Nba nerd

MUST a Team Shoot From the Mid-range?

In last week's article about the midrange shot, this old argument reared its ugly head in the comments:

However, what if a team thinking to maximize their expected value simply stopped taking any shots that weren't 3-pointers or at the rim? So much so that an opposing team could rely upon those shots not being taken? In that scenario, the opposing team could change their defense to put all focus on 3-pointers and shots at the rim so that the expected value of those shots ends up much less than the average on which Wages analysis is based. 

This has come up before. There is a ton of anecdotal and logical evidence that this is false. We're also going to be putting up a ton of emprical evidence that this is false, but not until after Sloan.

The obvious anecdotal counter to this argument is, once again, Houston.  According to stats.nba.com, the entire Rockets team has taken 167 15-19 footers this year. To understand just how low that number is, consider that LaMarcus Aldridge has taken 303 all by himself. Further, note that every defense that plays against Houston knows that they won't try to get those shots; they only take them about 4 times a game, or on about 4% of possessions.

And yet the Rockets, playing against defenses that know exactly what they are trying to do, manage to shoot 47.2% (above average), 53.5% from 2 (significantly above average), almost average from 3 (34.2%) and the team achieves Jeff Hornacek's "magic" threshold of > 51% eFG (52.8% eFG), as well as a 56.7% true shooting mark (again, way above average). It's worth noting also that the Rockets shoot only 31% from 15-19 feet. This idicates that perhaps these are "desperate" shots taken when the clock is winding down; that would make sense given the frequency.

In other words, running an offense where you only shoot threes and layups does not appear to make it easier to guard that team. At least, it doesn't appear to make it easier to guard the Rockets. In fact, the opposite appears to be true. The main reason is that pretty much all NBA defenses already focus on 3-pointers and shots at the rim. It's not like teams march into Houston and the coach says "Hey guys, I know all year we've been saying we'll just let 'em fire from beyond the arc or under the basket, but tonight, we're going to defend the rim and the three-point shot!"

I'll re-iterate that we've got a lot of empirical data on this. The Rockets just provide an interesting natural experiment.

This is intuitive if you think about geometry; the area one needs to cover to guard mid-range jumpers is much smaller than the area you need to cover to guard 3 point shots. So it makes sense that defenses have to scramble lots to cover the whole three point line, and it further makes sense that the fact that when players aren't stopping to pull up from 16 feet when they drive, that doesn't actually help the defense; in fact, since most defenses are trying to make the offense do just that, the fact that a team like Houston is relentlessly trying to get the rim makes the defense work harder; defenders trying to "help the helper" or to rotate after a penetrate-and-pitch drive followed by a swing pass have to cover more ground when the ballhandler gets closer to the rim before pitching.

Of course, a team has to get players that are good at finishing at the rim and/or getting to the free throw line. Otherwise the "layup" part of the offense's threat is easy to prevent, and the helpers and rotaters don't have to scramble as much. This type of offense works well when you have James Harden. On the other hand, it's pretty clear that there are players with similar skillsets to Harden who could become more efficient by being more relentless with a "three or the rim" approach (Mr. Anthony, my gaze is fixed firmly upon you and your 203 midrange jumpers).

I don't think the evidence provided by the Rockets is as clear as you make it. Your analysis looks at the Rockets 15-19 foot shots, but I don't think that makes sense if you are trying to see how often they shoot from areas that are not at the rim or from three. There are plenty of shots that are closer than 15 feet but not at the rim (as well, there are shots that are longer than 19 feet but not from behind the line). I think you are looking at the wrong tab on the stats.nba.com site. You instead shoot look at the "By Zone" tab, rather than the "5 foot distance" tab. In the "By Zone" tab, you can see 3-pointers and at the rim (or, at least, in the restricted area) directly. For Houston, 39.44% of the shots were in the restricted area, 32.70% of their shots were behind the 3-point line, and 27.85% of their shots were either from midrange or in the paint (outside the restricted area). Note that both midrange and in the paint (outside the restricted area) have FG% below 42%. In other words, more than a quarter of Houston's shots are NOT "three or the rim" and thus can be classified as "bad" (i.e., have an eFG% less than 51%).
The Austin Toros (Rockets' D-league affiliate) certainly think it's not a problem:

http://m.bleacherreport.com/articles/1901342-the-houston-rockets-are-doing-it-again

What a beautiful shot chart. Average at the rim, but the 3 point percentages are insane.
Al,

I dislike the NBA's way of categorizing these. For instance, the have a category 16-24 feet. Which is moronic, because teams shoot tons of 16-24 footers; 24-footers are worth 3 points. Clearly, 16-23 should be a different category than 24+, but it isn't on that site.

