Midrange shots and efficiency

[Dr Positivity]

Here's an idea I've been playing with lately

Midrange shots improve the efficiency of inside and outside shots - the more the d is positioned to stop midrange shots, the less they are at the rim and from 3. Particularly at the rim cause the defender is stretched out. Importantly, the midrange scoring has to put pressure on the defense for this to apply. Teams want Luol Deng to shoot 20 footers so he won't change the defense or help teammates much. Whereas Dirk, Kobe, prime KG are nightmares to guard from midrange and pressure the defense to guard these shots, at the cost of opening up the rim and 3pt line

But midrange shots themselves are not very efficient TS% wise.

So should we trust TS% in judging elite midrange scorers? Or should we give the players a bump for helping other players efficiency + from a 'replacement level' type perspective. If the average midrange shot is far less efficient than from the rim or 3, is there more value in a .60 TS% shot from midrange than at the rim or from 3? Cause it's much harder to have the former. Arguably if you have a midrange, inside, and 3pt scorer all producing at the same efficiency, the midrange scorer has way more impact on his team's midrange efficiency, than the inside guy has on his team's inside efficiency, or the 3pt guy has on his team's 3pt efficiency. The latter two are producing at a norm efficiency for inside and outside shots. But it's the midrange scorer producing at a much higher level than the norm, if he can match inside/outside %s from that range

Perhaps this is why Dirk, arguably the GOAT midrange scorer, has anchored dominant team ORTGs his whole career. I bet all his Mavs teams have elite if not league best midrange efficiency. You put that together with shots inside and from 3 which are always efficient, but especially on one of the a dominant midrange teams, and no surprise the offense is great cause all 3 shots are good.

[kabstah]

I don't buy the theory that mid range shooting opens up the 3 point shot. Generally speaking, the further away from the basket, the lower the percentages are for the shooter but the harder it is for the defense to contest the shot. I could see a situation where defending the 3 point shot results in an open mid range shot instead -- defender closing out hard on a 3 point shooter gets blown by, then the shooter pulls up for a jumper -- but that happens on a relatively tiny amount of plays compared to the number of 3 pointers and mid range jumpers taken.

Should mid range shooters get extra consideration? No, that doesn't make sense. Points are points, no matter how you score them. I don't even think mid range shooting has a strong correlation with OTRG either. The two greatest offensive dynasties in the modern era, the SSoL Suns and the Showtime Lakers, were more about scoring efficiently on the fast break than getting a quantity of mid range shots. The Suns were also an elite 3 point shooting team, coming in around 39-40% collectively.

[Ripp]

Yeah, I don't see how midrange helps 3 point shooting. Helps you get layups and dunks, yes. You could probably test this statistically...see if high team mid-range shooting is a good predictor for 3 point shooting or close shots. Sort of an interesting hypothesis...might want to test it out and see.

[Jimmy76]

Yeah, I don't see how midrange helps 3 point shooting. Helps you get layups and dunks, yes. You could probably test this statistically...see if high team mid-range shooting is a good predictor for 3 point shooting or close shots. Sort of an interesting hypothesis...might want to test it out and see.

I'm not sure that would work as a way to test it since we're only talking about the most elite mid-range players

[Ripp]

I was thinking test it on the team level....use attempts and FG% from midrange as predictors variables for attempts (and then FG%) from close in. Say, download 4 years worth of the 30 teams in the nba from hoopdata.com, see if higher midrange attempts and FG% from midrange leads to higher attempts (or FG%) at the rim. Well, I guess first you'd want to compute an empirical covariance matrix to see whether the variables are related, then try to do some regression to tease out a relationship (if one exists)

Dunno if it would work, but might be interesting to see if something interesting pops out.

[Jimmy76]

I was thinking test it on the team level....use attempts and FG% from midrange as predictors variables for attempts (and then FG%) from close in. Say, download 4 years worth of the 30 teams in the nba from hoopdata.com, see if higher midrange attempts and FG% from midrange leads to higher attempts (or FG%) at the rim. Well, I guess first you'd want to compute an empirical covariance matrix to see whether the variables are related, then try to do some regression to tease out a relationship (if one exists)

Dunno if it would work, but might be interesting to see if something interesting pops out.

