Basketball Statistics

[whereinthehellami]

I'm reading Moneyball by Michael Lewis right now and it is interesting how missleading some statistics are in baseball (fielding precentage, RBIs). The work that Bill James did to revolutionize statistics for baseball is amazing. Has anyone done this for basketball? It seems that basketball could use some new stats and a field simliar to sabermetrics. where is basketball's VORP, OBP?

Which basketball statistic is overrated? I've always thought that blocks are somewhat misleading when you consider that many shot blockers block the ball out of bounds. maybe a stat that tracked down the blocks that result in a change of possesion would be more informative.

[uh_no]

Which basketball statistic is overrated? I've always thought that blocks are somewhat misleading when you consider that many shot blockers block the ball out of bounds. maybe a stat that tracked down the blocks that result in a change of possesion would be more informative.

Tell that to 2 time defensive player of the year hasheem Thabeet

a block not only is a missed shot, but the shot clock doesn't reset, so its like the shot never happened....its also a huge mental thing....if your team gets 10 blocks a game, you think twice about driving the lane and putting up some weaksauce layup

[airowe]

Tell that to 2 time defensive player of the year hasheem Thabeet

a block not only is a missed shot, but the shot clock doesn't reset, so its like the shot never happened....its also a huge mental thing....if your team gets 10 blocks a game, you think twice about driving the lane and putting up some weaksauce layup

Or Shelden Williams. A blocked shot that stays in bounds and becomes a turnover is obviously much more effective than a shot that goes out of bounds.

[Tim1515]

I'm reading Moneyball by Michael Lewis right now and it is interesting how missleading some statistics are in baseball (fielding precentage, RBIs). The work that Bill James did to revolutionize statistics for baseball is amazing. Has anyone done this for basketball? It seems that basketball could use some new stats and a field simliar to sabermetrics. where is basketball's VORP, OBP?

Which basketball statistic is overrated? I've always thought that blocks are somewhat misleading when you consider that many shot blockers block the ball out of bounds. maybe a stat that tracked down the blocks that result in a change of possesion would be more informative.

I think it's much harder to do for basketball...there are two HUGE differences (great book btw)

1) length of schedule. The college basketball season is long...NBA longer but 162 games is hard to compete with. Over that length of time there is less unknown so to speak...things even out more.

2) Baseball, to a large degree, is a 1v1 sport. You never have double teams or zones which i'm sure makes it significantly harder to runs stats for.

[kmspeaks]

I'm reading Moneyball by Michael Lewis right now and it is interesting how missleading some statistics are in baseball (fielding precentage, RBIs). The work that Bill James did to revolutionize statistics for baseball is amazing. Has anyone done this for basketball? It seems that basketball could use some new stats and a field simliar to sabermetrics. where is basketball's VORP, OBP?

This link was posted back in February.

http://www.nytimes.com/2009/02/15/magazine/15Battier-t.html?_r=1

The Rockets don't want to give away their secret just yet but it seems they have come up with some sort of alternative statistics. Not that this comes as a suprise to us Duke fans but these new stats show that Shane Battier is pretty much in a league by himself.

[uh_no]

Or Shelden Williams. A blocked shot that stays in bounds and becomes a turnover is obviously much more effective than a shot that goes out of bounds.

Obviously. But this does not mean that blocked shots is not a more telling stat than blocked shot which stays in bounds.

[theAlaskanBear]

I'm a big baseball fan, and while I do appreciate the innovations by Bill James, if you look at the Billy Beane "Money Ball" approach, in baseball terms it has been an abject failure.

[DevilHorns]

I'm a big baseball fan, and while I do appreciate the innovations by Bill James, if you look at the Billy Beane "Money Ball" approach, in baseball terms it has been an abject failure.

as a huge baseball fan, you have to agree that somehow teams that can "buy" wins arent necessarily the ones who win it all every year. Look at the Fla Marlins who won the world series with no payroll (specifcially their second championship). i think this speaks to the inefficiences of how baseball is run since the best teams arent necessarily the ones who have the highest payroll.

[uh_no]

I'm a big baseball fan, and while I do appreciate the innovations by Bill James, if you look at the Billy Beane "Money Ball" approach, in baseball terms it has been an abject failure.

