Team 3pt FGA / mid-range FGA ratios

[Crow]

I might write more about this later, but am pressed for time at the moment:

Team 3pt FGA / mid-range FGA
HOU 1.38
NYK 1.06
LAL 0.83
POR 0.75
ATL 0.74
BKN 0.71
SAS 0.71
MIA 0.71
DEN 0.70
LAC 0.67
OKC 0.61
TOR 0.60
SAC 0.58
CLE 0.57
DET 0.56
IND 0.55
MIN 0.55
DAL 0.54
ORL 0.53
GSW 0.53
NOR 0.52
MIL 0.51
CHA 0.49
BOS 0.46
WAS 0.46
PHO 0.45
PHI 0.43
UTH 0.42
CHI 0.39
MEM 0.35
Average 0.59

Houston with by far the best overall team "shot" distribution by this ratio with only Denver having a different argument with their league leading at the rim frequency and pretty high 3 pt FG / mid-range FGA ratio.

HOU 2.34 * the average
MEM 0.6 of the average
Hou / Mem ratio 3.90

Top 4 teams in league (overall) are tightly bunched on this ratio. Is that the optimal range?

[knarsu3]

Thanks Crow. I decided to expand on that and look at the correlation between 3pt FGA/mid-range FGA and Team ORtg. Correlation Coefficient was 0.609. I was expecting higher but I suppose that's a fairly strong correlation considering shots near the rim are not being considered.

Also, the standard deviation for 3pt/mid-range is 0.21 which means that Houston is nearly 4 standard deviations above the average for this ratio. Is this just a statistical oddity or have Morey and Co. simply refused to shoot mid-range shots? Either way, Houston is 6th in ORtg so they clearly are doing something right.

I have to say its surprising to see analytic heavy organizations like Memphis and Boston so low. I mean its not surprising if you watch them play (KG seems to only take mid-range shots these days) but from the standpoint that you would think an analytics heavy organization would be more aware of this? The only possible explanation is the % of open mid-range vs. guarded 3 available to the team (and still, I don't know what the %'s are but I believe Evan Z did a study on this awhile back) However, I would imagine that for each team over a large enough sample size, that'll all average out. I doubt any other teams are different from Boston and Memphis in seeing more guarded 3 opportunities and more open mid-range shots. I imagine it all averages out over the long haul. So are these teams just not listening to their analytics teams? For the life of me, I've never understood why KG just doesn't take a step or 2 out and just shoot 3s instead of consistently shooting the lower EV mid-range shots.

Code: Select all

		       ORtg
ATL	0.74	104.9
BOS	0.46	102.9
BKN	0.71	108.4
CHA	0.49	101.1
CHI	0.39	103.4
CLE	0.57	104.5
DAL	0.54	105.6
DEN	0.70	110.2
DET	0.56	103.7
GSW	0.53	106.4
HOU	1.38	109.8
IND	0.55	104.5
LAC	0.67	110.6
LAL	0.83	108
MEM	0.35	104.9
MIA	0.71	112.4
MIL	0.51	103.6
MIN	0.55	102.9
NOR	0.52	105.8
NYK	1.06	111.2
OKC	0.61	112.5
ORL	0.53	101.8
PHI	0.43	101.8
PHO	0.45	101.2
POR	0.75	106.1
SAC	0.58	106.3
SAS	0.71	108.8
TOR	0.60	105.5
UTH	0.42	107
WAS	0.46	100.1

[crsofa]

I have to say its surprising to see analytic heavy organizations like Memphis and Boston so low.
[/code]

In Boston it has gone up w/o Rondo. By NBA.com numbers http://stats.nba.com/leagueTeamShots.ht ... N&splits=Y (they use every non-paint shot as mid range), they have a 704÷1221=0.58 ratio up to January 27, and 631÷909=0.69 ever since. I guess some of that had to do with the way that Rajon plays.

[Crow]

Thanks both for the feedback with further analysis.

I should mention that I was using hoopdata information for my original post where anything not at the rim or from 3 pt land was considered a mid-range shot. In the nba.com data I would count the in the paint (non-restricted area) as mid-range.

It appears that the Magic were the first to have a ratio better than 1.0 in 2008-9. They did it again in 2009-10. I believe they were the only ones to do it before the Rockets. The Magic just missed last season.

Houston year to year team shot chart per game change
2011-12 2012-13
At Rim 23.6 30.2
3-9 Feet 12.7 7.8
10-15 Feet 8.3 3.0
16-23 Feet 19.2 10.1
Three 20.2 28.8

They had to convince the coach and players that 3-9 foot and 10-15 foots were not good (on average) / not part of the strategy. They also convinced them to to get to the rim more. Why did it take until 2012-13? Why is no one else achieving this right now? I would hope / think there should be 5-10 teams doing it with a few years. It would take more analysis and presentation of evidence to argue that this is not the right way to go in the abstract or for a particular team, I think.