Well you're clearly more knowledgable than me on this so if you think that would work do it and post the results

Could you get the correlation confused though with the high at rim fg% leading to the mid-range?

[Ripp]

naw, basically the y-axis variable is high FG% at the rim, the x axis variable would be say FG% from 16-23 feet. Actually, i'd have a lot of x axis variables, and just that single y-axis variable.

The idea is to hopefully find out how increased attempts, efficiency at midrange affects FG% at the rim.

Won't be able to do it anytime within the next few days, so if anyone else here reading this wants to do this, feel free to run with it.

[tsherkin]

It's worth pointing out that Boston tends to game to GIVE UP mid-range shots so that they can guard higher-efficiency baskets around the rim (higher FG%, more likely to get a foul) and from behind the arc (better point pay-off for those going in). San Antonio does a lot of this, too.

You might want to look at correlating how much less likely the offensive team is to rebound a missed mid-range shot compared to a three or a shot in the key, and factor that in with the lower points-per-shot rating that you'll get from the lower DrawF and no opportunity for the 3rd points per basket.

It's great to have the game in that range so that you can always get a good look at the rim, but there's a reason teams game to go inside or shoot the 3 for the most part.

[Ripp]

Could you get the correlation confused though with the high at rim fg% leading to the mid-range?

Sorry, just now realized I misunderstood your remark. Do you mean correlation versus causation? What do you mean?

[mopper8]

It's worth pointing out that Boston tends to game to GIVE UP mid-range shots so that they can guard higher-efficiency baskets around the rim (higher FG%, more likely to get a foul) and from behind the arc (better point pay-off for those going in). San Antonio does a lot of this, too.

I think that's precisely the idea though...a team with good midrange shooting is going to make the strategy difficult to stick with because it will be scoring effectively, forcing the defense to come out and contest and opening up the lane some. No? At least, that's what I think the OP was trying to say.

[Ripp]

I went ahead and started this. Downloaded year-by-year team shooting stats from hoopdata.com, and put them in CSV format:

[2010] https://spreadsheets.google.com/ccc?key ... y=CPX59OQF
[2009] https://spreadsheets.google.com/ccc?key ... y=CN7sg5EC
[2008] https://spreadsheets.google.com/ccc?key ... y=COChh-cD
[2007] https://spreadsheets.google.com/ccc?key ... y=COvp__MI

The goal is to predict FG% at the rim (column G) from eFG% from below 10 FT, 10-15, 16-23, and 3 point range. I think that I should actually also try to use volume (I.e., fraction of shots taken from <10ft, 10-15, 16-23, and 3 point range) as predictor variables, but for now let's focus on raw shooting ability.

I did a regression using the above data, and basically found that eFG% on <10ft shots and 3 pointers are the two best predictors of finishing ability at the rim....performance on midrange jumpers (either 16-23 feet or 10-15 feet) doesn't seem to give a better-fitting model, assuming I already know performance on <10ft shots and 3 pointers.
(Just to be very clear, if I have absolutely no information at all except for performance on midrange jumpers, this is probably going to be better than nothing [i.e., statistically significant]. My point is that if I have performance on 3s and <10 feet, then midrange jumper efficiency is not statistically significant.)

Anyway, here are plots of the results:


I also made a "movie" of the above 3d plot. Unfortunately, it is a gif file and somewhat big (~850kb), so I can only link to it rather than displaying it here:
http://s950.photobucket.com/albums/ad34 ... ovie-1.gif

Finally, the formula produced by the above regression is:

Code: Select all

FG% at rim = 37.8562   + 0.2609 *FG% under 10 Feet + 0.2203 eFG% on 3 pointers


In the above formula, FG% is a number between 0 and 100, and eFG% is also on that scale (i.e., not a number between zero and one, instead a number between 0 and 100).

So roughly speaking, a 1% jump in your team's eFG% from 3 means a 2.2% jump in your FG% at the rim. I'm a bit surprised how powerful 3 point shooting appears to be.

Some caveats:
1) Anyway, as I stated earlier, I should also look at fraction of shots taken from different spots on the floor, and probably also add FTA/FG. Maybe with the addition of those variables, midrange might become statistically significant (or other variables drop out.)
2) It would be much better if this were done game-by-game, rather than season averages for teams..I'd be going from a sample size of 120 to one of nearly 5000. I don't really feel like grabbing this data right now, though...will do it later.

