Wednesday, April 05, 2006
State of the stats
When it comes to the savvy use of stats in college hoops, this year we "reached a tipping point," "turned a corner," "made a breakthrough," and "used a lot of clichés." To recap....
Basketball has made do with per-game stats for a few decades now even though, due to the varying tempos of games, those stats lead to inherently flawed comparisons. Consider: at halftime of the national championship game when Dick Enberg gave his "essay" (was it just me or did Enberg not particularly enjoy Greg Gumbel's comment about his tie?), I was reminded how relatively new the shot clock really is in the college game. Tempos must have varied even more in the old days, no? Pity the player whose team went into the four corners midway through the second half in any game where they had a lead. ("What's the matter with you, Mikan? You're only getting eight points and four boards a game!") Still, while the shot clock may have brought the slowest teams closer to the mean, we continue in 2006 to see a big difference between a Maryland (75 possessions a game) and a Princeton (55).
But then along came this nifty stuff which I like to call tempo-free stats. Said nifty stats sprang, we think, from the fertile hoops minds of Frank McGuire and Dean Smith in the 1950s, were independently codified and committed to box score-friendly equations by Dean Oliver starting in the 1980s, and have now been made to truly sing in a college hoops setting by Ken Pomeroy.
So we now have the ability to compare Maryland to Princeton directly. The ability, yes. But how much longer until we can rest confident in the knowledge that the Biggest National Names on the air or on the web are actually basing their comments on this kind of ability? Good question!
In many ways, the outlook is dang sweet. First, tempo-free stats now bloom from a thousand, mostly bloggish, locales. (Want a teachably extreme example? Note the college hoops-inspired--and, as far as I know, unprecedented--adaptation of tempo-free stats to another possession-based sport, lacrosse.)
And yet these stats aren't just for blog geeks (redundant) anymore! Sports Illustrated led off its NCAA Tournament preview last month with a lengthy feature by Grant Wahl on the possession-based Tao: its roots and current applications. One Division II conference has started posting tempo-free stats on its site. And maybe the most encouraging sign in this saga has been the occasional yawn from a big-time coach--a yawn that says: "Why the fuss? I've been using this stuff forever."
Speaking of "why the fuss," Jay Bilas had this ($) to say in December at ESPN.com:
The old Dean Smith concepts of points per possession (and offensive and defensive efficiency) are starting to take on cult status, but we need to be careful not to overdo it. Several of us have been using that formula for a long time and it can be revealing; however, it is not the be all and end all of basketball. It is one tool in evaluating teams and players that can be useful. One of Dean Smith's books, "Basketball Multiple Offense and Defense," which I read in the 1980s, sets it out pretty well and is great reading for any serious basketball fan. "Moneyball" was all the rage in baseball a couple of years ago, and the concept is a useful tool as well, but there is no one statistical measure for basketball, the ultimate team game. There are too many variables, moving parts and dependencies in the game to look to any statistic to tell you the whole story.
I've always liked Bilas (yes, I'm flip-flopping already from yesterday's flip-flop--feel free to email the Big Ten Wonk "reader's representative") but this paragraph struck me as unintentionally revealing. If I may paraphrase Bilas: "Pay no attention to these faddish yet insufficient stats which, by the way, insiders like Dean Smith and I have been using way longer than you have."
Let's term Bilas's position the "they're just a fad and I used them first" paradox. My response to this paradox is pretty much the proverbial everlasting yea: Tempo-free stats are a fad, if by "fad" you mean something that's being perceived as "new" (though, in this case, the perception is incorrect) and is being adopted at an ever-accelerating rate. I would hazard a guess that batting average started as a "fad."
But I think the most common misconception related to this particular fad is a belief that these stats must surely represent something irreducibly complex and esoteric. I think the truth is pretty much the exact opposite. The value of tempo-free stats is precisely that, like a batting average, they enable us to perform the most mundane and least esoteric of descriptive housekeeping, to wit: "That team has a good defense." "This player is a better rebounder than that player." "That team turns the ball over a lot."