But I think the fact Houston takes so few 15-19 footers is definitely a sign that the offense is trying to avoid shots outside the paint and below the 3.
looking at the "by zone" tab, the rockets shoot the fewest from mid-range in the league, by a wide margin. the same story is told looking at the "5ft distance" tab, they have the fewest shots by far from 15-19, and the fewest shots, albeit by a less sizeable margin, from 10-14.

"MUST a team shoot from mid-range?" is a bit of an exaggeration, because, at times, obviously, a team MUST shoot from mid-range or just turn the ball over, but the rockets this season are pushing the envelope on just how few shots they MUST take from mid-range, shooting the fewest shots from 10-19 feet away in the league by a hefty chunk while maintaining a very efficient offense. This is just more evidence that reducing 10-19ft shots to as few as possible doesn't appear to have a downside. Sure they shoot over a quarter of their shots from mid-range, but that's the low-water mark for the league.
Patrick, I agree. As mentioned, I recommend the "By Zone" tab on the page you linked to (instead of the "5 foot distance" and "8 foot distance"). It has a different way of categorizing shots: Restricted Area, In the Paint (not Restricted Area), MidRange, Corner 3, Above the Break 3. That's what you want.

The most interesting thing from my perspective on that page is that shots that are In the Paint but outside the Restricted Area are bad shots too! (Only the Heat shoot them at >45%, and most teams shoot them at >40%.) And teams seem to take a lot of shots from that distance (Houston takes 16.5% of its shot from that distance, but only shoots 41.7%).
I would add further that the Rockets are still dead last at attempts from the midrange, with half as many as Philadelphia (which has a coach and GM that hate midrange shots), and that, despite this, Houston is still way above the median at FG% in the paint (7th), and the restricted area (4th). They are bad at above the break 3s and about average in the corners, but still far above 50% eFG on threes.

Two lessons from this:

1) the above-the-break three isn't a great shot but it breaks even, so is probably a great shot to "keep the defense honest"
2) it still doesn't look like the defense has it any easier by knowing it can "ignore" the midrange against Houston
Al, agreed on the shots in the paint but outside the restricted area. I suspect many of these are heavily contested shots from someone trying and failing to get to the rim and not making a smart decision like pitching to the wing. Clearly, open shots from the free throw line would be great shots for most teams, but NBA defenses won't give you that often.
The Heat and Spurs who knew the value of the 3, especially the corner 3, last season (Heat even led the league in 3 pointer defense and kept the Rockets to low 3 point % in both meetings). But the also saw great value in the midrange shot, at least some of their players. Green, Leonard, Manu and friends took nothing but layups and 3s, while Parker and Duncan were given the green light on mid. Lebron, Bosh, and Wade liked the mid, but Ray, Battier, etc. treated that area of the court like lava. http://grantland.com/the-triangle/courtvision-how-the-heat-and-spurs-score/

You can go back a few years and see Kobe and Gasol living in the midrange. Same with Dirk who beat the Heat. MJ lived in the mid range second half of his career and did alright. Hamilton on the Pistons lead them in scoring (?) by taking midrange shots off of screens

It seems for those who excel at midrange shots there is a definite premium for those skills and highly successful teams seem to see a value in those shots that the Rockets and this blog don't. I'd say the midrange shot from those types of players is very valuable, but I'd never let players like Wall/Beal or Josh Smith take them. I'd go into why I think the mid range has high value for good midrange shooters , but this comment is long enough already.
"Clearly, open shots from the free throw line would be great shots for most teams, but NBA defenses won't give you that often."

Patrick, isn't that the point of this post? :)
DG22,

You don't know us very well if you think the fact that Kobe and Hamilton took tons of mid-range shots is a counter-argument for us.

