APBRmetrics

[schtevie]

As I have perhaps too repetitively noted, the typical NBA franchise shoots way too few three pointers. More to the point, the belief in the long, mid-range jump shot as an important part of the offense is a curious but inexcusable anachronism.

Consider the future and the foot-dragging past. Last night Orlando pasted Phoenix. OK, it was an away game for Phoenix and the second night of a back to back, and what does one game mean anyway? Well, it provides a very nice illustration of the general point.

Both Phoenix and Orlando are teams that shoot threes with well above average efficiency. They differ in the relative frequency with which they take the shot, Orlando is at the top of the table, making it a (the?) central feature of their offense; Phoenix, despite the skill, insists on being about average.

And what is the relevant trade-off? Clearly that would be fewer long twos for more threes. Let's go to last night's shot chart. Defining free throw distance (15 feet) to the three point line as the relevant area of offending shot selection, we see that Phoenix took 26 shots and made 9 (34.6%), and this 26 was out of 66 non-three FGAs (or 39.3%). By comparison, Orlando, had no more success with the long two, making just 33%, but they didn't waste their shots on them, taking just 12 out of 65 FGAs (18.5%) from this range.

My point is not that Phoenix would have won the game if they were organized to play like Orlando. A pasting is a pasting. But they almost surely would have put more points on the board if they had smartened up on their shot selection. Assuming that 14 more threes would have been taken by Phoenix - thereby equalizing the Orlando attempts - they might have been expected to add "only" 3.6, 6.4, or perhaps even 9.4 more points to the scoreboard (numbers reflecting the actual game 3pt%, last season's average, and this year's average to date).

The point is that the shot selection the Suns displayed is not at all atypical, both in terms of the frequency of the long two and its success rate. All the while that there is an apparent dominant strategy, exemplified by Orlando, that just sits unexploited by so many teams. I say dominant because the ratio between long twos and threes is almost strictly a choice for the offense. The defense extends outward; it cannot prevent the three.

Maybe there is some mitigating factor besides the dead hand of tradition. My hunch is that the remainder can be explained by fear of change/ridicule and the fact that the average expected gains are not enormous.

But they are large, probably on the order of one to two points per game, on average, enough to bring you up a level but not to make an average team a championship contender.

And what this implies this year, looking from the top down, is that if Orlando wins it all, it will be because of their being on the vanguard. Conversely, for each of the primary contenders, should they fall short it can be chalked up to this opportunity forgone.

[Mike G]

So far this year, eFG% from 3pt shots is .526, league-wide.
On 2FGA and FTA (combined) it's .537 .

If a few of those FTA are from 3pt attempts, they're about even.

When you draw fouls, you gain additional advantage on the opponent. Dwight was fouled out in 17 minutes at Detroit; probably not while defending the 3.
_________________
`
39% of all statistics are wrong

[Ryan J. Parker]

Certainly the 3pt shot is valuable, but this doesn't mean teams should start throwing up a bunch of garbage 3pt shots. Making a concerted effort to create quality 3pt shots is important, but so is driving in the paint and drawing fouls.

In the Magic's case, a guy like Dwight Howard helps create quality 3pt opportunities.
_________________
I am a basketball geek.

[deepak]

The threat of the 3-point shot can also open up space down the middle of the floor, making the interior more vulnerable. So I don't think you can just separate out efficiency on 3-point shooting and 2-point shooting. Excellent 3-point shooting can indirectly improve efficiency elsewhere.

I just quickly took a look using data from basketball-reference. Since the beginning of the 05/06 season, 3PA% (percentage of true shot attempts that were 3PA) has a .360 correlation to offensive efficiency. FTA% (percentage of true shot attempts that were trips to the line) had only a .064 correlation.

[schtevie]

Mike, perhaps I should have been clearer, though I don't see how I could have been. I am not talking about two point shots generally. Yes, dunks are good. I am talking about long jump shots. That is what Orlando minimizes, relative to the league.

Unless you are prepared to argue that long twos get fouled at the same rate as the average two (which would be foolish because it isn't true) you are comparing apples and oranges. And never mind the fact that the numbers you cite subsume fast breaks and half court offenses and ignore the differential effects of turnovers.

The fact of the matter is that long twos are typically terrible shots. They drag a team down. A player, or team, may be above average at that particular skill, but that doesn't redeem the strategy as returns are still relatively low. Yes, there are times within games (i.e. as last options as the shot clock is running down) when a long two can be appropriate. And, yes when a really good jump shooter gets a really open look. But the point is that otherwise - and it is mostly otherwise, as the numbers tell us - they drag the average performance of a team down. It is hard to find a game for any team - a single game - where long twos yield above average returns.

If you should doubt this, simply pick some games at random and let me know how many you go through before you find one.

