A simple theory of some current basketball surprises

Apply a dose of science and big data to a team sport such as basketball.  The big gains will come in cooperation.  Who should take the next shot?, when is a “corner three” worthwhile?, who should play with the second unit, how good is the pick and roll against this opponent?, and so on.  Big data also will bring some gains at the individual level, such as from better training regimens, but those moves were easier to spot in the first place.  The issues involving cooperation are those where simple intuitive observation, of the old school style, will miss a lot of potential improvements.

Cooperative gains are more fragile, however, because everyone has to get the strategy right to reap the benefits (think of Michael Kremer’s O-Ring model).  So the previous champion, San Antonio, has fallen off dramatically because Leonard is injured and Tony Parker is playing like his age (32).  Atlanta suddenly had all the pieces gel, and they now, to the surprise of almost everyone, have the best record in the East.  (They have learned the ball movement and shooting style which San Antonio perfected last year during their championship run, but Atlanta has no big stars.)  Golden State is a positive surprise too, with the best record in the league.  Cleveland has attempted to do “cooperation” (ha) on the terms of its stars, not on the terms of the data, and that experiment has fallen flat.

In Panama I watched an old Lakers game from the 1980s (vs. Portland) and was struck by how tall everyone was, compared to today.  There were fewer surprises that year, and I believe those facts are related.  The three-point shot has made players shorter and more cooperative and arguably increased the value of the coach and his assistants.

Some of these arguments should apply to areas other than basketball, so perhaps a higher value for data-driven cooperation will mean more surprises in the world in general.


Due to Big City bias, I doubt in the end though Atlanta will take it. I'm surprised San Antonio held together as much as they did. Typically refs will call the game to favor Big Cities like NY, LA, not San Antonio, Atlanta, Golden State (where are they? Oakland still? LOL I used to live near there and I still don't know). It's the way big sports work.

Nonsense. They might make calls for superstars or high-profile teams (Detroit fans certainly thought that in '88), but not cities. Do you really believe the NBA would prefer the Knicks over the Cavs in this year's finals?

Not a good retort, as the Knicks obviously won't be in the playoffs to tempt the NBA to fix things for them. If the Knicks were a good team they might get a little extra love from the NBA.

Really, it's not? So, when the Knicks have been in the playoffs three of the last four years, where was their edge? You can say, "Well, the NBA wanted the Heat and the Celtics more." But the no-name Pacers? And none of those games had a margin of more than 10 points, so there was ample room for favoritism to shift the game to NY.

Also, what is the NBA's motivation for favoring bigger cities? OKC-Miami was their highest rated five-game series since the games moved to ABC. The lowest rated series? Spurs-Nets, despite the the NYC market. What really brings in the ratings is a Game 7.

Oh wait, before somebody jumps on me, there was one 5-game series that was rated higher - Lakers-Pistons.

Caveat: In the 80s the rules permitted much more concentration of great players in a same team, causing less surprises.

If your decisions are "data driven" then everyone else will know what your decision will be. The comparative advantage will disappear and it will again come down to individual skill and experience. Oh, it will also make failures more spectacular.

Game theory strikes again! (pun intended)

"So the previous champion, San Antonio, has fallen off dramatically because Leonard is injured and Tony Parker is playing like his age (32)."

As with all sweeping judgements, this one probably is somewhere on the wrong side of the continuum. Injuries have taken a toll on the Spurs and Parker has missed about the same number of games as Kawhi had. Even after he came back, Parker seemed relatively listless in the games he played during this last week. However, yesterday, against the Hornets, he suddenly seemed to come alive and play like the Parker we all know. The Spurs coaching staff (worth their weight in gold) are keeping a tight grip on the minutes of players - possibly even their effort - after they get back from injury. For heaven's sake, they are handling even Belinelli with kid-gloves. Anyway, I spend way too much time analyzing the Spurs.

Tyler, you left out the best example of the importance of cooperation: Josh Smith and the Pistons!


Grantland recently had a data-centric article on James Harden.

IIRC, Tyler linked to that article a few days ago. So I think Tyler's keeping up, sort of, with the advances that are being made in NBA analytics. But only sort of, and he frequently mis-apprehends or mis-applies the new evidence. The posts of triclops and Ben directly below this one give examples, there's also Tyler's overrating of the Nets prior to last season.

OTOH Tyler's overall point about the increased importance of cooperation is almost certainly correct. The article about Harden focused on offense but Goldberry and other authors have written many articles about the complexity of playing team defense in today's NBA. So all aspects of the game have become more team-oriented and cooperative in nature. This may indeed imply that coaching staffs are relatively more important and that predictions now have larger standard errors, but that's an empirical question.

NBA players are shorter now? That is not my experience. Back to the basket centers have almost disappeared, but that hasn't had a huge impact on average height. Perhaps watching a team that had, by far, the tallest point guard, as well as an unusually tall center, might have biased your perception here.

And I'm not sure that those perspectives on the struggles of the Spurs and Cavs are great. The Spurs have played like this for several years now, pacing their aged stars for the playoffs, with great success.
And I don't think metrics can do much to help the fact that the Cavs don't have any good interior defense. Metrics are most useful when you have pieces you can do interesting things with. In the case of Cavs defense, the best you can hope for is a least terrible option, which isn't that much better than a random assortment.

This post is full of easily verifiable nonsense. Basketball Reference tracks reported height and weight in NBA going back to 1951 and the reported average height has been 6'7" since 1980. Specifically, the 2015 Lakers roster averages an inch taller than the 1980 team. Perhaps the players from 1980 look taller to Cowen because they are so much skinnier. The reported average weight is up from 205 lbs to 218 lbs, but my guess is the actual difference is much greater as reported weights today significantly err on the low side, likely more than in 1980.