[Jimmy76]

Could you get the correlation confused though with the high at rim fg% leading to the mid-range?

Sorry, just now realized I misunderstood your remark. Do you mean correlation versus causation? What do you mean?

Could high efficiency close range causing high efficiency midrange churn out the same data

And nice work

So to simplify up to this point there is no statistically significant correlation between midrange and at rim but there is clear relationship between 3 point shooting and at rim

[Ripp]

I don't feel completely comfortable with that statement, since I only have 120 data points. Hopefully sometime later this weekend I'll be able to do it on a game-by-game basis (roughly 5K data points points), and see how the other variables (fraction of shots taken from 3, etc) affect things...then I'd feel more confident about making stronger statements.

[Jimmy76]

I don't feel completely comfortable with that statement, since I only have 120 data points. Hopefully sometime later this weekend I'll be able to do it on a game-by-game basis (roughly 5K data points points), and see how the other variables (fraction of shots taken from 3, etc) affect things...then I'd feel more confident about making stronger statements.

Cool and thanks

Damn interesting stuff

[tsherkin]

Neat analysis, Ripp. It does seem to confirm the basic notion that spreading the floor opens up the interior, at least at first pass. Guys are less willing to rotate (or too far away to rotate over in time) down into the paint when they're being drawn out to cover guys with range enough to stick shots for that extra point. Synchs up well with qualitative analysis, too.

I think that's precisely the idea though...a team with good midrange shooting is going to make the strategy difficult to stick with because it will be scoring effectively, forcing the defense to come out and contest and opening up the lane some. No? At least, that's what I think the OP was trying to say.

Sure, but to be honest, I'd let them take their chances. When you're talking about 16-23 foot shots, you're talking about FGAs that GOOD shooters shoot at 40-45% FG and aren't easily rebounded by the offensive team. I'll play those odds all day long against high-payout opportunities at the rim or giving up open looks from 3 that are often at a similar percentage but with a higher scoring payoff.

You can be a fine mid-range shooting team, but you're going to have to shoot the three effectively to set that up anyway, or all the defenders will suck in beneath the arc and crowd the mid-range zones anyway, and the methods guys use to score in those zones require space... either lead-up space where they dribble and drive, clear-out space for triple-threat isos and post-ups or space created by screens... all of which are still dependent upon spacing. If your spacing is borked, the spaces you want to use are packed.

So at that point, if you're ALREADY making use of the high-payout shots, why would you emphasize the lower-reward opportunities?

[Dr Positivity]

Just another idea I was thinking about in a boring class

Let's hypothetically assume 3 teams shoot the same volume of shots from inside (includes FT line trips), midrange, and from 3. And for the purposes of a constant we'll say the 3 zones have the same volume. So the 3 teams all take 20 inside shots, 20 midrange, and 20 3s. These 3 teams each have a star leading them - Player A, Player B, and Player C.

Player A takes all of his team's inside shots at an average inside efficiency (which is extremely high)
Player B takes all of his team's midrange shots at an average midrange efficiency (which is crappy)
Player B takes all of his team's outside shots at an average outside efficiency (which is high)

The rest of the team scores at normal efficiency in the other zones, and again at the same volume for all 3. Player A's efficiency is insane, Player B's is horrible, and Player C's is very good.

All 3 teams end up with the exact same offense. This is despite Player A looking like the GOAT scorer, Player B looking like the GOAT inefficient chucker, with Player C looking pretty great but not Player A - with average teammates across the board.

Now let's account for the variable I ignored. My example assumed Player B's teammates can combine for all the inside points Player A scores and all the 3s Player C does. And likewise, that Player A's team will be forced to take 20 inefficient midrange shots. This can't be assumed. Most players in the league can't consistently create points inside. Great teams don't have to take as many midrange shots. And some teams have no 3pt shooting at all.

Nevertheless I conclude there's two ways to improve your team's offense

1. Improving the volume of your team's inside/outside shots vs midrange, compared to the average
2. Improving the efficiency of either inside, midrange, or outside shots, compared to the average

Going back to the above example, let's replace Player A with Player A-PRIME, who not only scores all his team's inside points, but drastically improves their volume of it compared to mid and 3pt range. Everything else the same with the other two teams, he now anchors the best offense. Now put original Player A back and instead replace Player B with Player B-PRIME, who not only takes all his team's midrange shots, but does so at an efficiency drastically higher than midrange average. If everything else stays the same, Player B's team now has the best offense, despite Player A still having by far the best stats.

