APBRmetrics

wiLQ · Joined: 23 May 2007 Posts: 22 Location: Poland

Hello,

I may be a new user to you but this forum is definitely not new to me.
I read it since summer of 2006 and that's why I'm writing this post.
Because I'm curious about your opinion...

In December 2005 I came up with the idea of MVP Points.
It was an attempt to find what winners of that award had in common and then
create formula which could describe it using various numbers and statistics.
While doing it I had in my mind only one goal:
predict who will win MVP on actual and every future season.
At that time I thought that I've succeeded.
Thanks to Roland it was published on 82games.com:
http://www.82games.com/wilq.htm
with the follow-up and one major tweak:
http://www.82games.com/wilq3.htm

Please, give it a read and tell me what do you think about it.
Is it anive and stupid idea? Good concept with bad execution?
OK, but useless? Nice curiosity but Nash's MVP in 2006 ruined it?
I appreciate every opinion especially those critical with justification ;-)

P.S. I want to apologize for all mistakes in this and all my future posts in this forum.
My English is definitely not as fluent as it should be but I hope all my intentions and
opinions will be understandable to you...
_________________
regards
wiLQ

Mark · Joined: 20 Aug 2005 Posts: 807

I recall the article and was impressed you found weights that fit with 20 years of MVP selections. But you may need at least one more criteria- I think Billups' failure to win shows that it is to some extent a popularity contest and the winner may have to come from top 5-10 most talked about players in general. Billups was probably well below top 10 on that and didnt have as strong a chance to win the regular season MVP as his performance warranted. Final MVP can overcome this. Billups and Detroit were considered an old story while Nash and Suns was a new one. Maybe you could incorporate a criteria for google references and maybe even media market size. Not sure how many points to award for this or deduct.

HoopStudies · Posted: Wed May 30, 2007 3:11 pm Post subject:

My memory is telling me that Billups was the first half MVP by consensus, but the fact that Detroit dropped a bit in the 2nd half hurt his chances. In that light, you might look into impact of 2nd half vs 1st half performance.

I'm not sure writers would like it if a statistical formula can predict their behavior Smile

Welcome to the discussion.
_________________
Dean Oliver
Author, Basketball on Paper
http://www.basketballonpaper.com

wiLQ · Joined: 23 May 2007 Posts: 22 Location: Poland

wiLQ · Joined: 23 May 2007 Posts: 22 Location: Poland

cwood · Joined: 30 May 2007 Posts: 5

what about including a lagged team improvement variable? i know with nash that there still was carry over in terms of what he "meant" to his team which should be represented by the increase in wins when he came to the team. perhaps only include this one season lag when its only the player's second season with the team or another route would be to include a different variable representing team improvment for a player on the first year with his team.

I like these articles, it takes a lot of the subjectivity out of the mvp talk. what was your method? did you tweak the values until you got a perfect fit for the historical mvps or did you do some type or regression?

Harold Almonte · Joined: 04 Aug 2006 Posts: 224

The only thing that comes to my mind is that it could have been the overrating of the Stoudemire variable. What was the big difference of Stoudamire being on, then off, and then on again?

kjb · Joined: 03 Jan 2005 Posts: 665 Location: Washington, DC

I have no idea whether this would have any application in predicting the MVP award, but have you considered inserting a category for race?
_________________
My blog

Mike G · Joined: 14 Jan 2005 Posts: 1509 Location: Delphi, Indiana

Does 'race' solve the Nash anomaly, when he was also behind Dirk in this 'formula'? Maybe 'continent of origin' as well?

Pettit and Bird were good white mvp picks. Cowens, Walton, and Nash were questionable.
_________________
40% of all statistics are wrong.

kjb · Joined: 03 Jan 2005 Posts: 665 Location: Washington, DC

I don't know if race has any applicability at all. I was just throwing it out there as something to consider. The issue of Nash's race in connection with his MVP awards has been raised -- at least on other message boards.
_________________
My blog

Mike G · Joined: 14 Jan 2005 Posts: 1509 Location: Delphi, Indiana

But it was a good point to raise, Kevin. Why did Nash win (particularly his first one)?

There are many ways to check white/black players on all-league teams, Hall of Fame, allstar teams, 'dream' teams, etc. There are disproportionately more 'underqualified' white members of these groups. They just tend to have lesser stats, whether by averages, PER, or whatever measure.

I read the original article and noted Nash's winning, counter to the prediction. What's good about such 'prediction formulas' based on subjective votes is that they can and do reveal the thought 'process' of a collective mindset. The voters themselves may well not know why they vote as they do. But where there's a pattern, there's a reason.

Shortness is also a plus in MVP voting. Probably in those other electables I mention, too.
_________________
40% of all statistics are wrong.

cwood · Joined: 30 May 2007 Posts: 5

wiLQ · Joined: 23 May 2007 Posts: 22 Location: Poland

Mark · Joined: 20 Aug 2005 Posts: 807

wiLQ, I was aware that google references would be negative and positive but I still think it would weed out guys who have too low a profile. But if you dont want to use that metric you could use previous year MVP votes or fan all-star votes or some other favorability metric. Media market size not decisive between Nash and Billups but media market size probably was a contributing factor is some previous year choices and could be in the future.

wiLQ · Joined: 23 May 2007 Posts: 22 Location: Poland