Other "contests"

[Crow]

Daniel mentioned kaggle contests recently. Are there other contests people think would be valuable to offer / engage in besides overall win prediction of one metrics / blends? Any teams want to sponsor a public challenge? Morey challenged for prediction of future 3pt fg% years ago. Anyone recall who won or if winning formula was released?

[sndesai1]

predicting which players will have the largest change from the previous year for a particular metric

[Crow]

A possibility. A test of aging curves, similarity systems, true skill estimation and regression to mean.

Others might be league average team 3pt attempts per game, final east / west interconference game win%s, league average efg%, most effective lineup over 250 minutes in season and projected into playoffs, etc.

[permaximum]

Well somebody will come out and say there should be at least 50 out-of-sample seasons, there can't be any blend (every metric is kind of a blend but you know you're a thief if you want to use a blend) and even if you qualify they can't be sure unless you explain the steps to reproduce the metric.

Kidding besides, I would like to see "beating vegas" contest. Without any rules...

[Crow]

Trying to think ahead for worthy topics / work but will put this on backburner for awhile to let folks concentrate on the one metric comparison first.