This type of mundane statement comes in really handy. It'd be nice if the announcers on the game we're watching on a given night could be trusted to make this type of statement dependably. And I'm not asking for the fully loaded tempo-free utopia all at once. But someday soon, games on ESPN could take a baby step or two:
--Replace FG percentage with effective FG pct.
--Never again confuse mere slowness for "defense," "allowing just 55 points a game," etc.
--Maybe even an occasional PPWS, especially in discussing tomorrow's Redicks and Morrisons.
Until that day, blogs will continue to enjoy, however improbably, a near monopoly on the information you need to talk hoops with the same level of minimal lucidity that's been taken for granted for decades in every other major sport. It's a monopoly I, for one, would love to see end. Soon.
In today's less Wonk-ish venues....
Hard to believe we're here already but I'm about to put the blog on its annual hiatus....
The farewell week continues!
Barring unforeseen developments, Friday's post should be the last one. I will then dutifully shut the old girl down for the off-season like a Bar Harbor lobster pound. So here's the season-ending clip show! A look back at the year's most notable one-game performances by or against a Big Ten team....
Fastest game: 84 possessions, Florida A&M vs. Indiana in Bloomington, November 21. (So the fact that the Hoosiers scored 100 points in that game indicates great but not ridiculous offensive efficiency--1.20 points per possession.)
Slowest game: 52 possessions, Minnesota vs. Northwestern in Evanston, March 4. Second-slowest? Iowa vs. Northwestern (55 possessions), in Evanston, February 8. Tied for second-slowest? Northwestern vs. Illinois (55 possessions), in Champaign, February 15. Wait, there's more: Purdue vs. Northwestern (55 possessions), in Evanston, January 4. You get the idea: the Wildcats were content to come in under 60 possessions in 10 of their 16 conference games.
Best offense: 1.55 points per possession, Ohio State against Penn State in Columbus, January 5. This is what dominance looks like: the Buckeyes scored 104 points in only 67 possessions. Thad Matta's team shot lights-out (43-of-68, 13-of-26 on threes, an effective FG percentage of 72.8) turned the ball over only nine times, and rebounded more than half (14) of their 26 misses.
Worst offense: 0.52 points per possession, Maryland-Eastern Shore against Iowa in Iowa City, November 14. This was the season-opener for the Hawkeyes and they held their overmatched opponent to just 41 points in what was actually a very fast (79-possession) game.
Best shooting: 78.6 effective FG percentage, Iowa against Michigan in Iowa City, February 4. Jeff Horner went 5-of-10 from the field and Justin Wieck was 0-for-1; every other Hawkeye (and Steve Alford played 12 that day) was better than 50 percent from the field. Iowa made 13 of 19 threes and 19 of 30 twos in a 94-66 laugher.
Worst shooting: (You guessed it!) 25.0 effective FG percentage, Wisconsin vs. North Dakota State in Madison, January 21. The Badgers made just 16 of 72 attempts and sank only four of 27 threes.
Wonk back!
Don't just mutter ineffectually; email me!
The (time-release) "poison pill" brings good results?
Yesterday I noted that Iowa was the "poison pill" in this year's tournament: their first-round loss to Northwestern St. set off a chain reaction of next-round losses for each successive victor (Northwestern St. lost its next game to West Virginia, West Virginia lost its next game to Texas, etc.) all the way up through the national championship game.
Hawkeye fans respond!
Iowa's and Alford's dubious showing as the Least Defensible Performers in the NCAA tournament actually portends well for the young men in black and gold.
Why, it was but two short years ago that a 5th-seeded Florida team lost to Manhattan (by 15!) in the first round. Manhattan then lost to Wake Forest, who lost to St. Joe's, who lost to Oklahoma State, who lost to Georgia Tech, who lost to UConn in the championship game.
By the way, if this e-mail at all increases the likelihood of Alford receiving a contract extension, please disregard.
Adam J.
