Traditionally football tipping competitions have involved tippers predicting the winners (and optionally the margins) of the games in any given week. The scoring systems are arbitrary and often are based solely on the number of correct tips, or sometime include bonuses for accurate prediction of the winning margin. The School of Computer Science and Software Engineering (CSSE) (formerly Department of Computer Science) has run a similar competition for a number of years. The new tipping competition rewards tippers in a mathematically sound way, and allows tippers to more directly express the degree of certainty in their tips.
In its current form the current tipping competition allows the user to enter a traditional competition, a probabilistic computation and a Gaussian competition.
The notion of probabilistic scoring for a binary event (i.e., an event with two possible outcomes, or a 2-way event) was first described by I.J. Good in 1952. To the best of our knowledge, it was first used on real data in Dowe and Krusel (1993), where it was used for binary events and where it was also extended to and used for 8-way events (i.e., events with 8 possible outcomes). The extension of 2-way to 8-way, 10-way or n-way is quite straightforward. Further 2-way logarithm of probability (log prob) scoring occurs in Needham and Dowe (2001) and Tan and Dowe (2002), the latter of which also introduces and uses 10-way log prob scoring. To the best of our knowledge, these works described above all pre-date other applications of log prob scoring. Tan and Dowe (2003) (section 5.1) surveys this area.
After Jon Oliver suggested in passing to David Dowe that he/they/we might apply this notion of probabilistic scoring to footy-tipping, rather than let it drop there and then, David Dowe contacted an excellent and obliging programmer, Kevin Lentin, and :-) the rest is history :-) .
The Probabilistic competition began in Round 3 or 4 of 1995, initially with 8 entrants in that very first round. It was run manually via email, with results being calculated in spreadsheets. Since its inception in 1995, the competition has been created, founded, organised and overseen by David Dowe - with the maintenance (and hard work) done by a variety of others as listed above (Kevin Lentin, 1995) and below.
At the beginning of 1996 John Hurst wrote football tipping software whereby tippers within the department could enter their tips using a UNIX terminal program and those outside the department (or around the world) could use an email interface. Tippers were automatically emailed a tipping template each week. When this completed template was emailed back to the email server their tips were automatically logged in the system.
In Round 4 of the 1996 season the Gaussian competition was introduced. The scoring system was devised by David Dowe so that, as in the probabilistic competition, the object of the tippers was again to choose parameters so that their expected Kullback-Leibler distance from the outcome of the game was minimised.
In 1997 the maintainence was taken over by an Honours student Matt Doran as part of his thesis project. The main improvement was an official WWW site and a WWW interface for the original server written by John Hurst.
Bernie Meyer talks to the MML group on 22nd April 1997 on "information theory, honesty, and multiple choice tests." Graham Farr introduces the notion of boldness.
The fact that the Footy Tipping program existed in many different forms (command line, email, web) meant that it was hard to maintain. This resulted in it being completely rewritten and maintained in 1998 by David Powell and Torsten Seemann. The main database engine and CGI scripts were written in Perl5.
Prizes now given to the top 10 primary or secondary students in the probabilistic competition.
About this time, Kevin Korb introduced a notion of calibration of the various entrants.
Maintained by Torsten Seemann. Mainly cosmetic changes and bug fixes to the web site to make it clearer to tippers from outside CSSE, especially secondary school students.
Maintained by Torsten Seemann.
Due to lack of funding, monetary prizes for top student tippers were discontinued.
Maintained by David R. Powell.
Those interested in learning more about the science of probabilistic football tipping can browse our list of relevant publications .
Clayton School of Information Technology,
Faculty of Information Technology, Monash University.
Web site originally designed and developed by David Powell and Torsten Seemann.
Please direct queries to firstname.lastname@example.org.
Safe Content Rating.