I'm guessing almost all players who make a living inside, have their impact in the volume premise, rather than improving inside efficiency which is so high that it's hard to raise (perhaps only Barkley, McHale and maybe some FT hounds like Durant can boost it). Shaq's case is interesting. Shaq's inside efficiency is probably either average or subpar cause of his FT shooting... but it's increasing the volume of inside shots where he brings his impact. Furthermore Shaq also boosted the volume of outside shots with his double teams and passing. With Shaq almost all your shots are inside and outside. Shaq is possibly the ultimate volume shifter

For midrange, a star player's job is to create inside and outside shots, which reduces midrange volume. So when a star decreases mid volume, he's really just increasing the other two. The players that increase their team's midrange volume are trouble, however (duck, AI!*) For the increasing midrange efficiency, I hark back to Dirk. For a long time I've looked at Dirk's run of top 3-5 ORTG every year and just scratched my head. Dirk doesn't really create shots at the rim or from 3 for teammates compared to average. If Shaq is the ultimate volume shifter, guys like Dirk and Melo are at the other end among stars. For a pure scorer I struggled to figure out where this massive impact came from. Maybe this is it. Dirk raises his team's midrange efficiency probably more than any player... ever. He's the GOAT midrange scorer. He's the closest guy we have to B-PRIME. Guys like Kobe and KG aren't quite as efficient as Dirk from midrange but I'd assume still give you more midrange juice than more teams. If Kobe and KG are replacing the average midrange chuck, there'll be some impact. Noteably KG's Wolves had an excellent run of ORTGs despite what felt like crap talent aside from him

Increasing volume and efficiency from 3 is more or less the same thing. If you create more 3pt shots, it probably means defenses are giving you more open shots from there. Usually to increase 3pt shots you break down the defense with penetration or post players who are looking to score inside. So this is very correlated to the inside scorers.

Finally wanted to mention two guys who put up 30ppg at league best efficiency and appeared to have little impact on ORTG: Adrian Dantley and Walt Bellamy. If I'm correct, one reason to explain this is they neither shifted volume or efficiency. I didn't watch Dantley play, but let's say his game is somehow taking away opportunities to score inside by teammates and he's scoring at the same efficiency as most players do inside. Now his only impact is shifting their inside shots to himself... thus achieving no impact. This is just a guess, there's a chance Dantley's supporting casts were just bad, but anyways. A more recent example of a high efficiency, is he really helping? guy is Corey Maggette. If Maggette's ballhandling is restricting teammates from getting good inside/outside shots, he's not helping his team's inside volume, and thus not helping them

* Noteably, Iverson played with no 3pt shooters or guys who could score inside. So at least on those Larry Brown teams he's arguably not raising the volume of midrange shots vs inside - especially considering he scored lots inside. Plausibly he's actually creating a positive inside to mid ratio. With that said, if Iverson played on a more talented inside/outside offensive team and still took that many difficult midrange shots, he'd have had negative impact. So putting defense/rebounding guys with no offensive ability around him was probably a smart move

[Chicago76]

Going back to the above example, let's replace Player A with Player A-PRIME, who not only scores all his team's inside points, but drastically improves their volume of it compared to mid and 3pt range. Everything else the same with the other two teams, he now anchors the best offense. Now put original Player A back and instead replace Player B with Player B-PRIME, who not only takes all his team's midrange shots, but does so at an efficiency drastically higher than midrange average. If everything else stays the same, Player B's team now has the best offense, despite Player A still having by far the best stats.

This is where you lost me. Putting some easy, but not too far from reality numbers out there to test this, assume 20 shots taken from all three locations: close (inside 16), mid-range (16 to the arc) and long (three point). Assume 20 FGA from all three with 55% from close, 45% from mid-range, and 35% from three. Also assume a "excellent" shooter from all three spots would shot 15% better from each location, or 63.25%, 51.75%, 40.25% from close, mid-range, and 3 pt territory.