The three point shot has not "made players shorter", rather there are more and taller people taking the shot now. The question of whether there are more "surprises" is checkable by comparing actual season win-loss records vs. pre-season Vegas over-unders. I doubt it is that much different today than it was back then.

Reading this post reminds me of the Gell-Mann Amnesia effect.

"Perhaps the players from 1980 look taller to Cowen because they are so much skinnier "

The post I came here to make. Also, because the shorts were so much shorter.

+1, I chalk it up to the shorts. Everyone's legs look longer in short shorts.

fwiw: centers and PGs are shorter now than for most of the 80s; wings and PFs a bit taller. The massive stiffs have all but disappeared (the price of having an offensive zero is huge in a modern offense). Agree with the weight theory, and I also think an increase in athleticism across the board disguises how tall they truly are.

So you have 1 guy who is obviously shorter today and 4 guys of more similar height, probably all more athletic and moving more fluidly. I can see how that would appear as shorter on average even if it's wrong numerically.

But players, especially tall guys, now come from all over the world, so the fact that average NBA height hasn't increased much in a generation means that it has fallen relative to expectations.

In 1981 the back of a napkin history of the NBA went like this:

1940s: George Mikan: 6'10"
1950s: Bill Russell: 6'11"
1960s: Wilt Chamberlain: 7'1"
1970s: Kareem Abdul-Jabbar: 7'2"

and in the near future:
1980s: Ralph Sampson: 7'4"

But that last didn't really happen.

1990s- Shawn Bradley, George Muresan, Manute Bol: 7'6"- 7'7"

2000s- Yao Ming: 7'6"

The trend has only stopped very recently.

Anyone care to elaborate on how the Cavs' approach to play is on "the term of its stars" rather than "the terms of the data"? I would agree completely that the roster-building left a bit to be desired (too many of Lebron's aging ex-teammates) and that the cooperative team play has been weak, but it's hard for me to see how to describe this as a failing of a star/data dichotomy. If anything, I would argue that Cleveland has a major talent deficit at every rotation position except for Love/Irving/James, and even then, only James is a top 20 player.

Compare to OKC, which has a similar talent deficit outside of Durant/Westbrook/Ibaka, but Durant and Westbrook are both top 10.

Lots of stagnant possessions where the play is 1) LeBron walk the ball up the court 2) high PNR (this is not the system Blatt played in Europe at all, fwiw) with the other three players stationary spacers 3) if this breaks down, bad isolation. The Cavs suffer from a major talent deficiency on defense, but their offense is far underperforming expectations. Lots of chatter about the Cavs not respecting Blatt, and LeBron's body language is horrific.

Agree with you 100%, just unsure how to fit this into star/data narrative.

re the spurs, there's another way they don't support the hypothesis -- they had about the most brutal month of play imaginable (given opponents and back to backs in december), and many of the games still were down to a coin toss (effectively) that could have resulted in four more wins (which would move them up the standings in the west).

I think Tyler's point about Atlanta is pretty right, they're taking advantage of newish ideas about spacing and optimal shot selection. But of course they do play in the East (86-142 right now vs. the West) so it's not like if they get to the finals they're guaranteed to get even one win against the West winner.

On the other hand, I think the "collapse" point is somewhat eccentric.

Games missed last year (out of 82) / player / games missed this year (out of 39):

14 / Parker / 13
16 / Leonard / 17
2 / Belinelli / 10
14 / Green / 0
3 / Diaw / 0
8 / Duncan / 4
1 / Mills / 30
23 / Splitter / 21
14 / Ginobli / 5

95 / all 9 / 100

Last year's point differential: +7.7. This year: + 3.6.

There is the sexy new big data but you can still learn a lot just by looking at the boring old small data for 5 minutes.

Conference dominance is fairly overrated, in the finals anyway. You could say the West dominated from '99-'11, but really it was the Lakers and the Spurs. The same goes for the East from '89-'98.

Tyler, I can tell you're an NBA fan since you've posted a fair amount on basketball. You really must check out the Youtube channel Bballbreakdown which has great video breakdowns and statistical analyses (e.g. ranking power forwards on pick-and-roll vs. pick-and-pop efficiency, etc). Very professional production values.


The details are wrong, but there is something about the cooperative play being a bigger deal in the NBA now. I think pretty clearly it is not just the 3-point line but the allowing of zone defenses (which actually leads to more team help and soft doubling). Until very recently an offense was mostly just the sum of its parts with a heavy bias towards the contribution of the best two or three players on the court. Now the offense is a much greater function of how the players complement each, the quality of their offensive scheme and the competency with which they execute it.

I would argue the issue with the Hawks is that they are not perceived to have any big stars. In fact, their starting 5 could basically form the all-underrated team on their own. Korver is one of the greatest offensive threats in the game, Horford is a top-5 center nobody realizes is a top-5 center (partially due to injuries), Millsap is so underrated it's become a cliche, Carroll is a classic 3 and D fit guy, and so on.

Perhaps the issue is that perception has not aligned with reality and won't until a deep playoff run or championship occurs, kind of like how Detroit did the "impossible" with the Wallace-Wallace-Billups team, but afterwards everyone suddenly agreed they were cooperatively great.

@Austin, very true. Teague has turned into quite a player too, but is never mentioned in point guard discussions. 5 really good starters who play off each other very well.

Could this not all be reduced to the old platitude that "chemistry" matters a lot. With chemistry being some residual of cooperation that is not captured in the standard statistics.

While the average NBA height has gone every so slightly down over time, that magnitude is largely explained by the fact that Manute Bol is no longer in the league.

What's the least valuable thing on earth? A first quarter NBA score.

Comments for this post are closed