Scorecasting

[Griffin]

Now that we have some time off, perhaps a lot of time off, before the Suns play again, how about some reading material:

Scorecasting: The Hidden Influences Behind How Sports Are Played and Games Are Won
By Tobias J. Moskowitz and L. Jon Wertheim

Anyone read this book? I just finished it and highly recommend it. Basically, the book attempts to statistically analyze various long-accepted sports myths and/or assumptions to see whether they are really true. Very similar to what Freakononmics did, but applied to sports. There is no focus on any particular sport or league, but the major US sports leagues are used as primary (but not only) examples.

Here are some of the results they have obtained and conclusions drawn:

• There is no statistical proof that defense wins championships. According to data, both offense and defense are just as likely to contribute to a team winning a title or winning in general, in all sports. Defensive-minded teams win titles, playoff games, and regular season games as often as offensive-minded ones.
  
  
• Home-court advantage is indeed real as home teams have a significantly higher winning percentage than road teams across all sports. However, the authors found that the main reason for home-court advantage is referee bias. Referees are much more likely to make calls favoring home teams, especially when the game is on the line. Referees simply, on average, make the call that tends to "agree" with what the majority of people think is the right call which happens to be the home crowd. Another reason given for home-court advantage which they found mostly only applies to the NBA is schedule bias. Road teams play a lot more back-to-back sets than home teams. Also, better teams usually have more favorable schedules.
  
  
• Having a superstar or two on your team makes a huge difference in winning a title, especially in basketball.
  
  
• High draft picks are often overvalued and future picks are often undervalued.
  
  
• Icing a player by calling a timeout before a free throw or a field goal attempt does not work.
  
  
• There is no such thing as momentum. Not on individual level, not on team level. A player who has made 10 baskets in a row is just as likely to miss his next attempt as if he missed ten in a row. In fact, players often are more likely to miss because the more shots they make, the more difficult shots they start taking. A team that has gone on a hot streak at the end of the game to force overtime is just as likely to win or lose as a team that blew a large lead. A team on a long winning streak is no more likely to win its next game than the same team on a long losing streak, etc.
  
  
• There are no curses. The Chicago Cubs are used as an example here to show how their failure to win a title in over 100 years can be attributed to mostly lack of sufficient incentive. Cubs are able to consistently sell tickets year after year regardless of whether the team wins games or not, so there is no financial incentive to make the team more competitive.

There's some other interesting stuff in there like the analysis of the value of a block shot indicating some blocks are a lot more valuable than others. While this may be rather obvious, they compared Dwight Howard's (DPY three years in a row now) blocks to Tim Duncan's and found that Duncan much better at getting more value out of his block shots than Howard. There's also an extensive discussion of psychological factors that influence fans, players, coaches and managers including loss aversion.

 

[BC867]

There is no statistical proof that defense wins championships. According to data, both offense and defense are just as likely to contribute to a team winning a title or winning in general, in all sports. Defensive-minded teams win titles, playoff games, and regular season games as often as offensive-minded ones.

Thank you for sharing the highlights of the article. 'Interesting statistical points.

Regarding defensive-minded teams vs. offensive-minded teams winning playoff games and titles, I come away feeling that it is probably true - - as long as the team is not deficient in either.

The Suns don't have to be a defensive-minded team to be an elite team in the post-season. Just not deficient at it.

Or, as we had to put up with D'Antoni, listening to him joke about the Suns playing defense.

 

[AzStevenCal]

Thank you for sharing the highlights of the article. 'Interesting statistical points.

Regarding defensive-minded teams vs. offensive-minded teams winning playoff games and titles, I come away feeling that it is probably true - - as long as the team is not deficient in either.

The Suns don't have to be a defensive-minded team to be an elite team in the post-season. Just not deficient at it.

Or, as we had to put up with D'Antoni, listening to him joke about the Suns playing defense.

I didn't really have a problem with Mike's approach to the NBA game (the first few years anyway). It was a fairly novel approach and it was nice to see us really focus on putting the best players out there by design as opposed to our normal patchwork small ball. With a little bit of luck it could have even won us a championship. My biggest beef with Mike is that he clung stubbornly to his model rather than modifying it once it became obvious that the longer NBA season and the postseason referee style warranted a few changes.

The SSOL style was fun and was probably a great thing for us early on but once we'd had a few years of being among the league's elite, we should have moved towards a more balanced lineup. I agree with this article that defense doesn't win championships any more than offense does but not having the ability to get stops in critical moments can definitely cost you a game, or a series, or even a season.