But I guess we have to see if anyone can win a title this way. Maybe, as I suggested earlier, the optimal ratio for a championship team is somewhat below the max achieved but perhaps still above the league norm? I believe that the highest ratio of a title winner in the 6 seasons was Boston in 2007-8 at .584 (not that high) . I haven't computed the seasonal averages for previous years yet to know if they were above then league average.
.

[deepak]

As far as Houston, I don't think its so much the front office having to convince the coaches where the shots should come from. Its a personnel thing. The two major scorers for the Rockets last season were Kevin Martin and Luis Scola. They both took quite a lot of mid-range jumpers, because that suits their game.

[mtamada]

As for Memphis, this articleis highly suggestive: it's a nice in-depth look at the pros and cons of Lionel Hollins as coach. Lots of both. One of the cons is that he's an old-school coach when it comes to analytics. So Memphis has to decide whether they want go with a coach who'll be on board with their program, or keep Hollins whose got several positives -- but these do not include being aware of the strengths of the 3-pointer and weaknesses of mid- and long-range 2-pointers.

[Crow]

Will a high 3pt FGA / mid-range FGA ratio figure into a HOU, NYK or LAL (or other) playoff upset win? Do the ratios tend to stay at regular season levels or change in the playoffs? More to watch and think about.

[Crow]

As far as Houston, I don't think its so much the front office having to convince the coaches where the shots should come from. Its a personnel thing. The two major scorers for the Rockets last season were Kevin Martin and Luis Scola. They both took quite a lot of mid-range jumpers, because that suits their game.

I don't know. and no one outside the Rockets does. But did they just change the players and then accommodate their strengths (whatever they were) or did they change / select the players with the specific purpose of fitting a strategy that radically changed the shot distance distribution to one very rarely even seen before? My interpretation is the latter but it will help to see if they sustain this novel team short chart long-term (including beyond this current wave of lead players).

[Notsellingjeans]

Great post. I agree that the Rockets' unloading of Scola contributes to their eye-popping ratio of 3's. So, here's an under-reported story moving forward:

None of the free agents the rockets are perpetually linked to are good shooters.

Josh Smith is the most egregious 'bad long two' jacker in the game. He's antithetical to everything the rockets do on offense.

Dwight Howard, pekovic, mill sap, david west or Bynum couldn't share the floor with Asik without mucking up the floor spacing they've worked so hard to cultivate, either.

Sure, they could trade Asik, but that would require finding a team that has significant cap space, a hole at center, an inability to fill that hole with smith, Howard, or Bynum, and the future assets the rockets would want in return for Asik, who is undeniably a bargain. I don't see that happening.

So...I guess it's kind of a bad year for the rockets to have cap space. Ideally, a player of Ryan Anderson's profile would be hitting the market right now. Long stretch 4.

What will the rockets do? Which free agent FA would allow them to continue to use their current, forward-thinking 3-happy philosophy?

[Jacob Frankel]

I think in a dream world they maneuver an Asik/Howard SnT and one of the 4 young PFs they've assembled (Motiejunas, Smith, Jones, Robinson) breaks out.

[Crow]

Harden's FGA/gm went about 70% last year to this year. His inside attempts went up about proportionately. Mid-range actually more than proportionately and 3 pt attempts less than proportionately.

Lin stayed about the same inside and from 3 pt land. Cut his attempts per game from 3-15 in half. The long 2 rate stayed about the same.

For Parsons the inside shot rate stayed proportional, mid-range stable per game but down proportionately. 3 pt rate up about 50%.


The overall Rocket shot chart trends are not rigidly adhered to for top shooters this season and don't seem to provide the bulk o fthe explanation of the change at team level. The impact of the supporting cast's shot profile change might be major.

[schtevie]

A couple of observations:

(1) I'm not sure what is the best measure to look at for an offense's propensity to shoot 3s. My first thought was that one should compare choices at the distance margin, therefore looking at the ratio of 3PAs to long 2s and preferably in the form 3PA/(3PA + long2PA). But then again, reading this 82games article (http://www.82games.com/locations.htm), perhaps the ratio of 3PAs to all non-rim attempts is preferred as it isn't clear that the relevant comparison is distance, rather overall efficiency gains.

(2) I think the accomplishments of Houston, in terms of shot selection (that is well above average share of shots both at the rim and from 3) should be put in larger context, both in terms of (recent) history and offensive efficiency. First the latter.

Looking at the data from both hoopdata and B-R, it seems that the entirety of Houston's above average offensive efficiency can be accounted for by its shot selection! That is to say that at each listed range, Houston's efficiency was about average, and then the effects of FT/FGA and TO% approximately offset. And in the instance, this "optimal shot selection" gain was about 3.9 points per 100 possessions. That's definitely not nothing.

Then looking for previous examples of such offensive orientation, there is really only one (using hoopdata numbers, which go back to 2006-07) and that is Don Nelson's Golden State Warriors, most clearly the 2007-08 team which was 4.3 points per 100 possessions better than average, owing mostly to its shot selection.