Hamilton has been awful and overrated for many years, and 2011-12 was by far Kobe's worst season, and not coincidentally the season where he loved the midrange shot the most.
Hamilton is bad, old and washed up for a while now, but I was taking about for the championship years where he was one of the the quality starting 5.

Spurs were your favorites for the titles last year and the Heat's success speaks for itself. The best players even nowadays are taking midrangers and are good at these midrangers and their teams are winning titles. The very good, quality, productive role players for these same teams never take them.
Will the eventual SportsVU data help with this discussion at all, will there will be a way to compare contested threes and at-the-rim shots to wide open mid range shots?
Just wanted to point out that NBA.com's 16-24 range only includes 2-point shots, even though it oddly includes 24 feet within the range. I checked their numbers against Basketball Reference's Shot-Finder, so using the 16-24 feet designation seems like the way to go, when discussing "bad mid-range shots" or "long-two's".

Furthermore, from what I understand, if you sort shot locations by "zone", NBA.com defines midrange as any shot that's not in the paint, or not a 3-pointer. This is problematic, as a shot from the free-throw line would be classified as a "paint" shot, despite being almost 16-feet out. Similarly, a shot from the side of the hoop, but not in the "paint" is classified as mid-range, despite being 8-feet out.
I am sure the empirical evidence will settle the argument, but let's focus on the anecdotal one. First of all, to understand the relevance of what Patrick says, it's important to know if 3s are included or not in the in the 20-24 ft range (btw, when are you guys switching to meters?). I think Patrick is right, and 3s are included in the 20-24 ft range. So his anecdotal range says that that Houston is shooting very little mid-range and still shooting quite well. Even more anecdotal evidence says that San Antonio takes a lot more of mid-range shoots, and their percentage is better (some times a lot better) from any distance but from 5-9 ft (where it is 42.5% against 43.2%). This tells us that the shooting percentage of a particular team compared to the rest of the NBA may depend a lot on that team's play and shooting ability. So maybe (MAYBE, it's just for the sake of the argument) if Houston was taking more mid-range shots their %s would rise. Finally, super-anecdotal, yesterday I went to the gym, made a little experiment and shot 87/150 mid-range shots. I was alone in the gym, and I'm a terrible shooter. This tells us that the defense it's quite important in influencing the %... By the way, I think Patrick is basically right, in particular I completely agree with what he says at the end on Harden and Anthony. But I don't think he proved it (he will probably do it with the empirical evidence he was talking about).
Is the answer not necessarily dependent on what another team does?

Let's ignore cumulative data and instead look at this from a geometric perspective: the maximum stress you can place on a defense comes from making them cover the largest area possible from which your offense can be effective, while having the personnel to credibly threaten that entire area.

This tells you a few things:

One, the magic of the 3 and rim strategy is that the rim is very far from the 3 point line. Making people defend both makes them run through the mid-range area, even if they aren't "defending" it, as NBA players as of yet cannot teleport between the rim and the 3 point line.

Two, this relies, as Patrick previously stated, on having credible threats in both areas. An offense that solely attacked the rim or solely shot the 3 pointer credibly while being poor at the other part (if everyone shot 3s like Josh Smith or finished at the rim like Rubio, for instance) would likely be trivial to defend.

Three, the real answer to "must a team shoot from mid-range" should always be "yes, when it is the best shot". There will be times (either out of desperation, or by design) when the mid-range shot will be available and the optimal outcome. I suspect these times are rare, or they are available to teams that have elite mid-range shooters (for instance, should your team entirely avoid long 2s if you have Dirk Nowitzki?), and can run offensive sets that use the abilities of these players to stress a defense.

So I suspect the conclusion that floor geometry will hint at is this: most players should not take long 2s (unless it is the only shot available at the end of the clock), however teams that happen to have elite mid-range shooters can probably use them as very effective weapons (and, this leads me to wonder if having a single elite mid-range shooter within a strong offense correlates to winning championships or with corresponding increases in that players' offensive efficiency elsewhere on the floor, as the presence of Bosh and Nowitzki on the past two champions is mildly suggestive).
"...In other words, running an offense where you only shoot threes and layups does not appear to make it easier to guard that team. At least, it doesn't appear to make it easier to guard the Rockets. ..."

Post hoc, meet propter hoc.

