6 ms·

Losing the best player on a team can improve it

26 points by dangoldin 17y ago

Actual paper: http://arxiv.org/PS_cache/arxiv/pdf/0908/0908.1801v1.pdf

jwb119 17y ago
Ewing Theory 101: http://sports.espn.go.com/espn/print?id=1193711 http://sports.espn.go.com/espn/print?id=1193711
- dangoldin 17y ago
  Thanks for the link, I just finished the arxiv paper and it refers to the ESPN article.
dangoldin 17y ago
Actual paper (maybe a moderator can point to it?) http://arxiv.org/PS_cache/arxiv/pdf/0908/0908.1801v1.pdf http://arxiv.org/PS_cache/arxiv/pdf/0908/0908.1801v1.pdf
callmeed 17y ago
The key word is "can" ... I'm sure there are more examples of a team losing their best player and getting worse. Does anyone really think Davidson will be better next year without Stephen Curry? Also, "best player" is so subjective in sports like baseball and football where players have specialized roles. Who is the Red Sox best player? You could argue 3+ different people.
- DannoHung 17y ago
  Perhaps Football would be applicable if you had a star receiver or running back. Not sure about a QB getting taken out. Perhaps in Baseball the first base position or maybe a DH or reliever?
v3rt 17y ago
The article mentioned that closing main roads temporarily could have a similar effect - could teams thus experiment with benching star players , or altering plays to "mix it up" more, without actually losing the player? This would seem to be the best of both worlds, by using the star optimally. In general, this seems like a lesson to not focus so much on the "strongest point", but try to squeeze out more returns from weaker points before diminishing returns set in so heavily. E.g. - don't make your strongest developer in a small team do all of the hard work, even if his code is much better than others'. His ninth hour of work is less productive than another developer's second.
jcl 17y ago
This article reminds me of another posted to HN about a basketball player with mediocre stats who greatly increased the effectiveness of his teammates: http://news.ycombinator.com/item?id=481392 http://news.ycombinator.com/item?id=481392
zach 17y ago
Oh boy, Malcolm Gladwell must be 3,000 words into a new essay already. This is right in his wheelhouse.
dkarl 17y ago
Unlike the Braess Effect for drivers, this is a phenomenon that a team can collectively understand and overcome. After all, if all the players pretended that the star was just an average player, the result would be better than if he really was an average player. And that isn't even the optimum way to handle it. In the pathological case, a team is like a codependent spouse that obsesses over the star and contrives ways to be dependent on him even in situations where he shouldn't be involved. In the healthy case, a team collaborates to exploit their best player by not depending on him. When the defense adapts to shut down their best scorer, his teammates get more openings. Trivial example: when the defense double-teams the best offensive player, the remaining offensive players get to play 4-on-3. The presence of a stellar offensive player makes it easier for his teammates to score. The Chicago Bulls with Jordan were a great example. If it didn't work for the Knicks with Ewing, don't blame the presence of the star. Blame the coach and the players for not using him effectively. Why a pro team would be unable to do this is perhaps a more interesting problem. Properly coached high school teams understand this and can execute it. Establish that Joe is an inside threat, then exploit the defensive adaptation by scoring from the outside. A pro team with a thoroughly scouted star gets to skip the "establish the threat" step and skip straight to exploitation. Why can't they do it? Maybe it has something to do with the Shane Battier article that appeared on HN several months ago -- if NBA teams and players are obsessed with individual statistics, perhaps they think like drivers in the Braess Paradox instead of thinking like a team.
jpwagner 17y ago
"For example, the actions of the defense aren't modeled at all." That's where I stop reading...luckily it was near the end.
akd 17y ago
The shot fraction graph doesn't hold any water. Ray Allen's shot fraction is likely to be the highest when his teammates are performing poorly -- thus he has to take on more of the burden himself. The effect may not be due to him at all, but rather that when 4 players are playing poorly, the team tends to lose.

26 points by dangoldin 17y ago

Actual paper: http://arxiv.org/PS_cache/arxiv/pdf/0908/0908.1801v1.pdf

jwb119 17y ago

Ewing Theory 101: http://sports.espn.go.com/espn/print?id=1193711 http://sports.espn.go.com/espn/print?id=1193711

dangoldin 17y ago
Thanks for the link, I just finished the arxiv paper and it refers to the ESPN article.

dangoldin 17y ago

Actual paper (maybe a moderator can point to it?) http://arxiv.org/PS_cache/arxiv/pdf/0908/0908.1801v1.pdf http://arxiv.org/PS_cache/arxiv/pdf/0908/0908.1801v1.pdf

callmeed 17y ago

The key word is "can" ... I'm sure there are more examples of a team losing their best player and getting worse. Does anyone really think Davidson will be better next year without Stephen Curry? Also, "best player" is so subjective in sports like baseball and football where players have specialized roles. Who is the Red Sox best player? You could argue 3+ different people.

DannoHung 17y ago
Perhaps Football would be applicable if you had a star receiver or running back. Not sure about a QB getting taken out. Perhaps in Baseball the first base position or maybe a DH or reliever?

v3rt 17y ago

The article mentioned that closing main roads temporarily could have a similar effect - could teams thus experiment with benching star players , or altering plays to "mix it up" more, without actually losing the player? This would seem to be the best of both worlds, by using the star optimally. In general, this seems like a lesson to not focus so much on the "strongest point", but try to squeeze out more returns from weaker points before diminishing returns set in so heavily. E.g. - don't make your strongest developer in a small team do all of the hard work, even if his code is much better than others'. His ninth hour of work is less productive than another developer's second.

jcl 17y ago

This article reminds me of another posted to HN about a basketball player with mediocre stats who greatly increased the effectiveness of his teammates: http://news.ycombinator.com/item?id=481392 http://news.ycombinator.com/item?id=481392

zach 17y ago

Oh boy, Malcolm Gladwell must be 3,000 words into a new essay already. This is right in his wheelhouse.

dkarl 17y ago

Unlike the Braess Effect for drivers, this is a phenomenon that a team can collectively understand and overcome. After all, if all the players pretended that the star was just an average player, the result would be better than if he really was an average player. And that isn't even the optimum way to handle it. In the pathological case, a team is like a codependent spouse that obsesses over the star and contrives ways to be dependent on him even in situations where he shouldn't be involved. In the healthy case, a team collaborates to exploit their best player by not depending on him. When the defense adapts to shut down their best scorer, his teammates get more openings. Trivial example: when the defense double-teams the best offensive player, the remaining offensive players get to play 4-on-3. The presence of a stellar offensive player makes it easier for his teammates to score. The Chicago Bulls with Jordan were a great example. If it didn't work for the Knicks with Ewing, don't blame the presence of the star. Blame the coach and the players for not using him effectively. Why a pro team would be unable to do this is perhaps a more interesting problem. Properly coached high school teams understand this and can execute it. Establish that Joe is an inside threat, then exploit the defensive adaptation by scoring from the outside. A pro team with a thoroughly scouted star gets to skip the "establish the threat" step and skip straight to exploitation. Why can't they do it? Maybe it has something to do with the Shane Battier article that appeared on HN several months ago -- if NBA teams and players are obsessed with individual statistics, perhaps they think like drivers in the Braess Paradox instead of thinking like a team.

jpwagner 17y ago

"For example, the actions of the defense aren't modeled at all." That's where I stop reading...luckily it was near the end.

akd 17y ago

The shot fraction graph doesn't hold any water. Ray Allen's shot fraction is likely to be the highest when his teammates are performing poorly -- thus he has to take on more of the burden himself. The effect may not be due to him at all, but rather that when 4 players are playing poorly, the team tends to lose.