Congratulations, in advance, to the Iowa Hawkeyes, National Champions of 2008! Coach Quin Snyder will be thrilled! ¶ 5:53 AM
When it comes to the savvy use of stats in college hoops, this year we "reached a tipping point," "turned a corner," "made a breakthrough," and "used a lot of clichés." To recap....
Basketball has made do with per-game stats for a few decades now even though, due to the varying tempos of games, those stats lead to inherently flawed comparisons. Consider: at halftime of the national championship game when Dick Enberg gave his "essay" (was it just me or did Enberg not particularly enjoy Greg Gumbel's comment about his tie?), I was reminded how relatively new the shot clock really is in the college game. Tempos must have varied even more in the old days, no? Pity the player whose team went into the four corners midway through the second half in any game where they had a lead. ("What's the matter with you, Mikan? You're only getting eight points and four boards a game!") Still, while the shot clock may have brought the slowest teams closer to the mean, we continue in 2006 to see a big difference between a Maryland (75 possessions a game) and a Princeton (55).
But then along came this nifty stuff which I like to call tempo-free stats. Said nifty stats sprang, we think, from the fertile hoops minds of Frank McGuire and Dean Smith in the 1950s, were independently codified and committed to box score-friendly equations by Dean Oliver starting in the 1980s, and have now been made to truly sing in a college hoops setting by Ken Pomeroy.
So we now have the ability to compare Maryland to Princeton directly. The ability, yes. But how much longer until we can rest confident in the knowledge that the Biggest National Names on the air or on the web are actually basing their comments on this kind of ability? Good question!
In many ways, the outlook is dang sweet. First, tempo-free stats now bloom from a thousand, mostly bloggish, locales. (Want a teachably extreme example? Note the college hoops-inspired--and, as far as I know, unprecedented--adaptation of tempo-free stats to another possession-based sport, lacrosse.)
And yet these stats aren't just for blog geeks (redundant) anymore! Sports Illustrated led off its NCAA Tournament preview last month with a lengthy feature by Grant Wahl on the possession-based Tao: its roots and current applications. One Division II conference has started posting tempo-free stats on its site. And maybe the most encouraging sign in this saga has been the occasional yawn from a big-time coach--a yawn that says: "Why the fuss? I've been using this stuff forever."
Speaking of "why the fuss," Jay Bilas had this ($) to say in December at ESPN.com:
The old Dean Smith concepts of points per possession (and offensive and defensive efficiency) are starting to take on cult status, but we need to be careful not to overdo it. Several of us have been using that formula for a long time and it can be revealing; however, it is not the be all and end all of basketball. It is one tool in evaluating teams and players that can be useful. One of Dean Smith's books, "Basketball Multiple Offense and Defense," which I read in the 1980s, sets it out pretty well and is great reading for any serious basketball fan. "Moneyball" was all the rage in baseball a couple of years ago, and the concept is a useful tool as well, but there is no one statistical measure for basketball, the ultimate team game. There are too many variables, moving parts and dependencies in the game to look to any statistic to tell you the whole story.
I've always liked Bilas (yes, I'm flip-flopping already from yesterday's flip-flop--feel free to email the Big Ten Wonk "reader's representative") but this paragraph struck me as unintentionally revealing. If I may paraphrase Bilas: "Pay no attention to these faddish yet insufficient stats which, by the way, insiders like Dean Smith and I have been using way longer than you have."
Let's term Bilas's position the "they're just a fad and I used them first" paradox. My response to this paradox is pretty much the proverbial everlasting yea: Tempo-free stats are a fad, if by "fad" you mean something that's being perceived as "new" (though, in this case, the perception is incorrect) and is being adopted at an ever-accelerating rate. I would hazard a guess that batting average started as a "fad."
But I think the most common misconception related to this particular fad is a belief that these stats must surely represent something irreducibly complex and esoteric. I think the truth is pretty much the exact opposite. The value of tempo-free stats is precisely that, like a batting average, they enable us to perform the most mundane and least esoteric of descriptive housekeeping, to wit: "That team has a good defense." "This player is a better rebounder than that player." "That team turns the ball over a lot."
