other twos are not all the same [Marc-Gregor Campredon]

The Murky World Of Other Twos Comment Count

Brian February 24th, 2021 at 12:10 PM

I read this deep dive into the problems with the Tennessee offense yesterday that mostly dealt with the fact that Rick Barnes is unusually enthusiastic about midrange shots, a distinction that is only growing deeper as more college offenses are analytics inflected. Then I got hung up on something. Here's a high-level view of the midrange across college basketball:

In 2010-11, the average top 25 shooting offense got nearly 30% of their shot attempts from non-rim two-pointers. In 2020-21, this number has fallen all the way to 20.9%, and it hasn’t cracked 25% since 2015-16. Yet Tennessee is still taking the same number of mid-range jumpers that they were taking in 2015-16. …

No team in the current top 100 of eFG% in 2020-21 gets more than 33.7% of their shots from non-rim twos, which is a full 2.3% less than Tennessee. In fact, only six of the top 50 even crack 30%, which was the average just ten years ago. Among the current top 25 in eFG%, no team is anywhere near Tennessee’s 324 mid-range jumpers on the season. Only one team (Campbell) even cracks 200 mid-range jump shot attempts on the season. Lastly, the expected Points Per Possession on a two-point jumper from 8-14 feet is roughly 0.732 (36.6% FG%) and 0.704 (35.2%) from 15-20 feet. An average team would have to shoot 24.4% or worse on threes for these shots to technically be advantageous. Tennessee makes 34.4% of theirs.

I looked up Michigan's standing in this department and found a huge discrepancy in my usual data sources. I wish this post had disambiguated where this data came from because there's a pretty big gap between what Synergy considers "at the rim" and what box score play-by-play—which is the source of Torvik's shooting split numbers—does. By Torvik's reckoning Michigan is a relatively heavy midrange team with about 31% of their shots there, which places just inside the top 100 nationally.

[After THE JUMP: Other twos are a big tent]

Synergy has a completely different take. They bin jumpers into three categories: short, long, and 3s. Here's Michigan' shooting chart* on all two-point jumpers to give you a picture of what is and isn't included:

image

Runners, floaters, and Wagner gumby layup attempts are not included. Neither are short hooks from post players.

So while the above might look like a large number of midrange jumpers, this is only 94 shots. (Tennessee, for its part, has 204 two point jumpers in four more games than M.) The box scores Torvik relies up on report a whopping 298(!) other twos. The gap here is tremendous.

About half of the gap consists of post hooks and runners/floaters. None of those are included in the above chart and Michigan has 116 of those on the season. 58 of those are fabulously efficient post hooks, which Michigan is hitting at a 59% clip. 58 of them are grim floaters/runners, which Michigan is hitting at a 33% clip.

That still leaves us with 88 "other twos" unaccounted for. And here we get into some ambiguous waters. Michigan has 403 tips/layups/dunks on the season, per Synergy, and 339 attempts at the rim per Torvik. Assuming that nobody in their right mind is going to classify a dunk or putback as an "other two," this means 64 layup attempts that Synergy puts in an at-rim bucket are other twos to PBP. The remaining 22 attempts are a mystery.

A few upshots:

The "other twos" bin contains some vastly different shots. It's got 42 post hooks from Hunter Dickinson and Austin Davis that are going down at a 67% clip. It's also got 58 floaters, a shot that even NBA players eke out just 0.86 PPP from. Also there are 60-80 layup attempts in there. Those are probably the worst layup attempts Michigan gets but even so I'd be willing to bet that those are good shots relative to jumpers and floaters, especially when you account for the Kobe assist factor.

Rim/away/three is are rough boxes indeed. That doesn't mean they're not useful; it does mean that you should sanity check numbers if they don't line up with your eye test. Speaking of…

Michigan is not a heavy, or even average, jumpshooting team. Michigan is towards the bottom of the league in two point jumpers attempted:

Team 2PJ FGA 2PJ/FGA eFG
Nebraska 79 1255 6.3% 34.2
Purdue 111 1323 8.4% 36.9
Michigan 94 980 9.6% 39.4
Iowa 145 1470 9.9% 39.3
Indiana 129 1236 10.4% 33.3
Illinois 137 1279 10.7% 43.1
Ohio State 166 1313 12.6% 41
Maryland 168 1287 13.1% 38.1
Minnesota 185 1406 13.2% 39.5
Wisconsin 208 1376 15.1% 37.5
Rutgers 194 1261 15.4% 42.3
Northwestern 189 1187 15.9% 35.4
Penn State 216 1252 17.3% 38
Michigan State 219 1252 17.5% 38.8

This is probably a more accurate rim/away from rim take than those derived from box score PBP. And this fits much more in line with my personal observations of Michigan basketball, where the Smith or Livers pull-up is a relative rarity.

