Re: Predicting a champion in a sports league

On 25 Jan 2006 08:08:06 -0800, jdoiii@xxxxxxxxx wrote:

> I have various sports statistics (offensive and defensive performance
> measures) for multiple teams over multiple years in a single league.
> At the end of each year, a champion is determined by playoff among the
> top teams, thus the champion is not necessarily the team with the best
> winning percentage that year. I am trying to determine which measures
> are significantly associated with being the champion for any one year.

Who will win the playoff between the *top* teams?

That varies between sports, but "luck" usually has
a big role. Baseball, for instance, is so indeterminate
that teams near the bottom will win 40% of their
games during the regular season. Who wins the World
Series? Maybe the Yankees were the perfect "clutch" team
when Yogi Berra was their catcher, but most Series are a toss-up.

Maybe it is a superb goalie in ice hockey or a superior
quarterback in (American) football, or the coaching staff.
Coaches have been credited in basketball, too. Coaches
are responsible for both the "game plan" and for keeping
their athletes sober and sane, for one game or for days or
weeks of playoffs... which reminds me, "They've been
there before" is frequently mentioned.

Are you going to quantify those?

Are you looking to select between the teams that are
"Champs", or are you looking to describe the general
requirements of being superior?


>
> My understanding is that logistic regression is not appropriate because
>
> the outcome observations are not independent; that is, if Team 1 is the
>
> champion (=1), then all the other teams must =0. (Also
> the various performance measures for any one year are all interrelated
> between the teams, not sure if that makes a difference.)
>
> Would be glad to hear any recommendations/advice/references on how to
> best
> go about this.

I suggest, read up on prediction of sports, or your sport in
particular. Come the finals, newspaper columnists will
regale you with theories, but other people are probably
writing articles or books about the same things.

"Chance", a magazine of the American Statistical Association,
has published at least one article on "point spreads" for betting
in American football. But I think the message of the article was
how unpredictable the outcomes are, *given* the superiority as
shown by eventual season records.


--
Rich Ulrich, wpilib@xxxxxxxx
http://www.pitt.edu/~wpilib/index.html
.

Relevant Pages

  • Re: Predicting a champion in a sports league
    ... >I have various sports statistics (offensive and defensive performance ... >measures) for multiple teams over multiple years in a single league. ... The model assigned each team an offensive and defensive strength. ...
    (sci.stat.math)
  • Re: Whos Counting: Accounting for Lower Girls Math Scores
    ... though I don't think it has anything to do with gender per se. ... >>personally care very little about organized sports statistics. ... > protect womens rights in the educational systems. ...
    (sci.geo.geology)
Quantcast