Tell that to the boston red sox two world series this decade......theo epstein is a moneyball disciple

[pfrduke]

I'm reading Moneyball by Michael Lewis right now and it is interesting how missleading some statistics are in baseball (fielding precentage, RBIs). The work that Bill James did to revolutionize statistics for baseball is amazing. Has anyone done this for basketball? It seems that basketball could use some new stats and a field simliar to sabermetrics. where is basketball's VORP, OBP?

Which basketball statistic is overrated? I've always thought that blocks are somewhat misleading when you consider that many shot blockers block the ball out of bounds. maybe a stat that tracked down the blocks that result in a change of possesion would be more informative.

One really simple one is raw fg%, because it weights 2 point field goals and 3 point field goals the same, even though one is, quite obviously, worth 50% more. For example, during their senior seasons, Shelden shot 57.8%, while Redick shot 47%. But they both scored 1.16 points per shot (Shelden was a touch better, but only slightly).

That's why efg% is better than raw fg% - it equalizes all shots from the field. The efg%'s for JJ and Shelden were 57.7% and 58.0%, respectively, much closer than the raw fg% would lead you to believe.

Not that this is revolutionary or anything, but it would at least allow guards to compete for fg% titles.

[theAlaskanBear]

Tell that to the boston red sox two world series this decade......theo epstein is a moneyball disciple

Theo Epstein also has the second highest payroll in baseball after the Yankees, and his two sluggers were caught using steroids (Manny and Ortiz).

Bottom line is, the Red Sox and Yankees can buy whatever player they want. It doesnt really matter if Epstein is a disciple of anything, as long as he has half a brain he can field winning teams.

[uh_no]

I'm not sure what the sluggers using steroids has to do with anything.....everyone was on steroids

also take a look at the mets who can spend a lot of money and not win.....there is something to be said for sabermetrics

[Duke09]

I'm a big baseball fan, and while I do appreciate the innovations by Bill James, if you look at the Billy Beane "Money Ball" approach, in baseball terms it has been an abject failure.

Sorry to take this off topic, but ABear you are way off.

What Beane in th early 2000s did was find an undervalued asset in the market place, namely hitters who could get on base, take walks, and drive up hit counts, measured by stat on base percentage (as you know, the percentage of time a hitter gets on base, no counting errors etc.) This was in contrast to some teams who would have hitters with decent averages (.300-280) but relatively low on base precentages (.320... .400 is considered very good). So a .270 hitter can be better than a .300 hitter if he gets on base much more often. Thus by only spending 40mil a year, the oakland As won several divisions and, if I'm not mistake, a best record in baseball with 103 wins in 2002. As you also know, they had mvps like miguel cabrera and jason giambi.

But they never won a championship you say? Yes, because unlike 162 game regular season where you get enough chances to find a true average (some statistician is going to kill me for that description), in a 5 game series there is a lot of luck involved. Its the difference between flipping a coin 100 times, versus 5. The closest to 50% heads you can have with 5 flips is 3-2, or 60 percent heads (or conversely 40% heads). The average of 100 flips will likely be much close to 50% heads. The A's had very little luck, running into the upstart Twins and who can forget the Jeter tag out of giambi at the plate that is one of the plays of the decade. 5 games does not a season make.

So why aren't the As any good now? The league adjusted. Now everyone values on base percentage and its cousin slugging percentage (total bases per plate appearance). A team with a small market cannot compete for these guys who 10 years ago were undervalued. Lately what has been undervalued are young players who are cheap. So the Marlins sign Hanley Ramirez long term at under market rate (his salary is 11.6 mill over 6 years. If he were a free agent he'd easily get 20 mil a year. Rockies did the same with troy tulowitzki. In return for reduced salary they aren't stuck on a rookie contract that pays 1 million at most, likely under 500k.

So, in conclusion, Moneyball wasn't a failure, just the market corrected and Beane hasn't found a new way to strike gold. But the league worships at the alter of on base percentage and slugging.

[UrinalCake]

...unlike 162 game regular season where you get enough chances to find a true average (some statistician is going to kill me for that description), in a 5 game series there is a lot of luck involved.

Some of that is luck, but some of it is an intangible quality that cannot be measured by statistics. Certain players are going to excel in pressure situations, while others falter. That's one of the big criticisms of moneyball - you simply cannot reduce human beings to numbers and expect it to be accurate.