Steve

 

[Errntknght]

I may have to get that book and see how they come to their conclusions. Just because someone concludes something from statistics doesn't mean they're correct - people with an agenda conclude whatever they damn well please... i.e. they choose which data to ignore and which to emphasize and what statistical method to employ. If none of that works they simply proclaim the desired result anyway.

I'm not claiming the myths are correct but I have some doubts about momentum - it sure seems like it plays a role. For years statisticians (or people employing statistics) have claimed there are no such things as hot streaks in shooting. I know perfectly well that random data typically shows more 'runs' than we expect so we conclude its not random but I'd love to get my hands on the raw data and analyze it myself. I wouldn't be surprised to come to the same conclusion but I also wouldn't be shocked to find a way of analyzing the data that indicates there are more hot (and cold) streaks than random data shows.

I'm not convinced about the absolute parity of offense and defense in winning titles in the NBA. How did they measure them for one thing? Did they actually show, for example, the parity exists for all games or just use the the playoff games. Someone might conclude that since offense and defense have parity in point differential computations and point differential correlates highly with winning games, offense and defense are equally important. (In fact, that's not a terrible argument and I use it internally to arrive at the belief that defense isn't that much more important than offense.)

But there is an argument that there is slight difference - the idea that offense falters more than defense does as pressure mounts because of the fine motor skills that are involved in shooting.

 

[mojorizen7]

There is no statistical proof that defense wins championships. According to data, both offense and defense are just as likely to contribute to a team winning a title or winning in general, in all sports. Defensive-minded teams win titles, playoff games, and regular season games as often as offensive-minded ones.

BS

 

[95pro]

data used and how its used, is one thing in determining the statements made in this book. ahhh, the art of statistics...

 

[dreamcastrocks]

There is no such thing as momentum. Not on individual level, not on team level. A player who has made 10 baskets in a row is just as likely to miss his next attempt as if he missed ten in a row. In fact, players often are more likely to miss because the more shots they make, the more difficult shots they start taking. A team that has gone on a hot streak at the end of the game to force overtime is just as likely to win or lose as a team that blew a large lead. A team on a long winning streak is no more likely to win its next game than the same team on a long losing streak, etc.

Not buying this one AT ALL.

 

[elindholm]

This kind of book is, unfortunately, preaching to the converted. Either you believe in the authority of objective data over hunches, or you don't. Those that trust their gut in the face of factual evidence are going to continue to do so.

 

[AzStevenCal]

Not buying this one AT ALL.

I thought that paragraph was poorly worded and I'm not sure the specifics warranted the conclusion that there is no such thing as momentum. Or, maybe I should say, I don't think momentum is proved or disproved solely (or even primarily) by shooting statistics. I think momentum affects energy levels, confidence, effort, referee calls along with many other aspects of the game. I guess I can accept that just because someone has made 10 straight that momentum won't necessarily increase the likelihood that the next shot goes in.

Steve

 

[elindholm]

I guess I can accept that just because someone has made 10 straight that momentum won't necessarily increase the likelihood that the next shot goes in.

Even then, I wouldn't be surprised if "momentum" does play a factor, because players can be more relaxed and in balance when they are confident and "feeling it," and those subtle physical differences can (and most likely do) lead to improved accuracy. (Similarly, I think there is little doubt that "choking" is a real phenomenon for some athletes, because tension can interfere with their breathing and rhythm.) But I wouldn't reject the authors' conclusions just because they disagreed with my subjective impressions.

 

[Errntknght]

This kind of book is, unfortunately, preaching to the converted. Either you believe in the authority of objective data over hunches, or you don't. Those that trust their gut in the face of factual evidence are going to continue to do so.

I believe in objective data but I haven't seen that... people's conclusions drawn from their viewing the data is something quite different. I'm not pronouncing them wrong without reading the book but I have my doubts.

If you've played sports haven't you at times suddenly found yourself playing much better than usual, feeling exhilerated by it and your confidence soaring. I was crappy at sports so it was very noticeable when I played well. I even felt better coordinated when it happened. I saw it happen to teammates as well - getting in the zone, as we called it.

Players talk about similar subjective feelings during games - the basket looked like it was six feet wide; I felt like I couldn't miss, etc. We talk about players having the 'hot hand' and yell at the TV when his teammates don't get him the ball - or if he is an opponent we curse his defender for not playing him tight enough. At those moments we certainly believe in 'hot streaks'.

Outside of sports we report the same kinds of things - we can be 'on a roll' in almost anything.

It would be interesting to know if our subjective sense of feeling 'hot' had any relation to what happens or is it always after the unusual success is history, we project our feelings back to the moment. We are perfectly capable of rewriting our internal 'history' to accord with what we desire it to be or think it should be, so it could be totally illusory.