Two point jumpers might be good shots for particular guys in particular situations but they're never good overall. Illinois has the best performance here and they're at 0.86 PPP. That understates the actual PPP because there's always the chance of an OREB; it's telling that there's no real correlation between how many two point jumpers you take and how good you are at them.

Lots of two point jumpers seems to put a ceiling on your offense. The six teams that take the most 2PJ/FGA are 7th, 9th, 10th, 11th, 13th, and 14th in conference eFG. Penn State creeps up to 7th in offensive efficiency because they pound the boards harder than anyone else in conference play. Nobody else of the bottom six is appreciably above average in either OREB rate or FT rate in league play—there are a couple of sixth-place finishes and a lot of red.

This isn't all because of the jumper rate; rather a high rate of jumpers implies that you're not great at getting to spots on the floor where fouls happen—and even when you do you have a lesser chance to have a positional edge that leads to calls. It implies that you're not great at finding shots with high OREB chances, which usually means involving help defense.

The flipside is not necessarily a guarantee of success. Nebraska, the 2PJ avoidance king, is also horrible at offense.

*[I posted a similar Synergy chart in the OSU game column and noticed that a couple people had questions about the coloring. Red is more efficient than the NCAA average at a particular spot; blue is less efficient. I find the desire to color these at all to be farcical since each bin has a few shots at most.]

Comments

stephenrjking

February 24th, 2021 at 12:22 PM ^

There are different kinds of "other twos" and, by my watching, there are also different kinds of "2 pt jumpers." A 15-foot jump shot with a guy in your face is a terrible shot, and Michigan doesn't take those. They do seem to take: 1. open 2-point jump shots where the defender has sunk and they have open space to pull up and get a clear shot; top of the key shots on ballscreens, that sort of thing (which, btw, appear to be those shots shaded orange in the diagram); 2. Livers fallaways from under 10 feet that look bad and seem to go in a lot.

Looking at the diagram, there are clearly a number of shots along the baseline on the right side, a few of which I remember but seems more numerous than I recall.

By my casual viewing there seem to be some 2-point jumpers that Juwan is ok with the guys taking, because they take them a fair amount. I would guess that the PPP for those is a bit better than some of the bad stuff, including those Eli Brooks floaters that just make me cringe. 

Ali G Bomaye

February 24th, 2021 at 8:42 PM ^

It's also worth considering the second-order effects of taking certain 2PJs. If a player blows by a defender at the arc and goes to the rim every time, the help defender knows to hang back. But if the player takes short jumpers sometimes, it could make him more efficient at the rim too because the defender doesn't know where he will shoot. 

Naked Bootlegger

February 24th, 2021 at 12:31 PM ^

This is a deep-take nitpick, but I wish some players would eliminate floaters from their arsenal.  It's an incredibly difficult shot since the player is typically moving forward and requires an adept touch to adjust for that momentum moving toward the rim.   And once you launch off one foot, you're stuck.  It's either shoot or pass before you come back to earth, lest you get whistled for a travel.   The solution?   A hard two foot jump stop alleviates that predicament and allows you to shoot in a more controlled manner, plus retains the ability to pump fake, pivot, etc.   Many more options.    

Now excuse me while I gear up for my next YMCA open gym hoops stint.   My two foot jump stop repertoire will be on full display, unless I pull a hamstring.

Naked Bootlegger

February 24th, 2021 at 1:04 PM ^

Yes, Eli does frequently employ the floater.   He's more proficient at it compared to previous years, but it's still such a tough shot.   But other players frequently launch off one foot amidst traffic and get stuck with the decision to shoot a highly contested floater-type shot or pass.   Franz has delivered a few nifty passes off one-foot launch floater moves, but has also been whistled for charges and gotten stuck in no-mans-land under similar conditions.   

I'm just a strong advocate for two-foot jump stops amidst heavy traffic since you can still typically pump fake defenders and pivot, but also retain your ability to shoot under control versus the one-footed floater.   OK...gotta run.   Craig Ross just challenged me to a 1 v 1 game.  First to 11 points wins.

AC1997

February 24th, 2021 at 2:14 PM ^

I think the floater in general is a bad shot for all of the reasons people have mentioned here.  Also because it isn't as easy to practice because they're all so random.  However, I do think that's part of the equation with this entire conversation - practice.  Watch a team warm up - they typically practice layups of different varieties and outside jump shots from deep.  No one practices a mid range jump shot as often in this age (for obvious reasons) and even less often would anyone intentionally practice floaters.  