Maybe it's the other way around - that is to say, perhaps the Rockets don't take those low percentage shots because their offence is particularly efficient at creating good shooting opportunities. (Yeah, I know 'shot creation' is a bit of a bugaboo, but if it exists, that's what it should look like.)

Naively assuming that the rate of 3-point and layup / dunk opportunities is roughly fixed for a particular style of play, we would expect that a team which "simply stopped taking shots that aren't 3 pointers or at the rim" to have a relatively low pace, but the Rockets' pace is in the top quartile of the league (and has gone up in the last 2 years.) This makes me think that Houston is leveraging some kind of scheme improvement beyond improving shot selection.
The fun part is this is really just the spoiler. The actual data and theory might make some heads explode :-) Can't wait till Sloan.
I am going to play a little devils advocate here, in two different ways relating to this. Patrick has used Houston as an anecdotal example of why the proposition above doesn't hold up. So I am going to use a different anecdotal example to broaden the discussion a bit.
In the preseason rankings Portland was projected to win just 28 games for the entire season. At the present time they have 31 wins, and the BasketballReference projection projects them to finish with 56, twice what they were projected to have.
Now Portland is an interesting case study, because rarely do you have the opportunity to find about as close as one will get to seeing the example of the often used caveat "all other things being equal, what is the impact of changing one variable". Portland made a number of changes, but in totality there was really one one change made that really mattered, all the other changes were so small that it in many respects, all other things do remain equal, or at least nearly so. The only meaningful change Portland made was losing JJ Hickson, and adding Robin Lopez. Now per WP JJ Hickson was the most productive Blazer last year on a per minute basis, and he played a substantial number of minutes. JJ Hickson was almost twice as productive as Robin Lopez was in NO, and Hickson was 2-1/2 times more productive than LaMarcus "Yay Points" Aldridge was in 2012-13. Yet Portland has removed Hickson from the equation and added Lopez, and they are winning at a rate twice what was projected, and 170% higher than Portland actually won last year. Does that then indicate that WP greatly overweight's the production of the things JJ Hickson does well, and greatly underweight's the things that LaMarcus Aldridge and Robin Lopez do well?

Now onto point #2. Houston shoots the fewest number of mid-range two's of any team in the league. Houston was projected by BSG group to win 59 games, and is actually on pace for just under 51 wins (per BBR projection tool). Houston has taken 167 shots from 15'-19' as a team.
Portland on the other hand was projected to win 28, and is on pace to double that projection. The single highest usage player on the team is LaMarcus Aldridge, who has a usage rate of 29.3%, and, as Patrick clearly pointed out above, has also shot 303 shots from 15'-19' all by himself, 181% of the entire team total for Houston. Now if mid-range two points shots are bad shots to take, and Portland has the player that leads the league in said shots, and that player is also the highest usage player on that team, then it would seem to indicate, anecdotally at least, that Portland should be a very inefficient offensive team. On the contrary, Portland is the highest scoring team in the league, and also the most efficient offensive team in the league, scoring 114.1 points per 100 possessions. All of this seems incongruous with the proposition that mid-range 2 points shots are necessarily bad shots.
Ziggy, while it's true that the Blazers shouldn't be efficient, they do have a couple of hyper-efficient guys playing like stars that make up the difference and then some.

Also, since neither of them were doing that last year, it's hard to say that all other things are equal. Having two extra star players on your team than the year prior is a fairly large jump.
The Rockets are not a team that only shoots threes and layups, they are a team that mostly shoots threes and layups.

The comment in question is wondering what would happen if a team was so focused on threes and layups that it could be relied upon to habitually pass up opportunities for completely uncontested long twos.
In that scenario, the opposing team could change their defense to put all focus on 3-pointers and shots at the rim

The idea that NBA teams are not doing this already is so ridiculous, it's hard to even take seriously.

Protecting the paint and closing out on three point shooters are the bedrocks of every coach's defensive scheme from middle school on up. It is basketball 101.

