Should defenses let long 2s go uncontested?

[Spicysheep]

Statistically speaking, long 2s are the worst shot in basketball (see sources 1,2, and 3 below). I can't think of a better way for a defense to encourage long 2s than to allow them to go uncontested. Is that a bad idea?

I read a study (source 4) that says that good defense on the shooter reduces his eFG% by 12%. Across the league the average eFG% is about 38% on long 2s. For shots that are not 3s or at the rim, the average is about 40%. Using some rough math, an uncontested long 2 would have 38% + 12% = 50% chance of going in, which would make it the third most efficient shot after shots at the rim (64% eFG%) and 3s (53% eFG%). Given these numbers, would it be a bad idea to encourage uncontested long 2s?

Long 2s have the worst eFG%
1. http://www.hoopdata.com/teamshotlocs2.aspx
2. http://www.82games.com/locations.htm
3. http://courtvisionanalytics.com/the-lon ... osh-smith/

Good defense reduces eFG% by 12%
4. http://www.slideshare.net/sloansportsco ... l-shooting

[xkonk]

One of the slides in the fourth link mentions that an open 19 foot jumper is (according to their model) the same as a loosely contested three or a highly contested lay-up. Presumably you would like any shot to be at least somewhat contested, but assuming that you really want to cut down on lay-ups and threes as the typically most efficient shot, a team should be willing to encourage long jumpers to the point of leaving them open on some occasions. It would also obviously depend on who you're leaving open.

[EvanZ]

Against most teams, probably. There are only a few players across the league (e.g. Dirk, Bosh, Curry) who, if left truly uncontested from 20 ft, could probably hit 55-60% of those shots every night. You wouldn't want to let that happen all night long.

[JonCBK]

Truly uncontested shots would go in at a far higher rate than +12% is my guess. Also it depends on who you are leaving open. And the suggested players above are not the only ones you have to worry about. Garnett and Dirk are able to shoot that shot in a game and be great at it mainly because of their height. In reality their release point and the relatively less mobile players guarding them means that they are generally uncontested a lot even in games where the other team is trying the best to contest the shot. But lots of shorter players in the league have far better jumpers.

Basically all of the guards could hit 17 - 20 foot jumpers at well over 50% if truly uncontested. But to be uncontested at 6' 2" in the middle of an NBA game probably requires you to have done a very fast move or for a play to have thoroughly broken down the opposing defense. The fast move results in a harder jump shot (almost no one can really dribble fast forward and then pull up quickly for a long jumper). And total break downs on D don't happen too often, especially not on the guard defenders who are less likely to be tasked with rotating on D back to the rim to protect it. If if you really stopped contesting Nate Robinson's 20 foot jumper, he would start hitting that at easily over 50% clip. Watch warm ups for a hint at uncontested shots. Everyone shoots over 50% and they aren't even trying.

[Jacob Frankel]

It depends. If not contesting long 2s is a way to bait the opponent into taking long 2s, then I don't think the strategy works. Letting them go uncontested takes away the inefficiency of long 2s so baiting the opponent into them is pointless. If it can strengthen up another part of the D, maybe having the big men not go as far from the basket or having less help off of three pointers, then maybe it is worth it.

[mbosset]

One quick note about your math. You add 12% to the 38% as a result of the shots being uncontested. However, at least some of the shots that account for the 38% would be uncontested. Your equation would be correct assuming that currently no long range jumpers are uncontested, which is not quite the case.

[Notsellingjeans]

There was a great YouTube clip a few years back of Gilbert Arenas making 81 out of 100 threes from the corner after practice. The Spurs in the Steve Kerr/Brent Barry had similar competitions where the winner broke 80. People who play basketball for a living are shockingly good when unguarded. Someone higher up in the thread hit the nail on the head here. NBA guards are going to hit 60+ percent of their uncontested long twos - high enough that the points per possession yield of this strategy would be defensively untenable at the pro level.

But high school and college? Absolutely,IMO. Most elite athletes haven't shot enough in their lifetime yet to burn you. John Wall didn't win a national title because he couldn't punish this strategy yet when he was 18-19 years old.

[knarsu3]

It depends. If not contesting long 2s is a way to bait the opponent into taking long 2s, then I don't think the strategy works. Letting them go uncontested takes away the inefficiency of long 2s so baiting the opponent into them is pointless. If it can strengthen up another part of the D, maybe having the big men not go as far from the basket or having less help off of three pointers, then maybe it is worth it.

I've actually been doing an analysis of this (well a general analysis on shot location by shot defense) using Vantage's dataset. And I think you'd be surprised at how inefficient even open 2s are (Vantage defines it as no defender within 5 feet). Vantage also tracks contested shots (defined as getting your hand up and being within 3 feet) and I've found that contested 3s and contested shots near the basket have a better pay off than taking a wide open mid-range 2. So I do agree with the overall premise for a league wide philosophy. Obviously there may be some players (and I haven't looked at this yet) where that is not true. I only found about a 5% difference in open 2s vs. contested 3s/contested layups (accounting for the fact the 3 is worth an extra point). In general my study probably aligns pretty well with the source 4 study for 3 point shots but where getting a hand up and contesting really matters is near the basket.

