The end of the (stats) world as we know it

[DobbsRover2]

While plowing through some data visualization blogs, I stumbled on an article about how the NBA is launching into a new mapping technologies and data analysis system that will soon revolutionize the world of basketball stats. No longer will we be arguing about who is the AAC's most improved player or best defensive player based on pretty limited value stats like points per game or total blocks.

With the help of overhead cameras that map all the moves the ball handler and other players make on the court, the key stat will be the EPV (Expected Possession Value), which will spin off likely many dozen subvalues to show how valuable a player is on the court. Instead of being bound to an old box score set of stats, we will now have far more nuanced stats that will show us just how effectively the future Stef Dolson set picks or future Moriah Jefferson gets a steal when cheating to the left. And when the stats are analyzed, they will allow the future Geno Auriemma to map out the plays that will win him his 17th national championship.

Yes, for the WCBB the implementation of the new data capture and analysis system is still some ways in the future, but it is coming. And though giving up the box scores and composite stat tables will be tough in the end of the world as we know it, overall I feel fine. Details at Databall.

 

[rvwsleep]

While plowing through some data visualization blogs, I stumbled on an article about how the NBA is launching into a new mapping technologies and data analysis system that will soon revolutionize the world of basketball stats. No longer will we be arguing about who is the AAC's most improved player or best defensive player based on pretty limited value stats like points per game or total blocks.

With the help of overhead cameras that map all the moves the ball handler and other players make on the court, the key stat will be the EPV (Expected Possession Value), which will spin off likely many dozen subvalues to show how valuable a player is on the court. Instead of being bound to an old box score set of stats, we will now have far more nuanced stats that will show us just how effectively the future Stef Dolson set picks or future Moriah Jefferson gets a steal when cheating to the left. And when the stats are analyzed, they will allow the future Geno Auriemma to map out the plays that will win him his 17th national championship.

Yes, for the WCBB the implementation of the new data capture and analysis system is still some ways in the future, but it is coming. And though giving up the box scores and composite stat tables will be tough in the end of the world as we know it, overall I feel fine. Details at Databall.

Interesting. I'm most looking forward to new defensive metrics. Because, no matter how detailed the scoring stats are, it's still only half the game.

 

[huskybill]

Interesting. I'm most looking forward to new defensive metrics. Because, no matter how detailed the scoring stats are, it's still only half the game.

I can wait.

 

[DobbsRover2]

I can wait.

I know, I know. You've invested heavily in that hb steals to turnover stat as the key to all basketball stats brilliance, but you should come to your senses and see that the ratio of successful picks to successful block outs to successful hustle plays is really what it's all about.

 

[Scotter]

Even box score statistics aren't the same since the play by play record of the offical scorer is now archived and available online even for women's basketball. Just using that information results in stats pages for players that look like this.

 

[DobbsRover2]

Even box score statistics aren't the same since the play by play record of the offical scorer is now archived and available online even for women's basketball. Just using that information results in stats pages for players that look like this.

Yep, the play by play is invaluable for reconstructing the games and analyzing the ups and downs of the course of a game when a 10 point run by one team is followed by a 12 point streak by the other team. The box score flattens out the stats and hides a lot of the variation that 0ccurs in any game.

 

[ndmb]

Really interesting article.

The camera's will often help with current statistics, namely giving an accurate number of possessions (the ones based on the current box score are an approximation.

 

[DobbsRover2]

Really interesting article.

The camera's will often help with current statistics, namely giving an accurate number of possessions (the ones based on the current box score are an approximation.

Yeah, all though by "possession" the EPV is looking at things a little differently than how we now define possession. Every step a ball handler takes can drastically change the EPV value for what her team can expect to score from the present set up, so it is a free flowing value. And it will also be interesting to see how they define defensive EPV measures. There will be a lot of tinkering with this system during the next few years as they learn more.

 

[ndmb]

Yeah, all though by "possession" the EPV is looking at things a little differently than how we now define possession. Every step a ball handler takes can drastically change the EPV value for what her team can expect to score from the present set up, so it is a free flowing value. And it will also be interesting to see how they define defensive EPV measures. There will be a lot of tinkering with this system during the next few years as they learn more.