Well I guess the discussion is about "lowering the number of mid-range" shots to Houston levels and not seeing %tage drop. Because if the discussion is about declaring at the beginning of the season "will never ever take a mid-range shot" and hoping your efficiency will not drop even if the other teams are defending only the 3s and the rim, well that's nonsense. Just as an example, in that case, in a wide area of the court, defenders should worry only of dribble and pass, making dribble and pass more difficult. It's a silly example, but since I am a lot taller of basically the whole population of the country I live in, I've played a few times pick up games under the rule that I could not shoot in the paint. Obviously, as I enter the paint under those rules, defenders are just caring about don't let me pass the ball, and they defend in a completely different way. I mean, more seriously, a good defender focuses on stuff like "that guys always goes left" or "he shoots a lot from that position". When I was a bench warmer in a "serie C" (sixth division in Italy were the first two divisions are pro level) , my team mates that were good defenders always focused on this stuff, and I guess as you go up the ladder these details are more and more important. They don't show up in WP maybe, but that's because WP it's just an approximation (don't get mad, General Relativity is just an approximation too)
This is not equilibrium yet though. McHale should tell his players to shoot less of the 15-19 footers if they're only going in 31% of the time. If you postulate that these are all heaves with the clock running out (to which I'd say heave it from somewhere else) then there will never be an equilibrium where midrange shots are actually efficient.
To the other commenters, I understand that there are players who can make the midrange shot more efficiently, but they don't play for the Rockets. So the Rockets should try to minimize the midrange shots taken, because their team makes them infrequently. And all midrange advocates, although usually without saying so explicitly, rest their arguments on the premise that an equilibrium exists (you gotta take the midrange to set up the other shots). Well to get to this equilibrium (which I think may involve a negative amount of midrange shots and therefore not exist) you should take more of the shots that get you more points and fewer of the shots that get you fewer points until you reach an equilibrium where the expected points are equal. This is the basic premise of equilibria-the expected points from all choices are equal. No sane team would frequently take a shot that goes in less instead of a shot that goes in more, so the Rockets would be defying all logic by increasing or failing to decrease their midrange shots taken. We should have plenty of test cases according to game theory, because it dictates that teams would shy away from taking shots with fewer expected points. Unfortunately, thanks to the logic-defying behavior of players and coaches, we haven't been able to see what an equilibrium would look like. I hope everything I said made those of you who support midrange shots say duh, because most of it is obvious. But I also hope that seeing it put together helps you understand how an equilibrium analysis dictates that teams should shoot fewer midrange shots.
I'm not sure Nathan is right. Correct me if I am wrong. Let us call x the average number of 3s, (1-x) the number of mid range shots, g(x) the percentage of 3s, f(x) the percentage of mid-range. The average production is 3x g(x) + 2 (1-x) f(x). The derivative is 3 (g(x) + x g'(x)) +2 ((1-x) f'(x) -f(x)). If we assume f(x) and g(x) to be constant, then x has to be increased if 3 g(x)>2 f(x). That's Patricks and Nathan maths. But the discussion that a post by @aeneas started was based on the hypothesis that, due to different defense strategies based on the opponent's shot selection, f(x) and g(x) are not constant. If we assume in particular 3 x g'(x) + 2 (1-x) f'(x)
(half of my post didn't show) If we assume 3 x g'(x) + 2 (1-x) f'(x) > 2 f(x) - 3 g(x) (i.e. if g(x) decreases much faster than f(x) grows with growing x), you have to decrease x to improve production. To know how to improve production you need to know the dependence of g(x) and f(x) in a neighbourhood of x (keeping other factors fixed) and not only the value in x
Well I didn't have any pen and paper available so maybe I didn't derive correctly, but I guess you need to know how the percentages will change if you change shot selection (my g'(x) and f'(x)) to optimise. Nobody assures that g'(x)=-f'(x) or 3 g'(x) =- 2 f'(x) or similar
Definitely teams are trying to take more threes. And the rate seems like it is going to increase as folks copy the Rockets strategy and get used to the math. But I think three point percentage varies widely based not the quality of the look. It is fairly common during the course of the game for a player to get a wide open, feet set, three point shot. Usually this comes at the end of a play that drew extra defenders and then a nice pass. That shot is a high percentage shot, especially if in the corner.