[Jacob Frankel]

I've actually been doing an analysis of this (well a general analysis on shot location by shot defense) using Vantage's dataset. And I think you'd be surprised at how inefficient even open 2s are (Vantage defines it as no defender within 5 feet). Vantage also tracks contested shots (defined as getting your hand up and being within 3 feet) and I've found that contested 3s and contested shots near the basket have a better pay off than taking a wide open mid-range 2. So I do agree with the overall premise for a league wide philosophy. Obviously there may be some players (and I haven't looked at this yet) where that is not true. I only found about a 5% difference in open 2s vs. contested 3s/contested layups (accounting for the fact the 3 is worth an extra point). In general my study probably aligns pretty well with the source 4 study for 3 point shots but where getting a hand up and contesting really matters is near the basket.

Very interesting. The Vantage dataset has huge potential for teaching us a lot about the game. I'd guess some teams already do the not-contesting-long-twos thing on a situational basis. I'd imagine some forward thinking defense like Indiana won't contest midrange shots when they play Detroit this season.

[Mike G]

Shouldn't you also control for time on the shot clock?
If everyone knows the long 2 is the worst shot, then it might be over-represented late in the shot clock -- just because it's the last available shot, because that's what the defense is giving you?

In the first 5 seconds of the shot clock, of course a long 2 is a terrible shot. In the last 5 seconds, it's a whole lot better to take a 40% FGA, with a chance at an offensive rebound, than to get no shot at all.

[knarsu3]

Shouldn't you also control for time on the shot clock?
If everyone knows the long 2 is the worst shot, then it might be over-represented late in the shot clock -- just because it's the last available shot, because that's what the defense is giving you?

In the first 5 seconds of the shot clock, of course a long 2 is a terrible shot. In the last 5 seconds, it's a whole lot better to take a 40% FGA, with a chance at an offensive rebound, than to get no shot at all.

I've started out looking at the baseline averages at each location but that's actually my next project: to account for the shot clock as well as the position of both the shooter and defender. Also going to look at long 2s off 0 dribbles (catch and shoot) vs. 1+ dribble to see if that makes a difference.

It's really quite wonderful looking at all the possibilities with Vantage's dataset. If you guys haven't checked out their blog yet, I suggest you do: http://blog.cacvantage.com/

[Dr Positivity]

Not sure how that 12% is calculated. Does it mean +/- 12% from average, for a total range of 24% between contested and uncontested? Or 12% from contested to uncontested? I’ll use the +/- 12% from average for this post as it seems more likely

Here is the eFGs for average, contested and uncontested for 2s/3s/at rim

Contested 2 - .26 eFG
Avg 2 - .38 eFG
Contested 3 - .41 eFG
Uncontested 2 - .50 eFG
Contested at rim .52 eFG
Avg 3 - .53 eFG
Avg at rim - .64 eFG
Uncontested 3 - .65 eFG
Uncontested at rim - .76 eFG (clearly grain of salt number)

In short, you shouldn’t give up the uncontested 2 because it’s still possible to force a contested 2, 3 - or a TOV, blocked shot, etc.

It’d work if the opponent took an uncontested 2 every time it was open. However if a rational agent the opponent will try for an at rim/3 attempt before them, as they’re more efficient.

So the result of not contested 2s, may be that they end up with a certain amount of x open 3s/at rim attempts, plus y uncontested 2s. The uncontested 2 is a much better ‘worst case scenario’ shot than a contested 3 or contested 2 in present defenses. So it could lead to a much worse defense. For it to work, the opponent’s amount of at rim/3 attempts would need to be reduced enough to make up for forcing less contested jumpers on them. While one could defend at rim/from 3 better if they’re not defending 2s and there may be ‘bait-taking’, I have my doubts about it being worth it. Or to put it another way, as defenses are imperfect they’re going to allow a certain amount of shots far better than league average eFG, when they give up shots from at rim/FT line/open 3s. To counteract this, you need shots far worse than league average eFG, contested 2s or 3s - hopefully bringing the total down to below league average eFG. Allowing uncontested 2s is giving up this powerful weapon of contested jumpers, leaving nothing to counteract the high efficiency shots the opponent automatically will get some of.

Guess one could argue, for leaving open 2s early in the shotclock. Then as the shotclock goes down and it becomes less likely the opponent will create a shot at rim/open 3, then you lock down uncontested 2s too. But you don’t want to leave players who are 42%+ midrange shooters (+12% = .54 eFG), if they’re that good of shooters you’re better off taking your chances with their offensive set. I think this is more or less what happens anyways. The good midrange shooters are covered from 2 early in the shotclock, but lots are left open.

