henry gently pwns two HOF coaches

[schtevie]

I like reading TrueHoop. And when they say something silly or wrong about a stats-related issue I enjoy discussing it here. Conversely, when they do a really good job on a stats-related subject, it merits special notice. This post by Henry Abbott is great and ought to be read by all who pass by these pages:

http://espn.go.com/blog/truehoop/post/_ ... mmon-sense

Ever so gently, he pwns not just Phil Jackson but Jerry Sloan, making the implicit points that just because you're a Hall of Fame Coach, doesn't mean you don't make very costly strategic mistakes and that the dead hand of tradition is not something that should be celebrated.

The excellence of the piece notwithstanding, there are a few points that could have been made better and a couple of things that weren't quite right.

(1) In the piece it is noted that the Jazz benefited mightily from the actual increase in 3PAs post-Hornacek. That is true, of course, but the point can be made stronger still. On average, in the four post-1993-94 years, more than all of the realized gain in offensive efficiency is attributed to the Jazz' increased 3PAs. And yet they still lagged at the bottom of the 3PA rankings! I have to suppose that there was a basic lack of ability or interest in being able to multiply by 1.5. Kind of astonishing.

(2) Mention is only made of the related foregone opportunity in the 1997-98 Finals. This, of course, was not the only championship not won by the Jazz. In the previous year, they lost games in the Finals by 2, 2, and 4 points! But would three point shooting really have gotten them over the hump in both series? By my calculation, if you posit that the Jazz would merely have matched the Bulls' rate of 3PAs (the seasonal average) any reasonable estimate of the gain in offensive efficiency from taking more 3PAs would have been sufficient to have made them equal in expected strength to the Bulls in 1997-98 and their superior the following year (strength defined as seasonal OffRtg - DefRtg). So the bottom line is that is very reasonable to believe that there "should" have been one Jazz Championship in those two years.

But that is not all! The Sloan philosophy regarding threes was a characteristic of his entire (very successful) tenure. Exactly how many championships in expectation should have been theirs is an involved question that might not be worth addressing, but the answer is surely greater than one. And a final note in this regard is that Phil Jackson was disproportionately the beneficiary of this strategic error, not only as coach of the Bulls, but the Lakers too.

(3) But let's not leave Phil Jackson off scot free (and I say this not as a hater, being a native Illinoisan and growing up a Bulls fan.) There is good reason to believe along these same lines that 1993-94 was the year of the four-peat that wasn't. There were two very, very good three point shooters on that team (B.J. Armstrong and Steve Kerr, with an overall average 3P% of 0.432) whose talents were conspicuously underutilized.

Here's a "what if": If those two players shot 3s at a modern rate for players of their revealed ability (modern rate: for 2012-13, players playing 20+ minutes per game, and shooting greater than 0.400 from the line, the median 3PAs per game is 5.2) and you assume an opportunity cost of about 0.39 points per shot (what is a reasonable estimate and what was, in particular, their advantage over that of the 3P effectiveness of the other two primary 3P shooters on the team: Scottie Pippen and Toni Kukoc) you get an estimate of potential improvement of about 2.3 points per 100 possessions. And such a gain would have placed them 4th best in the league in terms of net efficiency (again, OffRtg - Def Rtg).

(4) In terms of what wasn't quite right in the piece, two points. First, a factual quibble. It has not been only three plus decades since the introduction of the 3 point shot, it's been almost 50 years. It is simply incorrect to forget the ABA. Their experience provided many opportunities for a teaching moment when it came to utilization of the 3. It is quite simply appalling how slow the adoption of the 3 point shot has been.

And second, the final thought offered I find a wee bit (and irrationally) optimistic: "But blending the right lessons of the past with the right innovations from the future can come with big rewards." What is the evidence or argument that there are big rewards for anyone in future? The point I've made many times before is the same one Henry (implicitly) makes in his piece: the low-hanging fruit is no longer there for the picking and most of it rotted on the tree anyway. So we have to climb higher up the tree now. What evidence is there that there's anything there that can help a team gain an important competitive advantage, even if they weren't too hidebound to act upon it?

[MW00]

An interesting piece, as was your analysis.

Quibbles with the quibbles:
- If you're going to get technical the ABL had the three point line before the ABA.
- I'm not sure that the final paragraph does suggest that there's low hanging fruit to be had. Abbott is a journalist and he doesn't want to seem like he's crowing about how he knows more about basketball than Phil Jackson. So the last paragraph doesn't seem to have anything egregiously wrong about it. It does seem to be waffling and/or a truism (teams that play smart can win "big rewards").