But that shot is really a defensive breakdown. That is only going to happen so many times during a game between evenly matched teams. So to increase the number of threes attempted, a team will also have to take threes off the dribble or off of isolations. My guess is those threes go in much less often. So I believe there is a limit (and we are generally still pretty far from it) to the number of threes a team can jack up against an even opponent (against an overmatched opponent you can do what you want) and still have the shot have attractive percentages.
One more predictive point. If teams start following the Harden/Rockets drive and draw a foul strategy across the league, the Refs (perhaps with or without official league instructions) will start calling the game differently. The NBA already gives way too much deference to the shooter in that it regularly hands out fouls to defenders when the shooter is the one initiating contact. That isn't actually consistent with the rules. Nor is it called quite that way in other leagues. And if you went down to your local playground and regularly pushed off with your off hand and claimed that you were getting fouled, you would fairly quickly have your fouls ignored or you would be asked to leave the court.

So not to be a spoiler on the study for Sloan (which I'm excited about), but the rules of the game might change. The fans and, I would guess, the league like threes. But no one really finds free throws very exciting. And this is especially the case in the NBA where the players (for reasons we don't need to discuss) are in general not exceptional at executing free throws (and in some cases shockingly bad). I don't need to see NBA players shooting a ton of shots less successively than the guys from my saturday pickup game.
Aphos,
I do not dispute that the good 3 point shooting from Lillard, Matthews, and Batum does not have an impact on Portland's record, it most certainly does. Your assertion ("they do have a couple of hyper-efficient guys playing like stars that make up the difference and then some") is essentially ignoring the impact of Aldridge (they make up the difference and then some), and you cannot do that.

Regardless, my point in my comment was related to the question Patrick posited in this article "MUST a Team Shoot From the Mid-range?", and he used Houston as an anecdotal example. Portland is simply a team that seems to conflict with that question. I was attempting to lead the thread into a deeper discussion of this idea, but since I was unsuccessful, I will make the point more clearly now.

I fully understand the concept of TS% and eFG%, and I believe both are very good numbers at expressing a concept. I do believe though that both measurements are flawed to a degree, and let me explain.

I will use eFG in this example, because TS% includes FT, which is not germane to the question Patrick posited. If you have 2 teams, and one shoots exclusively mid-range two's, and makes exactly 1/2, they have a eFG of 50%, and PPS of 1. If you have another team that shoots exclusively 3's, and makes exactly 1/3, they have an eFG of 50%, and a PPS of 1. So they are the identical, and as such it makes do difference which approach you take.

In truth though they are not the same, at least in terms of net efficiency. The team that shot 2's missed exactly 50 shots, and in the NBA the ORB/DRB differential is 27%/73%. So of the 50 misses they were secure a second possession 13.5 times, but the opposing team secured the possession 36.5 times. NBA teams over that last 10 years have a points per possession of almost exactly 1. So the team shooting two's scored 100 points on their first possession, but their misses allowed the opponent 36.5 possessions which they converted into 36.5 points.

For the team shooting exclusively 3's they have missed 67 shots, which converts into 22 second chance opportunities for them, but gave their opponent 44 new opportunities, resulting in 44 points for the opponent.

So after just the first chance opportunity the team shooting exclusively 2's scored 100 points, but their misses led to 36.5 points for their opponents, a net of 63.5 points. The team shooting 3's scored 100 points on their first chance points, but allowed 44 points through their misses, resulting in a net number of points of 56. So even though each team has an identical eFG, the team shooting exclusively mid-range 2's will be 7.5 points better than the team shooting exclusively 3's. If you take this concept out to it's logical conclusion (essentially 4th chance opportunities), then teams shooting 2 points shots with an eFG of 50% will be net +/-4 points better off than teams shooting exclusively 3's with an eFG of 50%. So the eFG% is either overrating the effect of 3 point shots or it is underrating the effect of 2 point shots.

So back to Patrick's question,
"MUST a Team Shoot From the Mid-range?"
The answer seems to be yes.
Given that Portland has LaMarcus Aldridge which is the #1 mid-range shooter in the league in volume, if not in efficiency, and that they are the number one team in offensive efficiency, it seems that in fact there is a benefit to shooting some volume of mid-range two's. Correlation is not causation, so I will not make the claim that the example I provided above is confirmation of the answering yes to Patrick's question, but both viewing both together is indicative that a YES answer is at least more true than false.

Sign in to write a comment.