As far as useless statistics, so many people look at raw scoring average that it always frustrates me. A guy will shoot 10-31 and get top billing in the SportsCenter highlights because he scored 25 points.

[theAlaskanBear]

Sorry to take this off topic, but ABear you are way off.

What Beane in th early 2000s did was find an undervalued asset in the market place, namely hitters who could get on base, take walks, and drive up hit counts, measured by stat on base percentage (as you know, the percentage of time a hitter gets on base, no counting errors etc.) This was in contrast to some teams who would have hitters with decent averages (.300-280) but relatively low on base precentages (.320... .400 is considered very good). So a .270 hitter can be better than a .300 hitter if he gets on base much more often. Thus by only spending 40mil a year, the oakland As won several divisions and, if I'm not mistake, a best record in baseball with 103 wins in 2002. As you also know, they had mvps like miguel cabrera and jason giambi.

But they never won a championship you say? Yes, because unlike 162 game regular season where you get enough chances to find a true average (some statistician is going to kill me for that description), in a 5 game series there is a lot of luck involved. Its the difference between flipping a coin 100 times, versus 5. The closest to 50% heads you can have with 5 flips is 3-2, or 60 percent heads (or conversely 40% heads). The average of 100 flips will likely be much close to 50% heads. The A's had very little luck, running into the upstart Twins and who can forget the Jeter tag out of giambi at the plate that is one of the plays of the decade. 5 games does not a season make.

So why aren't the As any good now? The league adjusted. Now everyone values on base percentage and its cousin slugging percentage (total bases per plate appearance). A team with a small market cannot compete for these guys who 10 years ago were undervalued. Lately what has been undervalued are young players who are cheap. So the Marlins sign Hanley Ramirez long term at under market rate (his salary is 11.6 mill over 6 years. If he were a free agent he'd easily get 20 mil a year. Rockies did the same with troy tulowitzki. In return for reduced salary they aren't stuck on a rookie contract that pays 1 million at most, likely under 500k.

So, in conclusion, Moneyball wasn't a failure, just the market corrected and Beane hasn't found a new way to strike gold. But the league worships at the alter of on base percentage and slugging.

Listen, you wont find me arguing against OPS and sabermetric stats. They are useful evaluation tools. But since Moneyball a stats cult has built up where some people value statistical analysis over all else. I call it Fantasy sports syndrome. There are a lot of different dimensions that go into making good ball players, and good baseball teams.

Beane had a good five year run, but they never won anything. I agree that there is an element of luck in the playoffs, but you cant just say Beane was unlucky, you have to give him credit for his failures as well as his success. Thats my argument. There are many different ways to build successful playoff teams, and the Moneyball approach hasnt proven itself better than any else.

And to the person above who said "so what, everyone used steroids" thats just not true.

[brevity]

Beane had a good five year run, but they never won anything. I agree that there is an element of luck in the playoffs, but you cant just say Beane was unlucky, you have to give him credit for his failures as well as his success. Thats my argument. There are many different ways to build successful playoff teams, and the Moneyball approach hasnt proven itself better than any else.

You're absolutely right. Moneyball fans should acknowledge that Billy Beane steered Oakland teams with no business being in the playoffs into Divisional Series defeats. Because if you can't win it all, why bother, right?

Theo Epstein's variation of Moneyball -- sabermetrics worship plus deep pockets -- was a more successful model come October. It isn't all about money; the mid-aughts Red Sox (and the Yankees recently) have been credited with spending heavily but smartly. Most teams would probably love to mimic Epstein's approach, but can only afford to adopt Beane' approach.

I do agree with your concept of Fantasy Sports Syndrome. It's a virus. Like compulsive listmaking, it's an easy but deeply flawed way for a people to apply some sort of order to the chaos of sports, even though success can't be built on paper. I apply it in another thread to those DBR posters who obsess over Duke's basketball roster in future years, as if they have a magical hat that says "Coach" or "Sports Director" or something.

[theAlaskanBear]

You're absolutely right. Moneyball fans should acknowledge that Billy Beane steered Oakland teams with no business being in the playoffs into Divisional Series defeats. Because if you can't win it all, why bother, right?