Isolating average volume/even distribution as a base case (61 pts from 60 FGA) and high efficiency/average volume for each area (3 cases) and high efficiency/high volume from each area (3 more cases), I get the following expected points:

67.2 - excellent volume and efficiency inside
66.2 - excellent volume and efficiency outside
64.3 - average volume and excellent efficiency inside
64.2 - average volume and excellent efficiency outside
63.7 - average volume and excellent efficiency mid-range
63.3 - excellent volume and excellent efficiency mid-range
61.0- average distribution and efficiency

What types of percentages are you assuming when you make the statement above?

When you look at your analysis this weekend, one of the things you might want to consider w/ respect to mid-range shooting is pace. I don't have anything to offer empirically, but from an observation standpoint--

Teams that run more seem to have more open looks from mid-range and this increases their overall mid-range FG% enough to make this a more efficient alternative. If teams play slowly, the defense has time to set up and cover the floor better. In these cases, all shots are contested better, and the mid-range jumper loses enough fg% to lose out over the 3.

Also, you have some very interesting observations about Dantley that I hadn't really thought as succinctly as you laid out. Since you mentioned you never really saw him play, Dantley was the master of finding room 10-16 feet from the rim to work his magic. Incredibly efficient, but he could take something away from the offense. A guy like Thomas relied upon freedom to run the show and a mid-range shot where he needed constant motion and freedom to drive the line to shoot over and around bigger players. If you have a guy like Dantley playing fairly close to the basket, that can clog things up and slow things down too much for a guy like Thomas. The Pistons were built around Thomas/Tripucka/Laimbeer offensively, and Laimbeer could play a more perimeter role as needed at C to keep things open for Thomas.

Looking at Ortg impact before and after he arrive at the Jazz, there was a lot of and some low efficiency, high volume guys left like haywood and Maravich the year before, but still, there was a pretty big difference. Also, when he left, the Jazz didn't really improve, despite Malone and Stockton growing as players. Those two improving enough to offset Dantley at that stage in their career shouldn't be a knock against him.

Looking at the same impact before/after arrival at Detroit: Isiah's TS% took a hit, but everyone else stayed roughly the same. Dantley replaced Tripucka, who was a pretty efficient scorer himself, but he operated more on the perimeter, which gave Thomas more freedom. When Dantley left, not much change, but consider that Dumars was a slow developer and Aguirre was pretty efficient himself. It's kind of strange, because Aguirre and Dantley had similar games. Short, lots of twisting and turning in the lane to free themselves enough to score on bigger players. Oh yeah, and both had really huge asses they used to back guys down. It sounds ridiculous, but trust me, ask a guy who watched a lot of the NBA in the mid 80s about Aguirre or Dantley and some derivation of "huge ass" will get mentioned inside of 1 minute. Guaranteed. I wish Thomas would have figured out a way to adjust his game to Dantley a little better, because I felt that team w/ Dantley may have had 3 titles in them.

[kabstah]

Going back to the above example, let's replace Player A with Player A-PRIME, who not only scores all his team's inside points, but drastically improves their volume of it compared to mid and 3pt range. Everything else the same with the other two teams, he now anchors the best offense. Now put original Player A back and instead replace Player B with Player B-PRIME, who not only takes all his team's midrange shots, but does so at an efficiency drastically higher than midrange average. If everything else stays the same, Player B's team now has the best offense, despite Player A still having by far the best stats.

No, it doesn't work like that. You don't have enough data, empirical or hypothetical, to make such a statement. Player A-PRIME makes his team eschew lower percentage shots in favor or higher percentage shots, whereas player B-PRIME makes his team convert lower percentage shots at a higher percentage than normal. There's nothing, however, that would allow you to reasonably conclude that Player B-PRIME's team has the best offense.

Consider this scenario (ignoring 3's for now, since we're only talking about players A and B):
Everyone shoots 100% from inside.
Player A-PRIME forces his team to shoot only inside shots, which they convert at 100%.
Player B-PRIME forces his team to shoot an even distribution of inside / mid range shots.
Player B-PRIME's teammates still convert inside shots 100% of the time.
Player B-PRIME takes every mid range shot himself, which he converts 100% of the time.
Net result: Both A-PRIME and B-PRIME's teams convert every shot, meaning there is no difference in team offense.

