context, the Warriors were far and away the league leaders in fast break points per game (23.7) and nearly 18% of their plays came in transition. However, that emphasis on creating quick shots meant they were more about quantity than quality (1.13 PPP, 15th overall).
Defense: Post and Transition Need Work
Defense was the Bucks' meal ticket all season long (2nd in defensive efficiency), so it's not surprising to see the Bucks ranking seventh overall in points per play (0.88) and in the top ten in most categories. The Bucks excelled defending isos (0.80 PPP, 12.2% of plays, 3rd overall), which you can interpret either as a) the Bucks are better individual defenders than they get credit for or b) opposing players kept shooting even when the Bucks helped on defense. Or both. The Bucks' defensive rotations and consistent effort closing out on shooters also helped them notch the 5th best defense on spot-up plays (0.94), which is a big deal since those account for almost 20% of plays (the most of any type).
The Bucks were also very good defending the P&R, which says a lot about both the guards' dogged work fighting through screens as well as the frontcourt players' ability to show, hedge and recover. While some teams change up their strategies against P&R depending on the opposition, the Bucks were fairly consistent in that they help but never switch (there's more to it than just that, but that's probably the most basic aspect). Plays ending with the ball-handler either shooting, turning it over or getting fouled netted just 0.81 PPP (6th overall, 14.3% of plays). About 75% of P&R plays resulted in plays for the ball-handler, which is good since plays ending with the screener getting a shot had a much higher expected value (0.94 PPP, 4.9%). Even so, the Bucks' PPP in those situations was second-best in the league, which goes to show that when the screener gets a shot it's usually a high-percentage play. While Luc Mbah a Moute's overall numbers don't stand out too much (0.87 PPP), his P&R defense was exceptional with 0.71 PPP allowed when he defends the ball-handler (28th) and 0.63 PPP when he defends the screener (6th).
From a relative PPP standpoint, the Bucks were much weaker in the post and transition. Overall, they were 6th best in the league with 12.7 transition points per game allowed, so they were fine in terms of the net damage inflicted by opponents. Still, that masks the fact that opponents were extremely efficient when they did get out on the break, scoring 1.18 PPP (10.9% of plays, 23rd in the league).
The Bucks were equally bad defending the post, where opponents scored 0.91 PPP and shot 46.3%, just 23rd in the NBA (9.8% of plays). Andrew Bogut was the primary post defender with a little over 20% of those plays and was marginally better than the Bucks' average, allowing 0.89 PPP (just 158th in the league). Kurt Thomas was assigned about half as many "plays" but put up a rather incredible 0.59 PPP, good for 5th best in the league. Dan Gadzuric was the worst of the Bucks' three centers with 0.94 PPP, offsetting a good opponent fg% (.381) with an extremely high foul rate--over a quarter of post possessions against him ended in a shooting foul, compared to just 8% for Thomas and 10% for Bogut. The mediocre numbers for Bogut might be a bit surprising, but it's probably a bit unfair to compare his defensive outcomes with the backups given a) the level of competition he faced and b) Bogut frequently guarded the opponent's best post player, even if he wasn't a center. Still, it also suggests that Bogut still has room for improvement as a one-on-one post defender.
There are obviously limits to using play-by-play stats in a team sport, perhaps even more so when trying to assign individual responsibility on a gang-tackling defensive squad like the Bucks. And as important as the PPP rankings are, the Synergy data as its currently offered doesn't make it easy to look at distributional aspects of the play categories. For instance, even the worst team at scoring in transition will have a higher PPP on the break than in isolation plays. But fast break chances are harder to get than isos, and it's tough to tell just from the numbers how teams should be better balancing their play-calling. A big part of being efficient on both ends is forcing the opponent into lower percentage types of plays and getting more of the high-efficiency plays (transition, cuts) for yourself, but it's much easier said than done.
Synergy also doesn't make it possible at present to see all teams' stats on one page, and I haven't gotten around to copying down each team's stats into a spreadsheet for an offline comparison. Same goes for players, who can only be compared one to one. Hopefully they'll eventually allow users to see all players' ranked by category, so we can more easily see who is the best and worst at defending and scoring in the post, out of isolation, from spot up plays, etc.