[Mike G]

pwns?

[Mike G]

The ABA only demonstrated what a lousy weapon the 3-point FG was; with the rare exception of teams which had players who were good at it.

In 9 years and 90 team-seasons, there were 24 instances -- roughly the top 1/4 -- of a team shooting more than 500 3fga in a season. Of these, 13 had >.500 seasons, and 10 were lower.
This is not surprising: A team with good shooters is likely to be a good team. But the correlation with 3-pt frequency is not great.

Meanwhile, in the bottom quarter of 3-pt-shooting teams -- 23 teams taking fewer than 260 3fga -- 17 were winning teams, and just 6 were losing teams (not counting teams which folded early in a season).
The strong correlation between 3fga and losing suggests that in 9 years, a few hundred ABA players did not become effective long-range shooters.

In no season did ABA 3FG% clear .300 .
3FGA frequency topped around 6 per game, in years 2 thru 4.
In years 6 thru 9, it was below 4 per team per game.

Having tried and failed to make the 3 a viable and reliable weapon -- excepting the few players who could actually regularly make it -- the norm at the end was to use the 3 like the NFL uses the 2-point conversion after a TD: Just about never, unless it was truly necessary.

The NBA only caught on after a few years when Bird and Ainge used it well. But in the ABA, you had 29% 3FG shooters vs 47% on 2FGA, so the tradeoff was mostly not a good one.

[schtevie]

Is can't be February 2nd already, can it?

The ABA only demonstrated what a lousy weapon the 3-point FG was; with the rare exception of teams which had players who were good at it.

This statement is simply and resoundingly false, on its own and especially in the context of the point I was making about the ABA: that its experience should have better informed NBA practice.

In 9 years and 90 team-seasons, there were 24 instances -- roughly the top 1/4 -- of a team shooting more than 500 3fga in a season. Of these, 13 had >.500 seasons, and 10 were lower. This is not surprising: A team with good shooters is likely to be a good team. But the correlation with 3-pt frequency is not great.

Meanwhile, in the bottom quarter of 3-pt-shooting teams -- 23 teams taking fewer than 260 3fga -- 17 were winning teams, and just 6 were losing teams (not counting teams which folded early in a season). The strong correlation between 3fga and losing suggests that in 9 years, a few hundred ABA players did not become effective long-range shooters.

The issue is not now, nor has it ever been, about what the cool (or the uncool) kids are doing. The argument for adopting the three point shot is and always has been: if you replace a non-three point shot with a three point shot, is there a positive gain on the margin?

In no season did ABA 3FG% clear .300 .
3FGA frequency topped around 6 per game, in years 2 thru 4.
In years 6 thru 9, it was below 4 per team per game.

The relevant facts are not whether, on average, ABA players were adept three point shooters (in a more modern context) nor the actual number of three point shots taken. Again, the relevant issue is whether there were points foregone for not having taken more three point shots, and as to this, the data speak quite clearly.

3P% in the ABA ranged between 0.285 and 0.300, an effective shooting percentage of 0.427 to 0.450. Compare this to realized 2P%, what increased from 0.428 in the ABA's first season to 0.480 in the next to last.

As we all know, the overall 2P% includes a range of scoring opportunities, from dunks on the high end to contested mid/long range 2s on the low. And what do we know about the latter? Recent history tells us that they are very bad shots indeed, having an expected return much lower than 0.427 to 0.450, and that there are a heck of a lot of these shots taken. And recent history surely informs us about the past. Agreed?

So, it has ever and always been true that three point shots have dominated the worst two pointers (what there are a lot of). And it has also ever and always been true that there have been players able to hit three pointers at a percentage higher than the "crap 2" threshold. Let's call that 0.400. Over its life, the ABA had an average of 4.1 players per team with such exhibited three point proficiency (3P% of 0.267). Now let's raise the bar to the highest realized average 2P%, 0.480. Over its life, the ABA had an average of 2.7 players per team who accomplished that (3P% of 0.32)

Could the average ABA team have increased its 3PAs? Yes. Would it have been optimal to do so at the expense of the worst 2PAs? Yes. Did anyone in the NBA in the 1980s (with the ability to count and multiply by 1.5) pay any attention to the ABA experience? I am supposing not.

Having tried and failed to make the 3 a viable and reliable weapon -- excepting the few players who could actually regularly make it -- the norm at the end was to use the 3 like the NFL uses the 2-point conversion after a TD: Just about never, unless it was truly necessary.