You can tweak the numbers to make either A-PRIME or B-PRIME come out on top, but it's far from a decisive advantage for B-PRIME as you're making it seem. Furthermore, with the advent of the 3 point line, the mid range shot is becoming increasingly unimportant. The best offensive system in recent years are the Suns, who are built around Steve Nash's playmaking. According to 82games assist tracking, what Nash does well is create a lot of opportunities for either 3 pointers and inside shots/dunks.

[Dr Positivity]

I'm not comparing A-PRIME and B-PRIME against each other. A-PRIME is compared to original Player B, and B-PRIME compared to original Player A. My 3 comparisons, similarly ignoring the 3pt for now

Scenario 1
Player A
Player B

Scenario 2
Player A-PRIME
Player B

Scenario 3
Player A
Player B-PRIME

In Scenario 1 the offenses are equal and our 'standard' level. In Scenario 2 Player A-PRIME makes his team take more inside shots compared to midrange, thus giving his team a superior offense. In Scenario 3 Player B-PRIME raises the efficiency of midrange, again raising it above the standard.

[Dr Positivity]

Going back to the above example, let's replace Player A with Player A-PRIME, who not only scores all his team's inside points, but drastically improves their volume of it compared to mid and 3pt range. Everything else the same with the other two teams, he now anchors the best offense. Now put original Player A back and instead replace Player B with Player B-PRIME, who not only takes all his team's midrange shots, but does so at an efficiency drastically higher than midrange average. If everything else stays the same, Player B's team now has the best offense, despite Player A still having by far the best stats.

This is where you lost me. Putting some easy, but not too far from reality numbers out there to test this, assume 20 shots taken from all three locations: close (inside 16), mid-range (16 to the arc) and long (three point). Assume 20 FGA from all three with 55% from close, 45% from mid-range, and 35% from three. Also assume a "excellent" shooter from all three spots would shot 15% better from each location, or 63.25%, 51.75%, 40.25% from close, mid-range, and 3 pt territory.

Isolating average volume/even distribution as a base case (61 pts from 60 FGA) and high efficiency/average volume for each area (3 cases) and high efficiency/high volume from each area (3 more cases), I get the following expected points:

67.2 - excellent volume and efficiency inside
66.2 - excellent volume and efficiency outside
64.3 - average volume and excellent efficiency inside
64.2 - average volume and excellent efficiency outside
63.7 - average volume and excellent efficiency mid-range
63.3 - excellent volume and excellent efficiency mid-range
61.0- average distribution and efficiency

What types of percentages are you assuming when you make the statement above?

For the purposes of simplicity, I assumed no matter the particular %s, both teams led by original Player A and Player B score at a 'regular' efficiency. Player A shoots team's 20 inside shots at a regular inside efficiency (which is very high), and his teammates shoots their 20 midrange shots at a regular midrange efficiency (which is much lower). Player B is the inverse - He shoots his 20 midrange shots at the same regular midrange efficiency as Player A's teammates, and then his teammates shoot the 20 inside shots at the same efficiency Player A had inside

Now in Scenario 2, Player A-PRIME's team is compared to standard Player B's. Player A-PRIME's team takes more inside shots than midrange, let's say 25 inside and 15 midrange. Player B's has the same 20/20. Even though the inside and midrange efficiencies are the same, Player A-PRIME's team offense is better than standard Player B's cause he has a better ratio of efficient to inefficient shots

In Scenario 3, standard Player A is back compared to Player B-PRIME. Both have the 20/20 inside and midrange split. Their inside scoring is identical - despite Player A being the dominant inside scorer, Player B-PRIME's teammates combine for the same efficiency on those 20 shots inside. Both B-PRIME and standard Player A's teammates take 20 midrange shots, but B-PRIME's team converts at a higher efficiency. Thus B-PRIME's team offense is > standard Player A's. The inside volume and efficiency is identical, but B-PRIME converts them at a higher % than Player A's teammates. However Player A still looks like the better player on paper, since he is shooting the same volume of shots as B-PRIME, but at a much higher efficiency since all his shots come inside. Nevertheless if standard Player A is neither altering his team's inside volume or efficiency compared to if he wasn't there, he's not changing his team above the standard - Whereas B-PRIME improves his team's midrange efficiency above standard, and thus improves the offense

(As mentioned though it's probably easier to get stock teammates who can replace your midrange shots, than ones who can replace an inside scorer. But for now this is hypothetical)