By Jason Wojciechowski on December 21, 2009 at 7:30 PM
Today marks the first day that I'm abandoning ESPN's (basic) box scores and going with HoopData's advanced box scores instead. I know about some of the advanced stats the the real NBA geeks are using, and I know the kind of emphasis they've been placing on things like effective field goal percentage (or true shooting percentage) rather than vanilla (or naive, I guess you could call it) shooting percentage, on usage rate, on pace. To this point, I've essentially been estimating these things, trying to point out when a guy's points came only by using way too many offensive possessions, or when someone was ridiculously efficient or utterly greedy on the boards. But it's all been rough mental math, attempting to estimate possession counts in my head from a sense of field goal attempts + turnovers. Now, with the advanced box scores, that data will actually be presented to me.
Wolves 104, Celtics 122: For instance, in this entirely expected 18-point win by the Celtics at home over the hapless Wolves, I can tell you that the pace was quite high, with 101 possessions for each team, but also that Boston wasn't up to its usual standards of defensive efficiency, with a 103.0 defensive rating. That number is a simple one: it's just points per hundred possessions. (The definition of a possession ought to be mentioned: it's basically what you'd think it is, but an offensive rebound does not create a new possession -- it instead extends the current possession. As far as I can tell, this is basically so that both teams have the same number of possessions at the end of the game. (I don't see why there aren't sometimes one-off errors because of one team taking the final shot, but maybe we just round that out because the effect on the overall numbers isn't worth the pain of teams not having the same denominator in the rate stats.)) Here's some context: the current league pace average is 93, ranging from 101.2 (the Warriors) to 87.7 (the Blazers), so this 101-possession game was quite fast indeed. The league average offensive and defensive ratings (which are the same for reasons that I hope are clear) are 106.6. Offense goes from 114.3 (Phoenix -- it's not all pace with them) to 96.3 (New Jersey, natch). Defense ranges from 98.7 (Lakers, the only team under 100, and this shouldn't be any surprise to anyone who's watching so far this year) to 115.1 (Toronto, natch redux).
What else? There's turnover rate, which is the percentage of possessions that end in turnovers. The league average there is 13.8%. On offense, that goes from 11.2% (Atlanta) to 16.2% (Charlotte), so there's not actually a ton of spread there. The same lack of spread is found on defense: 16.1% (the Warriors, surprisingly) to 11.6% (Toronto again). There are the percentage numbers for rebounding. The offensive average is 26.9%, which is higher than you might think. (The defensive average is of course 73.1%, then.) On offense, the range runs from 20.3% (Warriors again, no surprise given their size) to 33.4% (Memphis, they of Marc Gasol). On defense, you've got 78.4% at the top (Orlando, Dwight Howard) and 69% at the bottom (Warriors one more time).
Then there's some shooting stats, like free throws per field goal. The league-average there is about 0.23. On offense, the Nuggets get to the line a lot (.33) and the Bucks never get there at all (.162). On defense, the Hornets don't foul (.186) and the Bucks (again) do (.294). Finally, there's adjusted shooting percentage, which is represented by two stats, Effective Field Goal Percentage (eFG%) and True Shooting Percentage (TS%). The point of both stats is to adjust for the fact that threes are worth more (so a lower percentage on them is acceptable -- the easiest example being the extremes, that 33% on threes is the same as 50% on twos, as a matter of the probability of scoring working out to averaging a point per shot). The difference between the two stats is that TS% takes into account free throws, while eFG% does not. I can't easily find a ranking of TS%, but in terms of eFG%, the average is 49.6% and the leaders track the efficiency ratings, from 55.1% at the top (Phoenix) to 43.7% at the bottom (Nets) on offense, and 45.4% at the top on defense (L.A. Lakers) to 52.9% at the bottom (Warriors). (It's not a tautology that eFG% would track the Offensive Rating and Defensive Rating numbers, though, because turnovers could factor in -- a team might harass another team into a lot of turnovers but not be very good at keeping them from missing shots, or vice versa. If you look at the actual rankings, the numbers track fairly well, although taking the TOV% leaders shows some discrepancies. For instance, the Warriors are last in eFG% allowed, but their league-best turnover rate allows them to be 25th in Defensive Rating.
A lot of what's in the advanced box scores, rather than providing context, as described above, instead just adds more information. For instance, I can tell you that Ryan Hollins drew two charges in the game, and that Kendrick Perkins had hsi shot blocked three times. I can tell you that there were a total of three and-ones in the game. I can tell you that Kendrick Perkins shot 7-10 at the rim and took no shots from farther away than "at rim". I can tell you that of Rajon Rondo's 15 assists, seven led to buckets at the rim, and five led to threes.
