The Statistical Analysis We Need

I was looking at this latest iteration of efforts to use adjusted +/- statistics to evaluate NBA players, and it served as a reminder of how frustrating I find it that such a large proportion of efforts to apply quantitative tools to the analysis of basketball are dedicated to these searches for magic formulae to assess player quality. There are other, more interesting and probably more fruitful, lines of inquiry where quantitative skills could shed some light.

For example, there's a popular conception of a link between pace and defensive orientation -- specifically the idea that teams that choose to play at a fast pace are sacrificing something in the defense department. On the most naive level, that's simply because a high pace leads to more points being given up. But I think it's generally assumed that it holds up in efficiency terms as well. The 2006-2007 Phoenix Suns, for example, were first in offensive efficiency, third in pace, and fourteenth in defense. But is this really true? If you look at the data season-by-season is there a correlation between pace and defense? When pace changes leaguewide, does scoring efficiency also change? Then there are lots of interesting team level issues to ask. Intuitively, some teams' offenses are optimized for the fast-paced style and will function less efficiently during games that wind up being played at a slow pace. And vice versa also probably holds. But are there some teams who are making a mistake? Squads who score more efficiently when they play slower, but usually try to play fast?

I'm too lazy to actually conduct research into those questions, and I'm not even sure I know how to calculate a coefficient of correlation correctly these days, but I'd read someone who wanted to do it.

Comments

On a different note; I like college and pro ball. However, I don't think you need to tune in to an NBA game until midway through the fourth quarter. Why is it there is no such thing as a safe led in the NBA? It can be 89 to 72 with 7 minutes left and you're somewhat assured the team on the short end of the stick still stands a punchers shot at pulling it out. If invested in either you're nervous your team may eventually lose, but you're especially concerned your team will commit a monumental collapse and blow that 17 point lead. What the hell is it NBA teams can't protect leads (even HUGE leads)?

Posted by steve duncan | November 29, 2007 1:56 PM

Steve -

It might be because the marginal differences in talent from team to team are very small, especially for the upper-echelon teams. In college basketball those differences are a bit larger, I think.

Posted by Kurzbein | November 29, 2007 2:31 PM

Kurzbein, if marginal differences are small how did one team manage to get that 20 point lead to begin with? Continuing that thought how does the deficient team (so "marginally" close to the score-leading opponent) manage to play with devastating superiority for a scant few minutes and wipe out that 20 point lead? Where was that level of play for 3-1/2 quarters? And why is the leader laying down when they've just demonstrated the ability to lay waste to their opponent? I can't see the reasoning each team is but a degree of small fractions better or worse than their opponent explaining this phenomenon. As much as I like the NBA it is the one aspect of it that really pisses me off. I don't even like it when "my team" is the one coming back when the tradeoff is some crew looking like idiots for their inability to hold a damn lead. A more suspicious mind might start looking at betting spreads and try to put two and two together.............

Posted by steve duncan | November 29, 2007 2:42 PM

matthew, i have no idea where this "generally held" idea is "generally held" (or even if it is), but it is most assuredly not universally held.

the classic fast-breaking champion, the russell celtics, were known to be a pretty efficient defensive team, and the most reason example of a fast-breaking champion, the showtime lakers, were also a damn good defensive team.

pace affects the score: it doesn't mean that a fast-paced team can't also maintain defensive excellence.

Posted by howard | November 29, 2007 2:55 PM

and it served as a reminder of how frustrating I find it that such a large proportion of efforts to apply quantitative tools to the analysis of basketball are dedicated to these searches for magic formulae to assess player quality

That probably because we are most interested in wins and losses, and usually think of the best teams has having the best players.

But if any statistician is taking requests, I would like to see the correlation, if any, between spending on player salaries and regular season win percentage in the 3 major sports.

Posted by Al | November 29, 2007 2:56 PM

But if any statistician is taking requests, I would like to see the correlation, if any, between spending on player salaries and regular season win percentage in the 3 major sports.

I believe this is the question that started Berri down his primprose path to idiocy - the WAGES of WINS, after all. I think he and his collaboraters started out with this question (and maybe a few others) and concluded that there wasn't much correlation (in basketball, at any rate).

Posted by Quarterican | November 29, 2007 3:09 PM

I too don't know whether "fast pace = bad defense" is universally held. I was also thinking about the Showtime Lakers, in that regard. However, I'd note that they were a better offensive team than they were a defensive team, even if we accept that they were a good defensive team. (I think if you look at the '85-'89 period, the Lakers were always the most efficient offensive team in the NBA, while defensively they were good but not the best.)