[Jon Nichols]

In the end, I think it depends on the abilities of your players. The Magic have a unique situation with a dominant post player that can suck in defenders and a power forward adept at shooting threes. Not every team is blessed with these players, so you have to adjust based on what you have.

I do agree with your point about long two-pointers, though. I see no real benefit to those shots if they are converted at around the same rate as three-pointers but are worth less points and have no other apparent benefits.

[deepak]

Have people observed any correlation between the more stats-savvy teams and frequency of 3-point attempts? I know the Rockets shoot a lot of 3s.

[schtevie]

Jon, I have to disagree with you that this strategy choice is player dependent. Yes, if you have better three point shooters or better low post players it makes the choice to increase the proportion of threes what should be beyond obvious. But it is not player dependent.

To note the possible exception that proves the rule consider Philadelphia last year. Clearly the worst three point shooting team in the league, with a eFG of only 47.7%. And what was their return on long twos (even adding in the foul shots earned thereby)? I don't know, but I would bet that it is lower than 47.7%. I say this because it is rare indeed to find individual games on good shooting teams that get this return, let alone season averages for a poor shooting team.

And then there is history. It is an amazing, not-often-remarked-upon fact how three point shooting over the last decade has been in essentially perfectly elastic supply. Offenses have shifted to the three with no diminishing returns. More shots taken, no problem, the yield is an eFG of about 54% (+/- not much). What this means is that success at the three hasn't depended on personnel. It has depended on coaches decisions. There have been good low post players in the past whose teams didn't shoot threes, and there will be mediocre low post players in the future whose teams will shoot a lot of threes (instead of crappy twos).

As for deepak's notes about correlations, one simple point. The higher math isn't required! Unless there are Xs and Os reasons why long twos need be taken (but then again, one has to explain the history) isn't it basically a case of comparing two averages? One doesn't need to be a supah-genius to master the concept of eFG and partition conventional fg% (or TS%) by shooting zones.

[gabefarkas]

Philly's eFG% on all 2PA last year (which is the same as their actual FG% if we're only looking at 2PA) was 48.6%. I'm not sure offhand what it was for just long jumpers.

On all 3PA, they achieved 0.954 points per attempt. On all 2PA, this was 0.973 points per attempt. Again, I'm not sure what it was for just long jumpers.

[Jon Nichols]

[Kevin Pelton]

[deepak]

[Kevin Pelton]

[schtevie]

[Crow]

It depends, but only somewhat, on how easy it is to get the next extra 3 point shot attempt and how good a look it is.

Not taking long 2s is mostly within the control of the offense, except when you waste time or get forced into the last few seconds of the shot clock. While you'd like that next 3 point attempt to be at near average 3 point efficiency to get a big payoff you only have to hit around 27% to beat the league average efficiency on all mid-range shots (not so close enough to be called close or from 3) of about 40% or the 38.5 % average on 16-23 footers cited at hoopdata. It should be pretty easy to do at least that and replace long 2s and improve efficiency bit by bit. It still adds up and is a significant opportunity that the insiders in charge on the strategy on offense - and the players- have not fully exploited yet.

The 3 point game has increased in frequency over time but probably hasn't reached its max potential. a speculation supported by the rise over time in the NBA and partially supported by its reported even greater emphasis in Europe. There may be a max where the defense actually denies even a half-way decent 3 point chuck but until you get there a 3 point attempt with a 27% or greater chance of going in is better than a long 2 for under 40% for almost all players on the court.

And yes the teams with stat analysts do generally have higher than average 3 point rates and 3 pt FG%s. At least when I checked and reported on that a year or two ago.


Teams have equalized 3 point efficiency with all other attempts but I tend to think that 16-23 foot shots have nothing to do with trying to get close or get fouled and should be disconnected from the results of the 'inside effort".

10-15 foot shots and shot less than 10 feet but not "at the rim" are barely better than 16-23 foot shots either at 40.9 FG% and 41.6 FG% respectively league-wide. You probably have to end up with a substantial share of these while trying to get good inside action and I think that is acceptable since it works efficency-wise over all those shots; but as much as possible either get within 5 feet or get fouled while threatening to get within 5 feet or pull or kick it out and try to hit a higher percentage shot inside or from 3 some way. A 3 any way you can hit more than 27% of them is better than everything other than the inside action. A mid-range is only better on average than a turnover.

The general thrust of the conventional way players think and managers manage in basketball has been closer is better. Hence we see players trying to get the 15 footer or the 10 footer. But only really close or threatening the defense rightly or wrongly into a foul is better. The NBA has learned to value the 3 but not fully and most offenses appear to not have learned to dislike the mid-range near enough. And most fouls on a 10 -15 foot shot or a player threatening to put one up are unwise.