Yeah, I get that. I was referring to other stats, like Usage percentage or turnover percentage (along with pretty much every statistic that adjusts for the pace of the game).

One question that I would have is how in depth does it go. The article focused on the ballhandler, but that's really just the tip of the iceberg. Imagine Moriah driving to the basket under two circumstances, one in which Kaleena is on the wing and another in which a generic 30% 3pt shooter is on the wing. If Kaleena's defender stays on her, Moriah has an easier shot than she would if the defender came to help. Kaleena should get some of the credit for that play, even if she doesn't actually do anything, but how much credit is a complicated question.

 

[Zorro]

Lor' luv a duck!

 

[Kibitzer]

While plowing through some data visualization blogs, I stumbled on an article about how the NBA is launching into a new mapping technologies and data analysis system that will soon revolutionize the world of basketball stats. No longer will we be arguing about who is the AAC's most improved player or best defensive player based on pretty limited value stats like points per game or total blocks.

With the help of overhead cameras that map all the moves the ball handler and other players make on the court, the key stat will be the EPV (Expected Possession Value), which will spin off likely many dozen subvalues to show how valuable a player is on the court. Instead of being bound to an old box score set of stats, we will now have far more nuanced stats that will show us just how effectively the future Stef Dolson set picks or future Moriah Jefferson gets a steal when cheating to the left. And when the stats are analyzed, they will allow the future Geno Auriemma to map out the plays that will win him his 17th national championship.

Yes, for the WCBB the implementation of the new data capture and analysis system is still some ways in the future, but it is coming. And though giving up the box scores and composite stat tables will be tough in the end of the world as we know it, overall I feel fine. Details at Databall.

I invite your attention to a recent issue of ESPN Magazine, with a big discussion of "Analytics." All manner of stats are folded, spindled and mutilated and the result is that some surprising conclusions result. For example (I am recalling, not always reliably), for example, that Dwyane Wade was judged to be the least dependable clutch perimeter shooter, since his 4th quarter percentage (or was it his horrible accuracy in very close games?) was so poor, unlike stats for first three quarters.

There is an abundance of grist for the stat mill for wcbb afficianados.

 

[DobbsRover2]

Yeah, I get that. I was referring to other stats, like Usage percentage or turnover percentage (along with pretty much every statistic that adjusts for the pace of the game).

One question that I would have is how in depth does it go. The article focused on the ballhandler, but that's really just the tip of the iceberg. Imagine Moriah driving to the basket under two circumstances, one in which Kaleena is on the wing and another in which a generic 30% 3pt shooter is on the wing. If Kaleena's defender stays on her, Moriah has an easier shot than she would if the defender came to help. Kaleena should get some of the credit for that play, even if she doesn't actually do anything, but how much credit is a complicated question.

This EPV system is still in its infancy, and at the moment I believe the focus is more on the ball handler. But as the system gets set up more fully, every move that any player makes on the court will be evaluated. You cannot give an accurate representation of a player's value without factoring the picks, box outs, and other non-ballhandler moves they make.

Of course if EPV becomes an obsession it is possible that players might try to game the system by just making moves they know are approved by EPV but not with a lot of effort, but if they stop working well the EPV value for the player will go down. And likely if the player does make EPV-valued moves, she'll likely get better results.

 

[DobbsRover2]

I invite your attention to a recent issue of ESPN Magazine, with a big discussion of "Analytics." All manner of stats are folded, spindled and mutilated and the result is that some surprising conclusions result. For example (I am recalling, not always reliably), for example, that Dwyane Wade was judged to be the least dependable clutch perimeter shooter, since his 4th quarter percentage (or was it his horrible accuracy in very close games?) was so poor, unlike stats for first three quarters.

There is an abundance of grist for the stat mill for wcbb afficianados.

I'll leave it to Meyers to proclaim on whether Suarez is better than Messi, but an 8-year-old kid back in 1965 can tell you that it's hogwash that Carlos Beltran is a better base stealer than Maury Wills was. Just no way. I mean one has not even played on the Dodgers.

Just so many sports stats and analytics to boggle the brain and mush the mind, which explains why certain posters on the BY are the way the are.