But even if we were to accept that faster pace means worse defense - that's only half the story. Also potentially faster pace means better offense. (And, hell, who knows, maybe faster pace means better defense or worse offense...) The question would be whether, on average, the loss in defensive efficiency is outweighed by the gain in offensive efficiency.

Posted by Al | November 29, 2007 3:09 PM

I believe this is the question that started Berri down his primprose path to idiocy - the WAGES of WINS, after all.

Huh. Who knew. Maybe I should buy the book. [Heads exploding.]

Posted by Al | November 29, 2007 3:12 PM

As ever, I make the simplest mistakes when insulting others. Primrose, not primprose.

Is there any reason to a priori suspect there's a positive correlation between pace and efficiency? It's just as likely that playing offense at warp speed will lead to a lot of mistakes as it is that it'll lead to running all over a slowfooted defense.

Posted by Quarterican | November 29, 2007 3:16 PM

Steve -

You're assuming that talent or athletic performance gets displayed in some static, evenly-distributed fashion. That's almost never the case. Add the dynamics of pro basketball like substitution patterns and the resulting matchups, and you have volitility within a given game but a close final score.

I'm certainly not opposed to the view that the nature of NBA basketball is such that players and teams dose their efforts, and that dosing explains the big comebacks. But relatively equal talent plays a role in big comebacks too.

Posted by Kurzbein | November 29, 2007 3:17 PM

Steve Duncan: If Team A has a 17-point lead after 41 minutes, that doesn't necessarily imply "devastating superiority". Maybe it means that the two teams played pretty evenly for say 37 minutes, but Team A had a nice 2-minute stretch in the second quarter where they scored 8 points in a row, and another nice 2-minute stretch in the third quarter where they scored 9 points in a row. Basically, they got lucky. Random variation and all that. And because the teams are evenly matched, it's well within the realm of possibility for team B to go on a lucky hot streak.

Posted by nbt | November 29, 2007 3:25 PM

"I can't see the reasoning each team is but a degree of small fractions better or worse than their opponent explaining this phenomenon." ...Steve

I think it is a question of intensity. Asking a basketball team to play at the same level of intensity for 48 minutes is like asking a baseball player to hit 1.000. Ain't gonna happen.
Playoff-bound teams & franchise players will gain & lose leads, but are able to turn on whatever is needed in the last minutes and win the game. Obviously these differences in clutch play or talent are more than a small fraction since people like Russell, Jordan, or Duncan have multiple rings.

Now whether that ability to win games with offensive or defensive spurts is identical with being a generally good offensive or defensive team is another question.

Posted by bob mcmanus | November 29, 2007 3:27 PM

Kurzbein basically made my point, more eloquently.

Posted by nbt | November 29, 2007 3:28 PM

Now that we're all warmed up with the basketball analysis, let's discuss how Jews went from allegedly dominating pro basketball 85 years ago, to mostly just watching it today. Jews comprise about a third of NBA owners but only 0.25% of NBA players (and that player is half black). Since everyone here knows that hereditary differences in any abilities between different racial groups are negligible, what gives? Why did Jews cede the hardwood to African Americans so completely?

Posted by Fred | November 29, 2007 3:42 PM

What about the Arabs? Rony Seikaly and Alaa Abdelnaby once roamed the painted area, but now, nuthin.'

Posted by nbt | November 29, 2007 4:08 PM

If anyone has done it, these would be the guys:

http://www.basketballprospectus.com/

Posted by Jar Jar Binks | November 29, 2007 4:25 PM

Bill James spent decades doing small studies of baseball statistics, only turning to his holy grail win shares after a quarter of a century. As Matt suggests, basketball research might get ahead faster with less ambitious studies for now.

Posted by Steve Sailer | November 29, 2007 4:47 PM

"I'm too lazy to actually conduct research into those questions, and I'm not even sure I know how to calculate a coefficient of correlation correctly these days..."

It's a shame you're not a highly paid pundit, because then you could give us a definitive answer in spite of this.