There are some stats that require some estimation (although I've read that, with a few exceptions, these estimations are remarkably close to what we'd get if we did a play-by-play analysis of each player): usage, assist rate, turnover rate, rebounding rates, points contributed, and possessions used. Usage is a percentage stat estimating the percentage of his team's possessions he used (by "used", we mean "had the ball last" -- either you took a shot, got fouled, or turned it over). Possessions used is a counting stat which incorporates 1/3 of the player's assists and subtracts out 1/3 of the player's assisted field goals made. (The idea of this is presumably that on about 1/3 of assisted field goals, it was really the work of the assister that should be credited with the possession, not the assistee. If Chris Paul works his man, drives to the bucket, and dishes to Peja on the wing for a catch-and-shoot three, who really used that possession? After all, if after all that work, Chris Paul lost the ball, he'd have used the possession in the stat sheets. Now, if Chris Paul does all the work and Peja misses the shot, then that's all on Peja: the possession is his entirely. This, on the one hand, is probably fair, and on the same hand, likely represents the limitation that we don't have a box score stat for "would've been an assist if the shot had been made".
I think that just about covers it! I'll try not to write any more than I usually do, because I could spend all day in these stats, but if I do that, I run the risk of losing my two readers. So I'll keep it reasonable, but hopefully more informative than ever.
So back to Minnesota's loss to Boston. I've already mentioned that the game was played very fast, so Boston's unusually high points-allowed total is more due to ten extra possessions in the game than it is to overall bad defense (103 rating versus 100 on the season). Neither team did a great job on the defensive glass, as both teams were above-average on the other end, especially Minnesota, with a 31.4% offensive rebound rate. Kevin Love, in particular, ate up 17% of the offensive rebounds available, for a total of five. The Celtics were carried by Paul Pierce, whose 29 points came on just 18 possessions. The Celtics were a ball-movement machine in the game (or the scorer was particularly generous, either way), as Rajon Rondo's 15 dimes were symptomatic of 34 on 42 field goals overall. All eight of Ray Allen's field goals were assisted, in fact.
Cavs 95, Mavs 102: That's a very solid home win for the Mavs. The big number that jumps out is just six turnovers, a rate of 6.7%, a number that's nearly half the league-best 11.2% rate of the Hawks. The Cavs got to the line (.338 free throws per field goal), but the Mavs made the shots (19-20), with Tim Thomas putting up the only miss, hitting 6-7. And the Mavs did all this without Dirk Nowitzki, who was still out with teeth in his elbow. Just wanted to continue to the gross-out for one more post. The Mavs did a nice job on LeBron, who got to the line only seven times on twenty-three shots, and who made just four of his eight shots at the rim. LeBron's shot selection also comes into play, as he hit 2-7 from the long-two distance (16-23 feet). Long twos are, of course, the worst shot in the game overall: you're far enough away from the basket that your percentage is low, but you're not so far that you get an extra point for making the shot. LeBron isn't the kind of shooter whose stroke is so pure that he can hit those shots with great consistency. Shaq was a monster on the boards with a 26% rebound rate, but he only played 18 minutes, perhaps in part due to fouls (he had three), perhaps in part due to ball-hogginess (32% usage, second only to LeBron, but finishing with just 4.3 points contributed on 10.4 possessions), perhaps just because he's old, or perhaps because Mike Brown is, after all, still Mike Brown.
Lakers 93, Pistons 81: Granted that they didn't play any Eastern big boys, that whole thing about how the Lakers were benefiting from a bunch of home games, etc.? That's a 4-1 road trip. The Lakers won this one with defense, as the pace was actually fairly high (97 possessions). Detroit managed just an 83.5 Offensive Rating, though (not that the Lakers' 95.9 was anything to write home about). The Pistons' excuse is that they're missing most of their good players. Thus, they have Rodney Stuckey, who actually got to the rim (eight shots there), but couldn't finish (four made), and took the rest of his shots from distance (2-5 on long twos). He also had just one assist in fifteen possessions, a pretty low rate for a supposed point guard. The Lakers' offensive excuse probably revolves around Kobe's broken finger, but that finger doesn't affect shot selection (3-6 from long two, 1-4 from three). L.A. did force turnovers, as Detroit handed over the ball 22 times, including 15 Laker steals, led by six from Ron Artest (Kobe added five, presumably with his left hand). Artest had nine assists, five of them leading to buckets at the rim and three leading to threes, so he had an excellent night. Andrew Bynum added 14 points contributed (PC from here on out, blogwide) on just nine possessions. (By some math I don't understand, Hoopdata claims that works out to 1.50 PC/PU.)
Adam Morrison, by the way? Still really trying to impress the brass. In his nine minutes of (by definition) garbage time, he used 38% of the team's possessions. Sadly, all that work ended in just one field goal and one assist.