Returning to the original point, the issue isn't whether the ABA optimized three point shooting in its leagues short life. It didn't, and no one has ever claimed that it did. In fact, there was demonstrated retrogression, in terms of fewer 3PAs per game.

The NBA only caught on after a few years when Bird and Ainge used it well. But in the ABA, you had 29% 3FG shooters vs 47% on 2FGA, so the tradeoff was mostly not a good one.

This first sentence is a very strange short history of the early adoption of the three point shot in the NBA. I've written about this many times in the past, with special (and different) mention about the curious non-factor that was Larry Bird. But really enough for now.

[Mike G]

The issue is not now, nor has it ever been, about what the cool (or the uncool) kids are doing.

Hmm, I was talking about how the better and best ABA teams did not shoot a lot of 3's, in general. You don't win more games by being "the cool kids". Or you could explain what you mean by that.

The Kentucky Colonels are a good case study. In their early years, they stayed competitive by shooting lots of 3's (relative to others), and better than most.
Once they got effective inside scoring, via Dan Issel and Artis Gilmore, their outside guys stopped shooting so many threes. It's hard to beat .600 efficiency on 2-pt FG and the associated FT.

Is can't be February 2nd already, can it?

It was already, and it isn't again yet. Maybe you know what you're referring to?

...three point shots have dominated the worst two pointers...

By definition, everything else is better than "the worst".

Over its life, the ABA had an average of 4.1 players per team with such exhibited three point proficiency (3P% of 0.267)

That's quite a claim. In 90 team-season, 4.1 per team would be 369 player-seasons.
Just 375 player-seasons were as good as .267 in a season in the ABA. Many of these were 1-3, 2-7, etc. The median is around 30 3FGA for a season.
If a player has hit 8 of 30 in a season, would you counsel him to take 300 next year?

[schtevie]

The issue is not now, nor has it ever been, about what the cool (or the uncool) kids are doing.

Hmm, I was talking about how the better and best ABA teams did not shoot a lot of 3's, in general. You don't win more games by being "the cool kids". Or you could explain what you mean by that.

I return to the subject matter of this post. The evidence is rather very strong that two coaches generally regarded as being better than most failed to maximize their team's competitiveness. That they felt they had this luxury (and poor Jerry Sloan didn't) and still to this day Phil Jackson doesn't see the error of his past ways owes primarily (I suppose) to the fact that they were coaching very good players. Who's to question success!

The point is simple: you should expect very little information looking just at a simple, gross correlation between win% and 3PAs. If you're good, the evidence shows little incentive to think outside the box, and if you're bad, you also have little incentive, for fear of being deemed an unreliable freak.

The Kentucky Colonels are a good case study. In their early years, they stayed competitive by shooting lots of 3's (relative to others), and better than most. Once they got effective inside scoring, via Dan Issel and Artis Gilmore, their outside guys stopped shooting so many threes. It's hard to beat .600 efficiency on 2-pt FG and the associated FT.

Without specific reference to the particular case, even teams with dominant post players shoot a lot of crap 2s, and these are what should be substituted for.

...three point shots have dominated the worst two pointers...

By definition, everything else is better than "the worst".

If the point you are making is that instead of the bad two point shots, better two point shots should have been taken. Well, I agree, and part of the observed, historical progress in the NBA was indeed just that (for example, being more patient with respect to the shot clock). This discussion, however, is about the exchange of bad (typically understood to be long) twos for better threes, what are understood to be rather close (and readily achievable) substitutes.

Over its life, the ABA had an average of 4.1 players per team with such exhibited three point proficiency (3P% of 0.267)

That's quite a claim. In 90 team-season, 4.1 per team would be 369 player-seasons.
Just 375 player-seasons were as good as .267 in a season in the ABA. Many of these were 1-3, 2-7, etc. The median is around 30 3FGA for a season.
If a player has hit 8 of 30 in a season, would you counsel him to take 300 next year?

My 4.1 calculation came from taking the average of the seasonal averages. As for not imposing any qualification for minimum 3PAs, I think that was absolutely the right thing to do.

The overarching point is this: there is no question that there was an elastic supply of 0.267 and above three point shooters in the ABA. In fact, I suspect you would agree with me that the primary reason the realized efficiency was so low did not reflect a lack of fitness or ability, rather it was because the shot was anything but an integral part of the offense that the realized efficiency was so low (perhaps more often than not taken as a close to desperation heave). If coaches would have built offenses that recognized the potential of the three point shot, the players would have come, just as we've seen and are still seeing in the NBA.