On the other hand I would expect people to have times when they do things better and times when they do them worse, with no particular awareness of it as its happening. That alone should cause a certain amount of streakiness in our performance. Shooting a basketball ought to be about as good of a arena as anything else for seeing if there any degree of streakiness above what is due to randomness. I'd love to have the data available to analyze... hmmmm, I could generate random data and also generate random date with streakiness injected into it to various degrees and see how much data it takes to detect the streakiness I know is there...

 

[Griffin]

I don't think anyone should necessarily accept any of the conclusions presented in this book, or any other publication for that matter, without looking at the data, but at the same time, to reject those conclusions without looking at the data is just as misguided, imo. Has anyone seen any statistical data in support of momentum, for example? Yet many accept these and other assumptions because they've been accepted as facts by the majority of people despite any objective evidence. Whether or not the conclusions drawn here are correct is definitely debatable, but so are the underlying assumptions themselves.

I'm a little surprised that the home-court advantage due to referee bias has gotten no arguments or discussion here. That to me was the most interesting point made in the whole book. But then I suppose we all kind of suspected it to be true all along.

 

[elindholm]

I'm a little surprised that the home-court advantage due to referee bias has gotten no arguments or discussion here. That to me was the most interesting point made in the whole book. But then I suppose we all kind of suspected it to be true all along.

Definitely. It's the same reason that teams with "superstars" have an edge in the playoffs, where such reputations are established through a self-reinforcing cycle.

 

[BC867]

Definitely. It's the same reason that teams with "superstars" have an edge in the playoffs, where such reputations are established through a self-reinforcing cycle.

You're right. It is frustrating for many teams, but it is good marketing for the league.

The Rolling Stones wouldn't have achieved the same greatness without Mick Jagger being put in the limelight, for example.

Unfortunately for the Suns, the "1" doesn't carry as much weight as a "4" or "5" superstar (literally and figuratively) in the eyes of the Press and the refs.

Steve certainly doesn't hasn't gotten the respect a 2-time MVP deserves from the refs (or Robert Horry).

 

[Manu4five]

I have read some scientific articles about momentum and I believe the data, that it is a myth. It is also very easy to explain why some people believe in momentum since it fits perfectly with other data within psychology. Basically our brains are trained to see patterns and give meaning to events, even if they are in fact completely random, that is, we use simple heuristics. Also if you invent a concept like "hot streak" people will tend to notice cases which proves the concept and ignore the other cases.

 

[Errntknght]

okay, I bought 'Scorecasting' & read 'hot hand'.

They cited the work of Huizinga and Wile, using seasons 2002-08. On average players (or at least perimeter players - they weren't clear) shot 3.5% worse after a made shot than after a missed shot; they also shoot 16% sooner after a make than a miss (this is for jump shots only.)

I'm inclined to think there are players that are streak shooters but who in their right mind would think it would show up in a league wide average, which is what they tested. Don't get me wrong those are amazing stats, but they say nothing at all about the presence or absence of streak shooters - except that there are not a large number of them.

What those stats tell me is that a large percentage of jump shooters believe that they have hot streaks and definitely act on that belief with the net result that they take worse shots following a make. We've all seen many examples of guys doing just that but I know I'd have never guessed it was that pervasive and had such a strong effect. It wouldn't certainly behoove a coach to make sure his players knew about this result and institute steps to discourage players from acting like they had a 'hot hand'. Presumably a coach could get his hands on this data and he could have someone track each of his players to discover the ones that have the worst histories.

Its very difficult to show there are no streak shooters - and equally hard to show there are some - for the simple reason that just by chance there is going to be considerable variation among the players for whatever measure of 'steakiness' you choose. With eight years of data you might get a strong indication one way or the other with a fairly complex analysis (I don't mean a complex way of measuring streakiness - their measure seems fine to me, though I'd test it against at least one other measure that is often used for this sort of analysis.)

 

[elindholm]

There are other corrupting influences on the data too. For example, the opposing team is probably more likely to increase its defensive attention on a particular player after a made shot. If the offense insists on going back to the same well, the defense will be better prepared. We used to see this phenomenon constantly down the stretch of close games in the D'Antoni era, of course.

 

[Errntknght]

Good point, Eric. I neglected to mention that factor, but they did discuss it in the book.

 

[chickenhead]

Interesting stuff, and it's nice to see these discussions happening more in basketball. I like seeing the data that prompts us to question out assumptions. For example, the idea of clutch shooting. IMO a clutch shooter is often either a great shooter who would make the same shot undefended 9 out of 10 times (Kobe Bryant), or a good shooter who is left inadequately defended because the great shooter is also on the floor (Robert Horry).