This type of mundane statement comes in really handy. It'd be nice if the announcers on the game we're watching on a given night could be trusted to make this type of statement dependably. And I'm not asking for the fully loaded tempo-free utopia all at once. But someday soon, games on ESPN could take a baby step or two:
--Replace FG percentage with effective FG pct.
--Never again confuse mere slowness for "defense," "allowing just 55 points a game," etc.
--Maybe even an occasional PPWS, especially in discussing tomorrow's Redicks and Morrisons.
Until that day, blogs will continue to enjoy, however improbably, a near monopoly on the information you need to talk hoops with the same level of minimal lucidity that's been taken for granted for decades in every other major sport. It's a monopoly I, for one, would love to see end. Soon.
In today's less Wonk-ish venues....
Hard to believe we're here already but I'm about to put the blog on its annual hiatus....
The farewell week continues!
Barring unforeseen developments, Friday's post should be the last one. I will then dutifully shut the old girl down for the off-season like a Bar Harbor lobster pound. So here's the season-ending clip show! A look back at the year's most notable one-game performances by or against a Big Ten team....
Fastest game: 84 possessions, Florida A&M vs. Indiana in Bloomington, November 21. (So the fact that the Hoosiers scored 100 points in that game indicates great but not ridiculous offensive efficiency--1.20 points per possession.)
Slowest game: 52 possessions, Minnesota vs. Northwestern in Evanston, March 4. Second-slowest? Iowa vs. Northwestern (55 possessions), in Evanston, February 8. Tied for second-slowest? Northwestern vs. Illinois (55 possessions), in Champaign, February 15. Wait, there's more: Purdue vs. Northwestern (55 possessions), in Evanston, January 4. You get the idea: the Wildcats were content to come in under 60 possessions in 10 of their 16 conference games.
Best offense: 1.55 points per possession, Ohio State against Penn State in Columbus, January 5. This is what dominance looks like: the Buckeyes scored 104 points in only 67 possessions. Thad Matta's team shot lights-out (43-of-68, 13-of-26 on threes, an effective FG percentage of 72.8) turned the ball over only nine times, and rebounded more than half (14) of their 26 misses.
Worst offense: 0.52 points per possession, Maryland-Eastern Shore against Iowa in Iowa City, November 14. This was the season-opener for the Hawkeyes and they held their overmatched opponent to just 41 points in what was actually a very fast (79-possession) game.
Best shooting: 78.6 effective FG percentage, Iowa against Michigan in Iowa City, February 4. Jeff Horner went 5-of-10 from the field and Justin Wieck was 0-for-1; every other Hawkeye (and Steve Alford played 12 that day) was better than 50 percent from the field. Iowa made 13 of 19 threes and 19 of 30 twos in a 94-66 laugher.
Worst shooting: (You guessed it!) 25.0 effective FG percentage, Wisconsin vs. North Dakota State in Madison, January 21. The Badgers made just 16 of 72 attempts and sank only four of 27 threes.
Wonk back!
Don't just mutter ineffectually; email me!
The (time-release) "poison pill" brings good results?
Yesterday I noted that Iowa was the "poison pill" in this year's tournament: their first-round loss to Northwestern St. set off a chain reaction of next-round losses for each successive victor (Northwestern St. lost its next game to West Virginia, West Virginia lost its next game to Texas, etc.) all the way up through the national championship game.
Hawkeye fans respond!
Iowa's and Alford's dubious showing as the Least Defensible Performers in the NCAA tournament actually portends well for the young men in black and gold.
Why, it was but two short years ago that a 5th-seeded Florida team lost to Manhattan (by 15!) in the first round. Manhattan then lost to Wake Forest, who lost to St. Joe's, who lost to Oklahoma State, who lost to Georgia Tech, who lost to UConn in the championship game.
By the way, if this e-mail at all increases the likelihood of Alford receiving a contract extension, please disregard.
Adam J.
Congratulations, in advance, to the Iowa Hawkeyes, National Champions of 2008! Coach Quin Snyder will be thrilled! ¶ 5:53 AM