Theo Epstein's variation of Moneyball -- sabermetrics worship plus deep pockets -- was a more successful model come October. It isn't all about money; the mid-aughts Red Sox (and the Yankees recently) have been credited with spending heavily but smartly. Most teams would probably love to mimic Epstein's approach, but can only afford to adopt Beane' approach.

I do agree with your concept of Fantasy Sports Syndrome. It's a virus. Like compulsive listmaking, it's an easy but deeply flawed way for a people to apply some sort of order to the chaos of sports, even though success can't be built on paper. I apply it in another thread to those DBR posters who obsess over Duke's basketball roster in future years, as if they have a magical hat that says "Coach" or "Sports Director" or something.

I sense your sarcastic tone in the first paragraph, and let me tell you that it is not what I am implying, or sports (especially baseball) would be a bleak world indeed. I am a die-hard StL Cardinals fan, and we have good years and we have bad years. A World Series isn't the only determinant of success.

Also, the Cardinal Manager and pitching coach, Tony LaRussa and Dave Duncan (both former A's coaches in the late 80s early 90s) pioneered a statistical approach to the game, especially in regards to situational relief pitching and player platoons, so I appreciate the statistical nature of the game.

Payroll is and will always the biggest factor to success in the MLB until there is a salary cap. Sure, there are teams like the Cubs and Mets who spend a lot and dont get results, but if you look back to the big spenders the last several years, you will see a direct correlation with payroll and sustained success.

[CDu]

Payroll is and will always the biggest factor to success in the MLB until there is a salary cap. Sure, there are teams like the Cubs and Mets who spend a lot and dont get results, but if you look back to the big spenders the last several years, you will see a direct correlation with payroll and sustained success.

I completely agree. Which is what makes the Moneyball approach taken by Beane and colleagues that much more impressive. He was able to use the Moneyball approach to consistently compete with the big spenders for years despite having a payroll of about one third that of the big boys.

[NSDukeFan]

I completely agree. Which is what makes the Moneyball approach taken by Beane and colleagues that much more impressive. He was able to use the Moneyball approach to consistently compete with the big spenders for years despite having a payroll of about one third that of the big boys.

And as Duke09 mentioned (not in these words) "imitation is the sincerest form of flattery" as every team now uses some form of updated statistical analysis to determine value vs. using wins, hrs, rbis, era, sbs, batting average, etc. as the sole judges of successful players. It has taken the advantage away from Billy Beane after his great success with low-market teams.
Unfortunately, this has also hurt my favorite player of all time's (Andre Dawson) chance at the hall when you look at the numbers a bit in retrospect. I am still hoping to see him get in, though.

[tbyers11]

Payroll is and will always the biggest factor to success in the MLB until there is a salary cap. Sure, there are teams like the Cubs and Mets who spend a lot and dont get results, but if you look back to the big spenders the last several years, you will see a direct correlation with payroll and sustained success.

I agree with this as well. The distinction that I see is that while having a high payroll doesn't guarantee you success it increases your chances greatly while having a low payroll makes your chances of success very, very small.

Here are team salaries (http://www.cbssports.com/mlb/salaries) for the 2009 season. Some of the big spenders were failures (Cubs, Mets, Astros) but 5 of the 8 playoff teams were in the top 9 salaries (above $100 million) with the Cardinals (13th), Rockies (18th) and Twins (24th) falling below $100 million.

Basically, if you don't have a high payroll you pretty much have a once-a-decade or so chance for success (2003 Marlins, 2008 Rays) and then you go back to mediocrity. This makes Oakland's run from 2000-2003 and the Twins five playoff berths since 2002 impressive.

[CDu]

And as Duke09 mentioned (not in these words) "imitation is the sincerest form of flattery" as every team now uses some form of updated statistical analysis to determine value vs. using wins, hrs, rbis, era, sbs, batting average, etc. as the sole judges of successful players. It has taken the advantage away from Billy Beane after his great success with low-market teams.
Unfortunately, this has also hurt my favorite player of all time's (Andre Dawson) chance at the hall when you look at the numbers a bit in retrospect. I am still hoping to see him get in, though.

Agreed on the first part, but not sure I agree about Dawson. I'm not sure that the sabrmetricians have really reached the Hall voting yet. I think they still vote on old school numbers, as evidenced by the fact that Bert Blyleven still isn't in, and Tim Raines actually got fewer votes this year than last year.