Now, improving efficiency by 3 plus points is a really big, competitive deal, and possibly achieving this by "deciding" to take shots from the better places is, of course, dependent to some (but unknown) degree on having the proper personnel.

It will be interesting to see the extent to which Houston's "model" is copied throughout the league if at all. One doesn't think of any team having been inspired by DN's GSWs. Why? Well, because they weren't winners (not even having made the play-offs, despite 48 wins). Clearly however greater attention is being paid to the machinations of Daryl Morey. But until Houston at least advances far into the play-offs, I wouldn't expect there to be that much influence.

[Crow]

When we talk about using analytics it is almost always focused on the activity of or directed at management (GMs, consultants, coaches). Another pathway is by players perhaps independent of team action. I have wondered how much agents are aware of and using the available (or possible) data with their clients and not just using it for contract talks with management but using it to shape their player's behavior and best "position" them for future contract talks. I know some are using it, I just don't know how aggressively. A simple example: Harden's shot distribution chart (pretty low on mid-range shots and very high on at the rim / foul shots and 3 pointers) would seem to be a good model for many wings to try (really hard and disciplined) to emulate. But players and agents could benefit from being pro-active in many more detailed ways. There could be conflicts over how a player plays with the benefit of analytics. The way that is best for the team and the way that is best for the player. The conflict is no surprise but it might be discussed / fought somewhat differently with more and better analytic data.

[dblack]

A couple of observations:

(2) I think the accomplishments of Houston, in terms of shot selection (that is well above average share of shots both at the rim and from 3) should be put in larger context, both in terms of (recent) history and offensive efficiency. First the latter.

Looking at the data from both hoopdata and B-R, it seems that the entirety of Houston's above average offensive efficiency can be accounted for by its shot selection! That is to say that at each listed range, Houston's efficiency was about average, and then the effects of FT/FGA and TO% approximately offset. And in the instance, this "optimal shot selection" gain was about 3.9 points per 100 possessions. That's definitely not nothing.

Motivated by this post I decided to look at each team's eFG% and see how much is gained or lost from shot selection and how much is gained or lost from just simply being a good shooting team. I'm new to all this so I just wanted to do something really simple. I broke down each team's shot selection into 6 "zones" - within 3 feet, 3-9 feet, 9-15 feet, 15+ feet 2 pointers, corner 3s and non corner 3s. I found the league average shooting % from each of these zones and then determined what each team's eFG% would be if they shot at the league average from each zone but kept the same shot distribution. The difference between this and the team's actual eFG% is the team's "shooting factor". The difference between this and the league average eFG% is the team's "shot selection factor". So the shooting factor represents how much above/below average a team's eFG% is as a result of their shooting ability and the shot selection factor represents how much above/below average a team's eFG% is as a result of their shot selection. I realize there is more to a team's eFG% and this is a pretty basic way of looking at this but I still found the results pretty interesting.

Code: Select all

team effective_FG% shooting_factor shot_selection_factor
Mia     55.2             4.8                   0.8
SA      53.1             2.9                   0.5
OKC    52.7             3.1                  -0.1
LAC    52.5             2.5                   0.4
Hou    52.5             0.0                   2.8
Atl      51.7             1.9                   0.2
Den    51.5            -0.6                   2.5
NY      51.5             0.6                   1.2
LAL    51.2             0.8                   0.7
Dal     50.6             2.4                  -1.4
GS     50.6             2.2                  -1.3
Bos    50.3             1.1                  -0.5
Por    49.8            -0.1                   0.3
Bkn   49.8             0.0                   0.1
Uta   49.2             0.5                  -1.0
Sac   49.1            -0.7                   0.2
NO    48.9            -0.5                  -0.2
Tor   48.8             0.4                  -1.2
Det   48.7            -2.1                   1.2
Orl    48.5            -0.6                  -0.6
Phi    48.1            -0.4                  -1.1
Ind    47.9            -1.3                  -0.5
Pho   47.7            -1.1                  -0.8
Mil    47.7            -2.3                   0.3
Was  47.5            -0.6                  -1.5
Cle    47.3            -1.9                  -0.4
Min   47.3            -2.7                   0.4
Mem 47.3            -1.4                  -1.0
Chi   47.0            -2.1                  -0.6
Cha  46.0            -3.6                  -0.1

[Mike G]

This is very interesting. I was almost motivated by Schtevie's comment about Houston shot selection, to do something like this.

But what we don't know is how FT figure into this. Some locations are more likely to produce FTA, which are almost always the most efficient.
Then, too, turnovers result from perhaps the same types of shots that create fouls; including offensive fouls.

After Houston, there's Denver; then a pretty big dropoff, in this "shot selection factor."
Denver gets a lot of shots at the rim. They're not that high in FT/FGA, avg in TO%