3 Pointers
John Hollinger swats down Doug Collins for suggesting that it’s a mistake for Richard Jefferson to take a third of his shots from beyond the three point line:
Setting aside for a moment the ridiculously small sample (24 shots over three games), or the fact that Jefferson shot a career-high 39.7 percent on 3s last season and ranked among the league’s top practitioners of the corner 3, there’s the simple matter that for a wing to take one-third of his shots from distance is completely unremarkable in this day and age. Last season, of the 63 small forwards to play at least 500 minutes, 30 took a third or more of their shots from beyond the arc. With Jefferson transitioning from a leading role in Milwaukee to a secondary one in San Antonio — a switch that, for perimeter players, usually leads to a spike in the portion of shots that are 3s — I’d expect his portion of triples to stay around this level all season.
I would go further: In general, there’s not enough three point shooting happening in the NBA. In the 2008-2009 NBA season the average possession resulted in 1.083 points. The league average on three point shooting, meanwhile, was .367 meaning that the expected value of a three point attempt was 1.101 points. Better than average. Indeed, last year only four teams scored at a more efficient rate than 1.101 points per possession. If you consider that 26.7 percent of missed shots become offensive rebounds, the long ball looks even better. The break even point for three point shooting, on average, is something a bit lower than 36.1 percent.
In general, two point jump shots are kind of a sucker play: dunks, free throws, and three pointers are where it’s at.
26 Responses to “3 Pointers”
Register
RSSLogin
Topic Cloud
ArchivesBlog RollAbout Matt YglesiasContact Me
Donate
November 7th, 2009 at 4:31 pm
Major leaps in logic here, of the kind you often point out in others’ thinking. It seems likely that a majority of the most wide-open, easiest-to-make shots are already taken, and are made with much more than the .367 overall average. The overall average includes all the weak, off-balance, last-minute desperation shots, after all. There’s no reason to believe that the marginal potential 3-point attempt that’s currently NOT TAKEN would have anywhere near the success rate of the median current 3-point attempt.
November 7th, 2009 at 4:44 pm
OldK is clearly right that we need to keep mean value and marginal value clearly separate here. Clearly the (typical) marginal 3-point shot not taken right now has an expected return well below the mean value of a 3-point shot.
That’s consistent with thinking that there should be a lot more 3-pointers taken. It’s even consistent with thinking that the means Matt provides are some (defeasible) evidence that there should be more 3-pointers taken. After all, the alternative to a marginal 3-pointer is often a marginal 2-point jump shot, or a marginal pass to someone not really open but trying to cut to the hoop. And those aren’t great plays either.
There are differences in other sports where there are plays that are purely voluntary. (Like the stolen base in baseball for instance.) There the difference between mean and marginal value is really crucial for analytic purposes. But given there’s a 24-second shot clock, so you have to make some play or other, it seems reasonable to use mean values of plays as guides to their (relative) marginal values.
November 7th, 2009 at 4:52 pm
If you consider that 26.7 percent of missed shots become offensive rebounds, the long ball looks even better.
???
I don’t get this. Is Matthew saying that 3 pointers are more likely to be offensive rebounded than other shots?
November 7th, 2009 at 5:23 pm
Al,
I don’t have stats on this, so if I’m wrong, I am, but in my experience playing and watching basketball, long shots result in long rebounds, which minimizes the advantage the defense has in being close to the basket, which results in more offensive boards.
Hopefully there’s numbers on this somewhere, but that’s my anecdotal take on it.
November 7th, 2009 at 5:57 pm
I don’t get this. Is Matthew saying that 3 pointers are more likely to be offensive rebounded than other shots?
No. Even if you assume that a 3 point shot is no more likely to lead to an offensive rebound, Matt’s point is still valid. He doesn’t have the average points from a possession that includes a 3 point attempt. So he estimates the value of a 3 point possession by making the following simplifying assumptions:
1)that every possession that includes a 3 point attempt ends with a 3 point attempt (ie. a defensive rebound or offensive foul)
2)that the number of free-throws made per 3 point attempt is equal to the number made per 2 point attempt.
He then uses the percentage of 3s made to estimate the mean points per 3 point possession.
3)that missed 3 aren’t more likely to lead to offensive fouls, or – more likely – long rebounds that lead to fast break points by the opposing team.
Assumption 1 leads to an underestimation of the mean points per 3 point possession.
However, assumption 2 and 3 may lead him to overestimate the effectiveness of the 3 point shot. I’m not sure.