I do, however, think it COULD be a good shot.  Xavier Simpson turned a true hook shot into an effective and efficient weapon despite being 6-feet tall.  How?  He practiced the crap out of it!  

To me, the key to a floater in the lane is to use the backboard (which no one does anymore).  Your momentum is going to be different every time and thus let the backboard even out some of the variables.  

Naked Bootlegger

February 24th, 2021 at 1:50 PM ^

True.   The floater can be a crafty shot that allows shorter players to get shots off in the lane.   

Regarding the Kobe assist potential: this is based on the premise that an off-ball defender slides over to help on the shooter and typically leaves his feet and tries to block the shot, thus creating an easier offensive rebounding opportunity for the player left open by the helping defender.   I would argue that a two foot jump stop, pump fake, and pass would be just as effective and create a higher percentage shot, mostly since a Kobe assist involves some degree of uncertainty and random chance (where the ball will bounce if it hits the rim?).  

theytookourjobs

February 24th, 2021 at 12:33 PM ^

If you watch enough big ten hoops, (excluding the once in a generation M vs. OSU game), you'll begin to see that contested mid range jumpers are nearly the same as a turnover.  They rarely go in, and often lead to long defensive rebounds which in turn lead to fast break opportunities.

teamteamteam

February 24th, 2021 at 12:33 PM ^

Great post and love the analytics.  I worry though you may be mixing up your cause and effect.  Isn't it likely that having a bad offense leads to more 2PJ rather than 2PJs leading to bad offenses?

maquih

February 24th, 2021 at 12:57 PM ^

I'm not sure Brian ever implies a causal relationship.  But yes, I think that's something you need basketball analysts/coaches for -- watching the game and understanding whether the players are looking for 2PJS too much, or if they're struggling to get efficient shots and getting forced to settle for them.

There's also teams that have success looking for 2PJS, I think in the NBA there's always a couple of playoff teams that thrive off of taking those shots that NBA defenses allow because they're so focused on denying paint points and threes.

Joby

February 24th, 2021 at 1:36 PM ^

I think Brian addressed (at least in part) the potential mixing of cause and effect near the end of the post:

___
This isn't all because of the jumper rate; rather a high rate of jumpers implies that you're not great at getting to spots on the floor where fouls happen—and even when you do you have a lesser chance to have a positional edge that leads to calls. It implies that you're not great at finding shots with high OREB chances, which usually means involving help defense.
—-

SHub'68

February 25th, 2021 at 1:32 AM ^

I think we may be overanalyzing this. It's a mix of player skill, running offense, and coaching/scheming. If your team isn't very good, good teams are likely going to be able to force you to take bad shots in the areas they want you to take them in. Good teams are going to be better at forcing other teams to take bad shots while getting good ones themselves...

Michigan State probably has a high rate of these this season because they have been bad at basketball (until recently!) and probably not because Izzo planned it this way.

Mitch Cumstein

February 24th, 2021 at 1:11 PM ^

WRT the bolded statement “Lots of two point jumpers seems to put a ceiling on your offense.” I’m curious about causation.

Is there data around distribution of “other 2s” as a function of time in the possession for a given team? I think there is a lot of “this coach doesn’t embrace analytics and has his teams shooting low % shots” type criticism (and in some cases it might be true and warranted). In some cases though, it might be that bad offensive teams can’t generate statistically better opportunities early in the shot clock and need to expand what’s considered a “good shot” as the possession goes on.

Alabama is the antithesis of this where they’ll take some questionable shots early in the shot clock as long as they’re 3s or layups (even if guarded or on the move, etc). Seems to have worked out for them.

EDIT: teamteamteam asks essentially the same Q above.

SHub'68

February 25th, 2021 at 1:41 AM ^

Yeah, it's really: the data indicates that if you take too many 2PJs, you have a ceiling on your offense. Why is an important question.

  • "Are you putting a ceiling on your offense intentionally because you scheme for 2PJs at too great a rate?"
  • "Do you have a ceiling on your offense because your players aren't good?"
  • "Do you have a ceiling on your offense because your coaching isn't good (scheme/teaching/recruiting)?"

 

 

mfan_in_ohio

February 24th, 2021 at 1:31 PM ^

1. "Gers?" Ugh.  

2. That's just the eFG on the 2 point jumpers.  Their offense is #61 on KenPom, which isn't great (11th-best in the B1G), but it puts them between Kansas and Tennessee, so when paired with a strong defense it makes them a tournament team, albeit one that plays some ugly games.  

Kilgore Trout

February 24th, 2021 at 1:27 PM ^

"Two point jumpers might be good shots for particular guys in particular situations but they're never good overall."

