Hi, fellow INDUCTIVE e-mail list people (and apologies for any
cross-posting).

To the very best of my knowledge, the world's longest-running
probabilistic competition is 15 years old and now in its 16th year,
going back to Round 3 of the 1995 Australian Football League
(AFL) season.

The competition resumes in a few hours from now (slight update
due to delay in posting: started a few hours ago, with one game
now having been played), but it is a small matter if you happen to
miss the first game or the first few games.   After all, the season
has 22 x 8 + 9 = 176 + 9 = 185 games, and the scoring system of
1 + log2(p) means that a score of 0 in the first game or so is not a
particularly bad thing.

The Australian Football League (AFL) footy season starts on Thu
25/March/2010, and with it so does Monash University's (now 15y.o.)
probabilistic footy-tipping (or football prediction) competition.   This
competition is free for all to enter, and/but entrants not only choose
a team, they must also choose a probability that a team that will win.
(If entrants don't put in their tips [i.e., their predictions] before the
start of a game, this defaults to a 50%-50% tip of 0.5 and a score of
1 + log2(0.5) = 1 + (-1)  = 0.)

Scoring is logarithmic, and (in the long run) this rewards those
who are best at choosing probabilities.  With information as the
negative logarithm of a probability, this can also be said to be (as
far as we know) the world's longest-running compression-based
competition.

Further info' on joining (which is free and open to all), the
(logarithmic) scoring system, the history of the competition (which
goes back 15 years to 1995), media coverage, etc. is available via
links from www.csse.monash.edu.au/~footy .

For even further information (such as the uniqueness properties
of log-loss scoring [in footnote 175]), check out
D. L. Dowe (2008a).  Foreword re C. S. Wallace, Computer Journal
(Oxford Univ Press), Vol. 51, No. 5 (Sept. 2008) [Christopher Stewart
WALLACE (1933-2004) memorial special issue], pp523-560
http://dx.doi.org/10.1093/comjnl/bxm117
http://comjnl.oxfordjournals.org/content/vol51/issue5
www.csse.monash.edu.au/~dld/David.Dowe.publications.html#Dowe2008a
and search on (e.g.) ``footy'' or ``football''.

The pointer to Chris Wallace on this e-mail list is quite pertinent,
as he wrote the seminal Wallace & Boulton (1968) paper beginning
the area of Minimum Message Length (MML) inductive inference,
from which this (15 year-old) competition is a spin-off.


Cheers (and apologies again for any cross-postings),

David.

Assoc. Prof. David Dowe, Ph.D.; School of Comp. Sci & Softw. Eng.;
Clayton School of I.T.; Monash Univ.; Clayton; Vic 3168; Australia
Tel:+61 3 9905-5776  www.csse.monash.edu.au/~dld
http://dx.doi.org/10.1093/comjnl/bxm117
http://comjnl.oxfordjournals.org/content/vol51/issue5
www.csse.monash.edu.au/~dld/David.Dowe.publications.html#Dowe2008a

Reply via email to