[Mike G]

If coaches would have built offenses that recognized the potential of the three point shot, the players would have come, just as we've seen and are still seeing in the NBA.

They'd have come from where? Once they instituted the 3FG in college, they came from college. But in the ABA, only 11 players in their careers attempted as many 3's as Steph Curry did last year (600). Only 3 of these managed to hit the 2013 NBA avg 3FG% (.359).

Because the ABA started out as a 3-happy league, which as it matured, largely abandoned the 3, there really was no continuity or shining example of success via the 3. Occasionally a player got hot and won a playoff game or series. I'm thinking of the Pacers' Bill Keller; but he was a .338 career shooter.

I can't imagine any coach that would not field a sound player who also made 2 of 5 threes on a typical night. There just wasn't any such player, back in the day.

[Statman]

I can't imagine any coach that would not field a sound player who also made 2 of 5 threes on a typical night. There just wasn't any such player, back in the day.

Why is that BTW? You always hear old timers (ie, older than me) talk about past players with great effective range that would have killed it in today's 3pt game. Yet, the ABA and early 3pt NBA didn't show much. Is the talk of the great shooting of the past players hyperbole?

Also - maybe the none break away rim was a factor - absolutely no margin for error on the shot? Maybe today's great three point shooters would lose .05 to .1 off their 3pt % with those rims/backboards.

[Mike G]

A shooter is 'great' relative to others in his league. A guy may be having a hot night and actually hit shots beyond what is the normal range, in a league without 3-pointers. But nobody did it in the ABA on a regular basis, at an effective accuracy.

The ABA wasn't known for overall great defense, so there were likely better shots available, in general.
The 3 became known as rather a gimmick, a tool of the underdog. Occasionally it won a game, but not consistently enough to establish a trend.

[schtevie]

If coaches would have built offenses that recognized the potential of the three point shot, the players would have come, just as we've seen and are still seeing in the NBA.

They'd have come from where? Once they instituted the 3FG in college, they came from college. But in the ABA, only 11 players in their careers attempted as many 3's as Steph Curry did last year (600). Only 3 of these managed to hit the 2013 NBA avg 3FG% (.359)..

I know this is a serious question, but yet I'm surprised. There is no magic here. They, of course, would have come from the existing player population (and incrementally from incoming cohorts). If coaches incentivize players to both acquire certain skills and structure their offenses such that these skills are seen to pay off, it should only be expected that players would deliver.

And yet I hesitate to pursue this line of inquiry. Why? Because it is diversionary. The experience of the ABA and early NBA experience with the three point shot clearly shows that the existing, "raw" talent pool was sufficient to exploit opportunities that existed. But it was not so employed. This should be the end of the story.

But OK, let's pursue the diversion a little bit. There was a natural NBA experiment that speaks to this very issue: the three year, mid-90s NBA effort to bribe teams to take more three point shots. If we believe that old dogs cannot be taught new tricks we should expect to see no positive returns to three point shooting on a cohort basis between the before year (1993-94) and the after (1997-98). But what do we see? Well, if you look at the medians of the respective age cohorts, following each across time, you see that 3PAs increased without a drop in efficiency (what means high quality shots were being substituted for inferior ones on the margin). Alternatively if you compare the medians of identical age cohorts across the four year gap, you see the same: increasing 3PAs without a drop in efficiency.

Should we be surprised by this?

Because the ABA started out as a 3-happy league, which as it matured, largely abandoned the 3, there really was no continuity or shining example of success via the 3. Occasionally a player got hot and won a playoff game or series. I'm thinking of the Pacers' Bill Keller; but he was a .338 career shooter.

I can't imagine any coach that would not field a sound player who also made 2 of 5 threes on a typical night. There just wasn't any such player, back in the day.

Again, I am bewildered by the characterization. The ABA started out as 3-happy a league as the late 80s NBA: #nothappy. And it never matured, in the sense of tending toward optimal strategy.

But again, again, this is not the point (at least of this post). The point is that the ABA experience should have been the first opportunity to inform the views of Phil Jackson and Jerry Sloan and so many others as to the potential of the three point shot for NBA basketball, and it clearly didn't serve that purpose.