Blazers 102, Heat 95: Both of these teams are in the bottom four in Pace, so they egged each other on in this one, resulting in just 85 possessions, below even Portland's league-slowest 87.7. Obviously if you score over 100 while playing that slow, you're playing pretty good offense. Brandon Roy led the team with 28 and eight assists, for an excellent 30 PC in just 18 possessions. LaMarcus Aldridge, by contrast, scored 23, had just one assist, and had half of his buckets assisted, so he managed just 20 PC on a team-high 24 possessions. Dwyane Wade used an astounding 36 of his team's possesions (remember, there were only 85 in the game), but his efficiency was merely ok, with 28 points and ten assists, as he missed eighteen shots and turned it over three times. Quentin Richardson, though, had a game you don't need an advanced box score to appreciate: 7-7 from the floor, 7-7 from three. Add in his assist and he actually averaged slightly more than three points contributed per possession. Which kind of makes sense in this case since every time he touched the ball he made a three. (James Jones had a fun efficiency rating, too, as he hit his one shot, a three, in nine minutes played. The only thing dragging him down is that the shot was assisted. It's a good thing he hit that three, though, because he was that close to putting up a nine-trillion.)
Bobcats 94, Knicks 98: And look at the Knicks! Yes, they're 10-17. Yes, they're 11th in the Eastern Conference. But they're only a game back of the Pistons! They're right in this thing! Charlotte plays bad offense and good defense. The Knicks play average offense and mediocre defense. What did this clash of styles add up to? A defensive battle, as neither team got over the average Offensive Rating for the game. The Knicks made it work by getting to the line, putting up an excellent .423 free throw rate, led by, of all people, Big Cock, who took ten free throws on 14 shots. Somehow, he ended up with zero rebounds, though. How does a guy his size do that? Also, from his shot chart, it seems like Cock was getting fouled in bad (for the 'cats) situations, since he only has one field goal on the books at the rim and just one more from less than ten feet. By contrast, he took seven threes. Maybe he was just the designated foulee down the stretch? Eh, doesn't matter that much. David Lee's line pops, though: 15/15/7. He struggled with the shot, but grabbed 38% of the defensive boards. Matching him on the other end was Tyson Chandler, who apparently responded to Gerald Wallace calling him out by grabbing 38% of his own, for a total of 14. Or maybe it was just because Gerald Wallace didn't play in the game. After all, Boris Diaw didn't respond (four boards, 10% overall rate), but Ray Felton did (nine boards, although that's more playing time than anything, since his rate was 13.5% (which is still nothing to sneeze at for a guy his size)). Good sign for the Knicks' defense? Four drawn charges in the game.
Hornets 92, Raptors 98: Hey, whoa, a Toronto game where the other team didn't blow them off the defensive end of the floor. New Orleans managed just a 95 Rating on offense, "led" by their three stars: Chris Paul shot 3-13, missing eight shots from the mid- and long-range, and turned it over five times (though he did have seven assists); Peja shot 4-15, including 1-7 from three; and David West hit only nine of twenty-one, although he did get on the defensive glass (25%). Chris Bosh had 25/11, but six turnovers, and Andrea Bargnani missed threes all over the place (1-6), so the offensive stalwart was Jarret Jack, who had 15/5/7 and no turnovers. He was the only Raptor to contribute more than one point per possession. Remember, Toronto won this game.
Nuggets 96, Grizzlies 102: Z-Bo just destroyed this game, scoring 32 on 21 shots, hitting a three, grabbing twenty-four rebounds, and even dishing three assists. He did all this while committing just two turnovers. So that's a 1.27 PC/poss., a 21% offensive rebound rate (context: the league leader is Ben Wallace with 16%), 36% on defense, all with a turnover rate of just 7%. The Nuggets just didn't get on the glass, as Z-Bo's 21% offensive mark topped all but one Nugget's defensive rebound mark (that one being Birdman, of course, who did great work on the boards in his 22 minutes: 28.5% offensive, 36% defensive). It wasn't just Z-Bo for the Grizz, as Sam Young pulled down 29% on offense as well, although his 2-5 shooting at the rim tells me he might have gotten some missed tip shots into the stat books. Denver was led by Carmelo's 41 on solid shooting (13-26), but he shot just 1-6 from three and turned it over five times. He did get to the line 17 times, though, and sixteen of his shots were of the at-rim variety. Taking 22 of 26 shots at the rim or from three is generally an excellent recipe for success, but the threes just didn't fall for Carmelo this time.
So that was a really long post. I may struggle to keep these things reasonable. But like I said, I'll do what I can. Right now, I'm just sort of reveling in the incredible amount of information and the picture you can gather. You know, being able to criticize LeBron's shot selection with actual data, to point out why Carmelo is maybe the deadliest scorer in the game, to show just how impressive Z-Bo's 24-rebound game is compared to Birdman's 14-board outing, and on and on and on ... it's a dream come true. I'll get better about condensing it down to just a few tidbits. Or else I'll keep writing Basketbawful-length posts and you'll (all two of you) just have to deal.
Also, if you see anything obviously wrong or confusing with my explanation of the stats I'm using, let me know. You know where to find me. I'll update this post as needed, and try to be clear in the future about what I'm talking about.