November 7th, 2009 at 6:13 pm
OldK at #1 has it. If teams attempted more 3PT, then the quality of their 3PT opportunities would become worse on average, because, in part, defenders would play tighter around the perimeter.
November 7th, 2009 at 6:18 pm
the fact that the game is already filled with 3-pointers, dunk shots, and fouls is why a number of us do not believe we are exactly in a golden age of basketball, assuming, as bill russell did, that there’s more to baskeball than the several minutes out of 48 that are the act of shooting themselves.
November 7th, 2009 at 6:29 pm
Gregg Popovich has four championships and last time I checked, Doug Collins has zero. The Spurs traded for Jefferson to replace Bruce Bowen who – get this – defended the opponents wing and shot a lot of three pointers. Coincidence? Any possibility that Jefferson is doing exactly what the SA coaches are telling him to do in the hopes he’ll fit into their existing (and highly successful) team system?
November 7th, 2009 at 6:51 pm
There is a big discussion in the Apbrmetrics community regarding three point shooting. Let’s say that one side maintains that teams should much take more 3s and the other says that there must be a yet to be discovered reason why teams don’t do it. Teams have been taking more and more 3s since the mid 80s and with increasing accuracy, so it’s not like we are approaching the point where the value of a three and a long two will be similar. Regarding the Spurs, their whole defensive system consists of letting the opponent shoot long twos and taking away threes and close shots. It works pretty well, I would say.
November 7th, 2009 at 7:02 pm
Matt, you just uncovered the Mike D’Antoni philosophy of offense, get yourself wide open threes and shoot them. Lots of em.
D’Antoni believes the most efficient offense is one that shoots a high percentage from behind the arc. Maybe he has been aware of this very persuasive stat for some time, or maybe it always felt right to him intuitively, but that is the essence of the offense philosophy he developed during his coaching stint in Italy during the 90’s.
High tempo basketball, run yourself into numbers, 3 on 2’s and 4 on 3’s, move the ball quick, hit the wide open guy behind the line and have him shoot on rhythm.
Lebron will probably find out next year how much fun it is to play for the guy. Do you think? They have mutual love for each other from their Olympic experience, I know that.
November 7th, 2009 at 7:09 pm
N, i basically agree with you, but the idea behind jefferson is to upgrade on bowen: that is, with bowen, the defensive harassment was the meat and potatoes and the scoring was gravy. SA will live with a little less meat and potatoes as long as jefferson gives them more gravy, lightening the load on the big 3.
which i think he has a good chance of doing….
November 7th, 2009 at 9:53 pm
I’d like to see them move the line further out. That means wiping out the corners as launching pads. Good. 3pters are antithetical to my idea of basketball as a game of movement and attack, not of long-range bombing.
While I’m at it, I’d make all fouls in the final two minutes 3-shotters. That period is often among the most boring in sports. Less fouling means more action.
I also endorse somebody else’s idea that basketball be turned into a best of 7 competition, by dividing each game into 7 7-minute mini-games. That eliminates garbage time, and means that one bad quarter doesn’t mean you have no chance to win the contest.
November 7th, 2009 at 11:53 pm
Except for the garbage time in the last three “minis” when you go 4-0?
November 8th, 2009 at 2:27 am
thanks for using expected value in shooting. far too often, people use things such as the “80-50-40″ metric to evaluate shooting completely ignoring that the point is to score. NBA teams take far too few threes especially considering the value of these shoots over two point jumpers
November 8th, 2009 at 2:32 am
oh and Max424, high tempo is how Doug Moe got Nuggets teams into the playoffs. I hesitate to say that his teams had little talent, but most of his players had their best years for the Nuggets (Mike Adams, Fat Lever, Calvin Natt, Wayne Cooper, and the rest)
November 8th, 2009 at 2:35 am
Two-point jumpers are overall bad strategy, but good tactics within a context where you need to attack a stingy defense. Dwyane Wade, for example, can completely control a game when his short and medium jumpers are falling. Same for Chris Paul and Gilbert Arenas. I think this works best for ball-handlers who can consistently punish a defense that gives them that shot, and can quickly shift to assisting other players when the defense adjusts. Rajon Rondo is a particularly odd case. He doesn’t shoot threes, hardly even attempts them, but is relentless is attacking the basket not to score, but to get shots for his teammates.
November 8th, 2009 at 3:22 am
Layups. You forgot layups.