I mostly agree with this, but I think there is some nuance to it. The ability to hit a midrange jumper allows you to get better shots at the rim and for three. If you refuse to take a midrange jumper, you will get blitzed at the three point line and very saggy drop coverage that won't allow you to get a good look at the rim. I think relying heavily on two point jumpers is a bad idea, but the Alabama extreme of only taking threes and shots at the rim has to have a ceiling, too. 

If an average three pointer is worth 1.14 points per shot (38% make rate) then you need to find two pointers that you can make at 57% to be equal. I think there is space in the game to find two point jumpers that can be made at a 57% clip (watch tape of Rip Hamilton) and I would argue you don't have to be equal to make it worth taking them because NEVER taking them will negatively affect your ability to hit threes and shots at the rim. 

mfan_in_ohio

February 24th, 2021 at 1:51 PM ^

I agree with this.  Certain players will make certain 2 point jumpers a part of their game because it can help the rest of the offensive plays, and they may be shots that the defense gives you.  I bet Zak Irvin ended up with a pretty good percentage on his 18-ft jumpers at the top of the key.  That shot seemed near-automatic late in his career, even when his 3pt shot was broken. The good thing about that shot is he practiced it, so he was always shooting from about the same distance.  I think 3s go in at a fairly high rate because they're all from about the same distance, so the muscle memory on that shot is strong.  Conversely, 2pt jumpers are shot from a variety of different distances, and there are differences in both launch velocity and angle.  

Think of it like a pitcher throwing a changeup.  There are a few pitchers that can throw an excellent changeup, and pair it with a strong fastball to become outstanding pitchers (Trevor Hoffman, for example).  For others, yeah, it's maybe his 3rd or 4th best pitch, but throwing it about 5-10% of the time can make his other pitches more effective if the other team isn't sure what they're getting.  You're still going to throw your best pitch over 50% of the time, but even Mariano Rivera threw something other than a cutter once in a while.  You just have to practice it enough so that you're throwing a decent enough pitch instead of a meatball.

A player having a decent midrange game will discourage hard closeouts on 3s and make those shots better, and it will draw defenders out to more closely defend 15 foot jumpers, opening up driving and passing lanes to the paint.  

kehnonymous

February 24th, 2021 at 2:24 PM ^

Agreed - if you never take midrange twos, you become a lot easier to defend because you've effectively given away about 2/5ths of the half-court area that your squad has to guard.  To paraphrase Robert Downey's character in Tropic Thunder, "You never go full Morey" 

Long twos are "bad" shots in a vacuum but in context they have their role and sometimes you have to take what the defense gives you.  Since we're still a football school(?), I'll use a gridiron analogy -your basic off-tackle dives can be useful to set-up play-action bombs, but a) if you are abjectly terrible at off-tackle runs or b) you keep going to those basic plays or c) all of the above (cf, 27 for 27, sorry y'all), then they are in fact symptomatic of a bad offense.

WolverineMan1988

February 24th, 2021 at 9:52 PM ^

Agreed. It’s the idea of being a “three level” scorer. The ability to take and make 2point jumpers when needed makes it tougher for the opposing defense to take away something. If you’re not a threat at all to make a 2point jumper, then teams have a chance to run you off the 3 point line and pack the paint if they have rim protecting big/s. With that being said, a mid range jumper is always going to be analytically the worst shot you can take. However, that doesn’t mean you should never take them.

I think Michigan does a fantastic job of knowing when to take them and also who should be taking them. Also, OSU is an example of a great statistical offense that takes and makes a fair amount of mid range jumpers. Liddell and Walker in particular are very good at them. In the case of both of those players, especially Liddell, I believe their ability to make midrange jumpers at a fairly high clip makes them tougher overall to guard.

notYOURmom

February 24th, 2021 at 5:25 PM ^

There is a modeling problem here in the form of selection bias, in that players presumably prefer to take shots they think they can make.  And their ability to predict this accurately also depends on the shot.  Someone ought to throw a Heckman-type model at it.

Cmon whose good at endogenity problems?  Don’t everybody shout at once 

JamieH

February 24th, 2021 at 6:08 PM ^

Yeah, I think the quality of the look determines a lot about it being a "bad" 2 pointer.

An uncontested 15-foot jump shot for a good shooter is not a ton harder than a free throw.  I would guess good shooters hit over 50% when they are unguarded from 15-feet.  If they are hitting 50-60% of those shots, they aren't "bad" shots.

A heavily contested 15-footer is a bad shot for all but a star player, and even then they should only take them late in the clock.



 

umchicago

February 25th, 2021 at 12:01 AM ^

Ya, I was going to say that livers is about the only guy that consistently takes 2 pt jumpers. And often they are turn around late in the clock. Several guys take floaters and tweeners in the paint.