I think Dawson's candidacy has been hurt mostly by the fact that (like Dale Murphy) he is a fringe Hall of Famer who lingered too close into the steroid era. Had his numbers been amassed a half-decade sooner, he'd be a shoe-in. But now his 437 HR don't look so glowing, and we're far removed from his glory days. He had some really great years in anonymity in the early 80s (actually better than his MVP season in 1987), but injuries and hanging around are what has hurt him I think. Which is sad, because he was one of my favorites too.

[Duke09]

After hijacking this thread, I'm going to try and bring it back a little.

The reason statistics are so useful in baseball it that it is very much a 1 on one sports. The pitcher has to throw the ball in the vicinity of the strike zone and the batter has a chance to hit it. So you look at things a players directly controls (hits, HRs, walks, pitches per at bat, HRs given up, walks given hit, batting avg against, etc.) it correlates well to player and team performance.

basketball is much more a team game. Where other guys are on the court dictates what you do. Did someone cut to get you open? Did you get the ball in a good position to score? Did you seal off a defender? Did you move a defender 6 inches away from where they want to be? Did the double team come quickly and did everyone rotate? There are a thousand different questions and permutations of a play.

Because everyone is interdependent, it is hard to for statistics to correlate to a players abilities. Steals is a commonly cited defensive statistic. Mario Chalmers gets his fair share of steals, but can't guard anyone. Every point guard who plays Miami has a huge night. Sometimes a guy hits a tough shot, but a better and higher percentage shot was a available and would be a better play. Over thousands of plays these things add up. If you haven't read the NYT Magazine linked earlier in this thread. It is an amazing read.

[ArtVandelay]

To steer this discussion back to basketball (although I love me a good baseball sabermetrics discussion any day of the week):

If you're looking for basketball statistical innovations, the first step would be to go to KenPom's site. The tempo-free analysis of basketball is a major step forward that most mainstream commentators almost never talk about. You could also point to things like +/-, which Jumbo posts after every game. Another good example of a slightly different lens on things.

As others have pointed out, the big problem with basketball is that everything in basketball is interactive, not one on one. While you can easily look at "counting" statistics (e.g. points, rebounds, etc.), it's much harder to measure things a player DIDN'T do, or things that "don't show up in the boxscore," as they say, and the impact those things have on the game. It makes it very difficult to isolate any one player's contribution. That said, things like +/-, efg, etc. do help us evaluate a player's true value to the team more than things like scoring average.

[MChambers]

there is this book, which is okay, but as others have said, baseball is better suited to statistical analysis than any other sport.

http://www.amazon.com/Basketball-Paper-Rules-Performance-Analysis/dp/1574886878/ref=sr_1_1?ie=UTF8&s=books&qid=1261081634&sr=8-1

[mus074]

I am personally a big fan of Ken Pomeroy's work (http://kenpom.com/rate.php) in breaking down the games in poss-by-poss tempo-free stats and then compiling them in a more meaningful way. For example, the A/TO ratio is cute and fun to quote, but it is not necessarily the best indicator of ball control. It requires both stats to be good to show a meaningful statistic.

I really like the TO %age rating which tells you the percentage of possession the player is on the floor and turns it over. Similarly, the Assist %age tells the number of assists as a ratio of the teammates baskets made while the player is on the floor.

To my mind, the best single stat improvement is the Reb %age. It tells you what percentage of available rebounds were gathered in by the defense, offense or individual player on each side of the floor. Comparing one team's raw offensive rebounds versus another's is really, really distorted by the number of shots taken and shooting percentage. Creating TOs by the other team will limit the number of shots taken and thus rebound opportunities.

Pomeroy has made some of his broader stroke team-by-team stats available for download in CSV format from 2004 season to present. I am working on a regression analysis project with them and am finding some really neat correlations. I will share one initial finding.

I start from the idea that the Pythag Rating, which is based on the very same Bill James baseball model, is an accurate rating of ability. The highest correlation of any tracked stat to the Pythag is offensive TO %age. More than offensive Eff FG %age, defensive Eff FG %age and much more so than defensive TO %age.

So in looking at teams right now and seeing who has the best makings of a successful season, the first and single best indicator is Offensive TO %age - holding on to the ball. That bodes poorly for UK (279th) and UNC (190th), for example, unless they can really get a handle on their ball control.