Watching the 2009 Blazers, I’ve been a little disheartened by the incredible number of shot attempts that have occurred between the arc and the paint. The only thing worse than seeing amarcus Aldridge shooting a fadeaway jumper (repeatedly) over a 6′6″ Chuck Hayes is seeing him get the ball at the arc, step over the arc and then shoot a long, long two with plenty of time on the shot clock. Andre Miller will do that move too and I don’t understand it at all.
November 8th, 2009 at 4:54 am
This sounds great in Spreadsheetland but in the NBA there are a surprising number of shooters who don’t play even average defense and you rarely draw a foul on a three point shot which restricts your ability to shoot free throws. Playing above average defense and getting to the line also help you win in the NBA so it’s a lot more complicated than ’shoot more threes’.
November 8th, 2009 at 9:07 am
I don’t have stats on this, so if I’m wrong, I am, but in my experience playing and watching basketball, long shots result in long rebounds, which minimizes the advantage the defense has in being close to the basket, which results in more offensive boards.
That would be my hypothesis as well. On the other hand, those long rebounds that do go to the defense after a three-point shot sometimes lead to a quick and easy break-out opportunity with numbers. I would like to see some data on the opponents’ points per possession following a three-point miss and a two-point miss.
Another factor that needs studying is that a team’s two-point and three-point offenses are complementary. Generating open three point shots depends on the viable threat of going inside or driving the ball to the basket for a two-point shot opportunity. And finding open lanes to the hoop, or single coverage on bigs, for high-percentage two-point shots depends on the viable threat of a three-point score and forcing the opponent to defend the perimeter.
Another factor to weigh is what type of shot does more to wear down the opponents’ defenders. Three point defense requires a lot of running, chasing and fighting around picks; two-point interior defense is a grueling wrestling match. Both fatigue the defense, which can pay off on a crucial late-game possession.
Finally, as Matt mentions, foul shooting is a very important part of the game. Paul Pierce is a pretty good three point shooter. But he has lived at the foul line for most of his career, and he didn’t get there by shooting three point shots. Ray Allen is a great three point shooter, but if he didn’t drive the ball to the basket so often, he wouldn’t get a chance to put his 95% free throw percentage in to action.
November 8th, 2009 at 4:27 pm
It’s never really been determined if the negative of having long rebounds from trey attempts leading to transition baskets for the other team is counteracted enough by the actual 3PT% makes along with additional offensive rebounds due to longer rebounds.
Obviously, the makeup of your team, the layout of your size, the emphasis of your defense, can be purposefully laid out to address these concerns, but even so, it’s still unclear which is better, more treys, or less treys?
And, in the end, do you need an elite post player to make it all worthwhile?
November 9th, 2009 at 2:00 am
This is silly. If the guards block out as they should — but many do not — long rebounds just make 3-on-1, 3-on-2, 4-on-2 or 4-on-3 transition easier. The scourge of basketball today — the sure sign of a weak player and a weak team — is refusing to take the high-percentage midrange jumper because… well, because it doesn’t show up on SportsCenter as 3-pointers and dunks do. There’s a reason why some players average 25 ppg per game but never win.
An open 3-pointer is a 35-40% proposition. An open 15-foot jumper is more like 65%. Hmmm. . .
November 9th, 2009 at 11:24 am
21: I don’t know if your numbers are right or not, but it goes back to an earlier point. The mean averages are not that helpful when the expected outcomes are so close in value. It is the marginal opportunities that matter and it is far from obvious that they correlate well enough with the means for us to use the means. If the difference is between contested 3s by bench players and contested 2s by your best player, the contested 2s might be more preferable.
November 9th, 2009 at 12:10 pm
He’s comparing apples to oranges here: points per possession versus points per three-point shot. If you remove turnover possessions, points per possession concluding with a shot may be higher than points per three-point attempt (assuming that turnovers outnumber offensive rebounds- which I believe that they do). This, along with prior stated arguments re: marginal versus average, make yglesias’s comments dubious at best.
November 9th, 2009 at 3:49 pm
You gotta attack the paint before the 3-point line becomes available. Also, attacking inside (off the dribble or otherwise) tends to lead to more defensive fouls, which can screw up a team’s substitution pattern, frustrate them, and cause them hesitation at the offensive end.
November 10th, 2009 at 3:16 pm
[...] observed the other day that the average points per NBA three point attempt in 2008 exceeded the average points per [...]
November 13th, 2009 at 1:20 pm
[...] Dr./Author/MC Adam Waytz of freedarko.com spoke to our group this week and we discussed this work by Matthew Yglesias on Richard Jefferson’s proclivity for shooting the three. [Yglesias] [...]