2s are tricky. They look so stupid from an overall efficiency perspective, but it’s all relative. Uncontested 2 looks great if the alternative is contested 2/3, TOV, or not getting a shot off. And yes while ‘getting a shot off’ is insanely overrated, it still exists. You can’t just take a shot at ANY time. The defender will get a hand on jumpshots if he’s close enough. So even with contested 2s, if the alternative is the shot getting blocked or a shot clock violation, that justifies taking them. Contested 2s are still taken too much for other reasons, but even in a perfectly rational NBA, they still get taken.

[Mike G]

What about FT?
They're still the best shots. How do you get them? By shooting 3's, layups, earlier on the clock?

[knarsu3]

Not sure how that 12% is calculated. Does it mean +/- 12% from average, for a total range of 24% between contested and uncontested? Or 12% from contested to uncontested? I’ll use the +/- 12% from average for this post as it seems more likely

Here is the eFGs for average, contested and uncontested for 2s/3s/at rim

Contested 2 - .26 eFG
Avg 2 - .38 eFG
Contested 3 - .41 eFG
Uncontested 2 - .50 eFG
Contested at rim .52 eFG
Avg 3 - .53 eFG
Avg at rim - .64 eFG
Uncontested 3 - .65 eFG
Uncontested at rim - .76 eFG (clearly grain of salt number)

In short, you shouldn’t give up the uncontested 2 because it’s still possible to force a contested 2, 3 - or a TOV, blocked shot, etc.

It’d work if the opponent took an uncontested 2 every time it was open. However if a rational agent the opponent will try for an at rim/3 attempt before them, as they’re more efficient.

So the result of not contested 2s, may be that they end up with a certain amount of x open 3s/at rim attempts, plus y uncontested 2s. The uncontested 2 is a much better ‘worst case scenario’ shot than a contested 3 or contested 2 in present defenses. So it could lead to a much worse defense. For it to work, the opponent’s amount of at rim/3 attempts would need to be reduced enough to make up for forcing less contested jumpers on them. While one could defend at rim/from 3 better if they’re not defending 2s and there may be ‘bait-taking’, I have my doubts about it being worth it. Or to put it another way, as defenses are imperfect they’re going to allow a certain amount of shots far better than league average eFG, when they give up shots from at rim/FT line/open 3s. To counteract this, you need shots far worse than league average eFG, contested 2s or 3s - hopefully bringing the total down to below league average eFG. Allowing uncontested 2s is giving up this powerful weapon of contested jumpers, leaving nothing to counteract the high efficiency shots the opponent automatically will get some of.

Guess one could argue, for leaving open 2s early in the shotclock. Then as the shotclock goes down and it becomes less likely the opponent will create a shot at rim/open 3, then you lock down uncontested 2s too. But you don’t want to leave players who are 42%+ midrange shooters (+12% = .54 eFG), if they’re that good of shooters you’re better off taking your chances with their offensive set. I think this is more or less what happens anyways. The good midrange shooters are covered from 2 early in the shotclock, but lots are left open.

2s are tricky. They look so stupid from an overall efficiency perspective, but it’s all relative. Uncontested 2 looks great if the alternative is contested 2/3, TOV, or not getting a shot off. And yes while ‘getting a shot off’ is insanely overrated, it still exists. You can’t just take a shot at ANY time. The defender will get a hand on jumpshots if he’s close enough. So even with contested 2s, if the alternative is the shot getting blocked or a shot clock violation, that justifies taking them. Contested 2s are still taken too much for other reasons, but even in a perfectly rational NBA, they still get taken.

how are you defining contested? is this the literal definition or SportVu's definition (which isn't actually contested)

[knarsu3]

Shouldn't you also control for time on the shot clock?
If everyone knows the long 2 is the worst shot, then it might be over-represented late in the shot clock -- just because it's the last available shot, because that's what the defense is giving you?

In the first 5 seconds of the shot clock, of course a long 2 is a terrible shot. In the last 5 seconds, it's a whole lot better to take a 40% FGA, with a chance at an offensive rebound, than to get no shot at all.

Just got the data for the shot clock, there's plenty of long 2s taken with 10 or more seconds left and the FG% is roughly around the same (i.e. worse than contested shots near the basket and contested 3s).

It does ultimately depend on the player though because having looked at the data for a player who is a really good mid-range shooter, his open mid-range % is definitely higher than taking a contested 3 or a contested shot near the basket. So I think it depends on who you leave wide open on mid-range shots. The league in general won't shoot well and is worth leaving open but a superb mid-range shooter is not worth leaving wide open. Edit: I haven't backed up this 2nd part with enough of a sample but just generally looking