Posted by Alan in SF | November 29, 2007 5:02 PM

Steve-
Although issues like intensity might be important, I'm pretty sure this is just a question of statistics. If you flip a coin 100 times, you will not alternate between heads and tails every time. At some point you are likely to have a streak of either heads or tails. Two streaks of five in basketball would translate to about an 18 point lead. Then you're just as likely to have two streaks of five for tails, although you'd obviously still rather be on the heads team if there are only a few minutes left in the game. This means that the talent parity between teams actually perfectly accounts for the frequency of blown leads. They're just one of those things that can happen in a 50/50 situation. By this logic, your frustration can only be alleviated by reducing the parity between teams (the Spurs rarely blow a 20pt lead to the Bulls, eg).

Posted by Ignatz | November 29, 2007 5:05 PM

Pace impacts gross defensive stats (points allowed). It seems to me the true measure of defense should be a percentage of stops. By this I would try to give credit for steals, turnovers forced and defensive rebounding in addition to the more tranditional measure of field goal percentage against.

Posted by phg | November 29, 2007 5:25 PM

Pace makes defenses impossible to compare, because one of the things that faster teams are trying to do is to use their defense to create offensive opportunities. The classic examples of this include John Wooden's UCLA teams and Paul Westhead's Loyola Marymount teams, which used aggresive defense to force turnovers that would lead to fast break baskets. The most successful Los Angeles high school teams do this too and have for many years.

The point is, these defenses aren't operating on the same principles as others. If you are going for a lot of turnovers, you are also going to be giving up a lot of baskets when the defense gets beat. But it still may be a net benefit because of the easy fast break baskets that you get on your end, the benefit of forcing the other team to play your tempo, and whatever conditioning edge you have. So how do you compare a defense that has that philosophy to, say, Ben Howland's defensive philosophy at UCLA where he wants to stop every single shot and slow you down?

There's just no statistic that is going to bridge that gap other than wins and losses.

Posted by Dilan Esper | November 29, 2007 5:26 PM

So, is basketball susceptible to analysis of the kind that SABRmetrics people have done for baseball, and the sport is just waiting for its Bill James?

Or is the problem that it's too fluid a game? And/or there are not enough adequate, detailed, and accessible statistics needed?

Hockey fans talk about this a bit, mostly (for now) settling on the fluidity thesis.

Posted by Nell | November 29, 2007 7:19 PM

Nell, i think there are things that we can learn from deeper statistical analysis of basketball, but i too agree that the fluid/flow aspects of the game makes it hard to achieve the same kind of fine-grained analysis that baseball, with its discrete "pitches" and "plays," can be subjected too.

soccer, too, has this issue.

in addition, there are some things - setting a good pick, moving well without the ball, blocking out effectively - that are somewhere between difficult and impossible to pick up with current statistical methods: you'd have to have some new kinds of data-collection to pick them up.

Al, the more the showtime years went by, the better the Lakers got on defense, and don't forget that they had michael cooper, the pre-eminent man defender of the period....

Posted by howard | November 29, 2007 8:29 PM

Re Steve Duncan's question above, and contra my earlier answer, I would say that if you find your team down by over 50 points in the fourth quarter, then yeah, you might be victim to "devastating superiority".

Posted by nbt | November 29, 2007 11:54 PM

I'm on repeated record as saying that basketball is unlikely to be amenable to full SABR style efforts because of the reasons mentioned by Howard. There is the additional complication that even if we had a measure of such nice things as picks, and accuracy/timing of swing passes and so on is that whereas in baseball a second baseman is pretty much a second baseman,* this is not really true in modern basketball and where a PG for say the Suns is a very different role than the PG for the Lakers, and this is only partially the result of the differing talent levels of Steve Nash and Derek Fisher.

*though see Bill James' Historical Abstract entry on Bill Buckner, I believe for a discussion of the effects of different styles of play at a position may show up in analysis of raw fielding stats

Posted by Pooh | November 30, 2007 4:06 AM

A while back, when I had more free time, I looked into this exact question. I only used stats from a single season, because exporting the stats from web pages where I could find them, into excel where I coul do sometging with them, was a real PITA.

Anyway, with a limited sample set, there was little to no correlation between pace and defensive efficiecy. If my memory serves, though, there was a correlation with offensive efficiency, which suggested that playing at a higher pace was beneficial.

As a side note, the one other interesting thing I found was that there was almost no correlation between team shooting percentage and the percent of team's baskets that had assists. So the typical assist provides almost no benefit to a team. The one exception there was the Utah Jazz, who had an unusually high FG% and assist numbers.

Posted by Doug T | November 30, 2007 9:43 AM