New work on measuring defense

[Crow]

http://arxiv.org/pdf/1405.0231.pdf

11 days after publication. Might be big. Get into to it while it is hot.

[nrestifo]

This is very big. I'd be interested to see how many times their algorithm mis-identifies defender-offender relationships, but their player evaluation results, at least the ones they showed in the paper (areas 5 and 1), pass the eye-test.

[Crow]

Thanks for checking it out.

Thr measure of defensive entropy (time split defending different players) was higher than I expected. The team values vary enough to be worth considering further (design, effort and effectiveness). Offensive similarity network for continuum of data there. Three sizes / roles of defenders deemed adequate. Very interesting to see defensive impacts in restricted area by efficiency and frequency. Are these adjusted or fully adjusted? Will have to look again. Few perimeter played affect restricted area efficiency by their defense; for bigs it is few affecting frequency of such attempts. Using the product of these makes more sense than just one, as some analysts are currently doing. Finding best defensive matchups and matchup types? This data set should earn lots of consulting fees.

I'll look closer at other articles as I have time and as others show interest.

[Crow]

Hypothetically players defense could be measured by activity preceeding shot instead of just shot data.

This article talks about finding similarity in behavior. http://ieeexplore.ieee.org/xpl/login.js ... %3D5476192

I haven't read it yet. But I can see value in video tempural slice analysis. What was defender doing 3, 2, 1 second before shot? Good to know results, better to know how everyone got there, especially the best defenders (and the worst).

[Crow]

Might have some commonality with previous citation. http://dircweb.king.ac.uk/readinggroup/ ... nition.pdf

[Crow]

the future is coming
http://ieeexplore.ieee.org/xpl/login.js ... %3D5995386

[Crow]

computer warnings about open offensive players
http://www.sciencedirect.com/science/ar ... 0312000995
the life of video analysts may get easier, but busier, more important?

[fpliii]

Thanks for sharing. BTW this link:

Might have some commonality with previous citation. http://dircweb.king.ac.uk/readinggroup/ ... nition.pdf

seems to be dead for me.

[Crow]

Yep, dead, at least right now.

Alt chances: http://citeseerx.ist.psu.edu/viewdoc/su ... .1.65.4787
http://dblp.kbs.uni-hannover.de/dblp/Se ... 2&mlt=true

Neither worked for me, but my phone is being slow and tempermental in general. Might work for others, sometime or maybe not.