APBRmetrics

[HoopStudies]

[wiLQ]

[wiLQ]

[Harold Almonte]

[Harold Almonte]

The Suns's seasons games differential (Amare on/off full season) has been 7 and 8 games. How many wins produced says WP is he worth?. Of course he worths a lot more. But, All what you (statistically) see from players is all what you'll get from them? in all situations?
Performance ratings talks about a zero sum players-team acomplishments, more usage more rating they take from low usagers, but hide player's potential, being clutch and value of replacing other player's usage two of them.

[Statman]

[jkubatko]

I don't understand why you're worrying so much about creating a model that perfectly predicts past results. By "forcing" your model to perfectly predict the past, you are modelling the variability in the data rather than the trend. Sure, your model would predict the past with no error, but would it be useful for new data?
_________________
Regards,
Justin Kubatko
Basketball Stats!

[Mark]

wiLQ, this is your study and you asked for ideas. You of course can pursue them or discard them as you wish.

I didnt suggest all-star voting influences MVP voting directly but all-star voting was a proxy I thought of for favorable opinion that might indeed be a useful variable in a set of many.

I do think that prior year MVP voting is a significant indicator or factor for current year MVP voting You didnt need it to get a good match, and there are reasons you may not want it in, but from the outside it seemed worth mentioning. Media articles on MVP race often say things like it is "a guy's turn" and some of these guys are voters.

As for media market size I still think it can have an effect, but I took a quick look at last 20 years and did find that small market Karl Malone did get first MVP the year he most deserved it and small market Kevin Garnett got his the year with his highest to date PER as well. There are many years where 3-5 guys have plausible cases and media market size or overall media attention may play some role but I won't do a detailed study of that. If you want to look at that or not that is up to you. Perhaps the media effect may show up more in how high the also-ran candidates were vs. what stats would suggest. There may be more distortions on that than who is the #1, the MVP.

[Harold Almonte]

I think Statman's was a very good point. It would need to question some of that year MVP voters to know what were they minded about Nash's previous MVP, Stoud. MVP points, Suns expectations, Billups, and the season second half.
An MVP in some way means a stepping up, you don't win it with your average stats, and to loose 7 more games only without Stoud. was "supposed" to be a big stepping up.

[wiLQ]

[jkubatko]

[wiLQ]

[Mike G]

You could almost certainly explain all mvp's with one formula, if you include height and skin color. But then you might get results saying Stockton 'should have' won a few. Or maybe not, since his teams never improved dramatically. Tops was 9 wins, from '96 to '97; Malone won that year.

Some people still swear Kidd should have won in '03 -- based almost entirely on the Nets' improvement. In a best-of-7 Finals, was he anywhere near the best/valuablest player? Was Iverson the b/v in the '01 Finals? What did they have that Shaq/Duncan didn't have? -- Shortness, and an inability to dominate.

I don't remember -- is incumbency a negative in your formula? Jordan was smaller than (mvp's) Barkley and Malone, but he sure wasn't less valuable in their Finals.
_________________
40% of all statistics are wrong.

[wiLQ]

[DLew]

What Justin is trying to say is that the point of a model generally is to try to predict the future, not describe the past. If you force your model to completely describe the past, including data points that are outliers, then you will probably be less accurate forecasting the future.