[schtevie]

I took another pass through the data regarding the aforementioned natural experiment. Why? Cuz it's really rather important to get this story straight. The NBA moving in the three point line was a critical point in the evolution toward the modern game (and not just regarding three point shooting, but that is a story for another day). Rather than just get an impression by looking at the medians of cohorts, I spent a little more time with basketball reference's database to better address the question of whether the NBA was supply-constrained in more rapidly incorporating three point shooting into its average offense.

There are basically two, competing explanations as to why it has taken fifty-ish years to realize its current share of threes in the average NBA offense. The first is that supply has always been a bottleneck, that fresh, three-point shooting blood from the college (foreign?) ranks (and the retirement of the existing dead wood within the league) was required to raise the share (and realized efficiency) of NBA threes, and there was only so much of that. The alternative view (my own, of longstanding) is that the fault lies almost entirely with management: supply has always been latent; it simply was never called upon.

Here are the relevant numbers for the before (1993-94) and after (1997-98) seasons of the experiment with a shorter line:

In 1993-94, there were, on average, 9.9 3PA per game. Of these, 8.5 (or 86% of the total) were taken by players who would be in the NBA in 1997-98.

In 1997-98, by contrast, there were, on average, 12.7 3PA per game. Of these, 8.1 (or 64% of the total) were taken by players who were in the NBA in 1993-94.

But, of course, over this time the pace of the game slowed (95.3 to 90.3 per 48 minutes) and if you do a simple adjustment for the relative number of FGA between years, you find that the 8.1 3PA per game in 1997-98 would be the equivalent to 8.6 in 1993-94. (And adjusting further for minutes per game would be another refinement, but not really required for present purposes).

So, the bottom line on attempts is that the continuing cohorts held their own, and all of the increase in 3PAs came from the new blood (those entering in the as rookies between 1995-96 and 1997-98) relative to the old.

But what about efficiency? Overall, 3P% increased from 0.333 to 0.345 (eFG%: 0.499 to 0.518). But what explains this? Who was bringing the gain? Was it an increase in efficiency of the continuing cohorts, or alternatively the relative efficiency of the retiring vs. the incoming ranks?

As for the first factor, those who played through, their 3P% increased from 0.330 to 0.349 (eFG% increasing from 0.495 to 0.524).

And this reveals what the story is for the old vs. the new: the eventual retirees shot threes in 1993-94 at a rate of 0.352 (eFG% of 0.527), whereas the young players in 1997-98 only shot 0.339 (eFG% of 0.508). So, it was not the case that the oldsters were holding back progress.

Case closed then, no?

The NBA moved the line in in 1994-95, and front offices/coaching staffs (and to lesser extent players), stirred from their torpor, were suddenly more obliged than before to confront the opportunities that the three point shot conferred. Players began to practice long-distance shots more; and offenses began to be reconfigured to better find players in those more profitable locations. Then, come 1997-98, when the line was returned to its original position, voilà: players magically had the ability to hit the shot in increasing numbers and increasing efficiency, despite the callowness of youth.

And a final note. It should also be appreciated that the defensive pressure that long-range shooters faced in 1997-98 was distinctly greater than four years previous (what was, of course, a consequence of the experience with the shorter line.) Defenses improved generally, and the realized increase in 3P% should be seen in the light that not all else was equal. But despite these accomplishments, foresight remained lacking, and we continue the long, asymptotic slog to the glorious future.

Same as it ever was.

[schtevie]

P.S. Just to bolster the idea that the basketball universe is an orderly one, let me also add the cohort exit and entry data for three point shooting efficiency.

So for players who left the NBA during the mini-era of the shorter three point line, here are their eFG%s in 1993-94 (going in chronological order): 0.506, 0.522, and 0.548. This compares to average of the remainder of the 1993-94 player list of 0.495.

So, retirement was anticipated by declining efficiency from the three point line, but still these older players remained above average.

Conversely for those players who had joined the league during same mini-era as well as who entered as rookies in the first year the line returned to its original position, the eFG% from the three point line in 1997-98 was as follows (rookie year to fourth year): 0.462, 0.491, 0.501, and 0.557. And this compares to the remainder of the league, whose average was 0.524.

So the younger cohorts saw increasing efficiency, and by the fourth year, this class surpassed the veteran average.

Perhaps clearer than the truth, but as expected.

[TheSpiceWeasel]

Have you ever thought about coaching? Your team would go undefeated!

[schtevie]

Have you ever thought about coaching? Your team would go undefeated!

Excellent idea! No, wait. Then I would be compelled to ignore my own advice. Too problematic.