Thursday, 14 March 2013

Numbers Crunched - 2013 Kick-Off

Regular readers of the FMI blog will know we don't consider our rankings to be the bees knees and completely accurate. And as transparent as we like to be (you see all our ranking numbers and changes), we also acknowledged that we can improve our rankings, and ran the Project: Renewal (now archived) page detailing the different results.

With our first off-season under our belt, and a solid year of experience behind us, we have altered our tipping calculations and algorithms numbers. Primary to this was improving not only tipping accuracy, but 'margins of tips' accuracy. The guide for this was to use the scoring system laid down by the "Monash University Probabilistic Footy Tipping Competition", under their 'Normal' conditions.
We wont describe the rules here... they are accessible at the link if you care to check them out.

With that as our guide, we were able to retrospectively amend the calculations and attune last years tips, and every years tips, right back to 1897. In doing this, we determined the optimal performance (over the longer term), and we will use these calculations for this season.

Accordingly, below are the new rankings data for the 18 AFL teams under the new system, compared to the previous system we employed.
We have also taken the liberty of adding the power rankings from Troy Wheatley's blog, as well as the mysterious numbers generated by the equally obscure "Robey" (from the BigFooty forums pages).
Just so you can see not only how our methods stack up, but also others as well.

click to expand

The key thing to note, other than the disparity in numbers compared to the old system, will be that the movements of teams rankings through the season should be more dynamic.

As to accuracy, the data changes boosted our tipping performance over the past 10 seasons against the measures in the Monash Uni system, including a few 2nd places, a few 3rds, and an equal 1st last year. Disappointingly  we also got a few 15th-20th place finishes too... proving yet again that tipping footy and wagering... its a mugs game.


  1. Looks like we now all agree that Sydney was a clear #2 who were good enough to upset the #1 team on Grand Final day. (Although hard to tell under Robey's system how clearly the Swans were in second place.)

    I noticed all the points totals are higher under the new system - what's the explanation?

    1. Correct... Swans into 2 is better. I think I always underrated them last year... 7th for first 15 rounds!

      As to the increased numbers, all (bar Bulldogs) go up, avg about 10%. Due to increasing the 'game weighting' which attributes more points per game, and makes the ranks more dynamic. Additional tweeking to the numbers also helping.

      Bulldogs have gone down as they were over rated cf performance last year, and this is a truer re-adjustment.