Currently, Duke is ranked 3rd in the nation in Off TO %age. That is a raw statistic and not adjusted for quality of opposition. Duke's offensive SOS (the quality of the defenses it has played) is ranked 69th. The two teams with better TO %ages have played the 256th and 273th toughest defenses.

One stat alone will not predict too very much at all. But being fantastic at the apparently most important stat can't hurt much either.

If anyone wants to work on this project with me, message me here and we can share emails and discuss my thoughts. I would love to hear other's ideas too. I am limited by the stats Pomeroy has made available for sharing, but the ones includes do make for some fun analysis. :)

[mus074]

After I posted, I read this article (http://basketballprospectus.com/article.php?articleid=829) by Gasaway. Take away the gratuitous frustration, and I agree 100% with his take on rebounding, as stated above.

[Olympic Fan]

Let me join the chorus of those who appreciate sabremetrics and the Bill James/Billy Beane school of "new" statisticians. I think the earlier posters nailed it -- Beane never won world championships at Oakland, but his teams won significantly more often (and with more consistency) than they should have with their payroll. When Theo Epstein applied the same principals to a team with deep pockets, his finally brought Boston a championship ... two championships, in fact. As a Yankee fan, I don't mind saying that he forced my team to adopt some of the same principles.

But none of this is new. I've read recent articles suggesting that John McGraw looked for many of the same things that Beane has emphasized -- especially OBP rather than BA. It could be just a coincidence, but as a player with the old Orioles, McGraw had one of the great career OBP numbers in baseball history.

As for basketball, I'm no Bill James, but I do want to make a few points:

(1) I agree that steals are perhaps the single most overrated stat in basketball. There are a lot of bad defenders who gamble on the steal and end up with impressive steal totals, but actually hurt their teams in the long run becaue when they don't get the steal, they leave their team in bad position.

(2) Blocked shots are different, largely because of the intimidation factor. The shots that are blocked are not the only ones that are altered by a shot-blocker. I remember an old NBA coach complaining that his team lost in the finals because none of his players could take a shot from anywhere on the court without worrying that Bill Russell was going to block it.

However, blocks are not always a positive. We saw that with Shelden that he would rotate on defense to block or alter a shot, but by doing so, he would leave his man in good position for the offensive rebound. I think this happened in the 2005 loss to UNC in Chapel Hill, when Sean May had 20 rebounds in a close win -- time after time, Williams would help out on Felton or McCants, force a miss ... only to have May complete the play with a rebound. It's amazing to look back at that 2005 season when Shelden was (rightly) the national player of the year and see how many opposing big men had big games against him for precisely that reason.

(3) Team offense and defensive stats usually measure tempo and not offensive or defensive efficiency. For years, Pete Carril's Princeton teams led the nation in scoring defense and were touted as great defensive teams. That was bunk -- they merely played at such a slow tempo that scoring was down.

I think that Ken Pomeroy has done a very good job of rationalizing stats. I don't agree with everything he's devised, but overall I'm a big fan.

(4) Historically, it's impossible to compare basketball stats across eras because the game has changed so much.

Just one example -- modern rebound leaders can't hold a candle to guts in the 1950s and 1960s. Why? Because the game was so different. Teams ran up and down the floor with what we would consider appalling shot disciple. There were a ton more missed shots and hence a ton more rebounds.

In addition, we've only been counting such stats as assists, turnovers, blocked shots and steals since the mid-1970s. So when you hear so-so set an ACC or NCAA record for blocks, for instance, keep in mind that we don't know how many shots Tommy Burleson blocked ... or Bill Russell.

Anyway, I welcome any effort to refine basketball assists. As I said, I like what Pomeroy is doing, but he's no Bill James.

PS Football stats could do with some refining too. I read a convincing article not too long ago arguing that the single most overlooked stat in football had to so with yards per pass attempt. Forgot all the ridiculous QB rating numbers, total yards, completion pct. and the like -- yards per pass attempt was the OBP of football passing stats (although the author did suggest that you also do quick check on interception rate to suppliment that).

[mus074]

Oly, I agree about Pomeroy. I admire and appreciate what he has done for revealing tempo-free stats, but I am not as overwhelmed by his application of that data. I think there is great room for improvement, a la Billy Beane's regression analysis. So I'm putting my energy where my mouth is and seeing what I can uncover.