Saturday, 30 November 2019


Its the annual fixture review time again. Because footy season never stops.

Past fixture reviews as per here
2019.     2018.     2017.      2016.     2015.     2014.     2013.     2012.

The well known raison d'être of the AFL fixture is to fix games so that top teams are handicapped by playing better teams, and the bottom teams get a boost along by playing other poorly playing teams.

And it works. Look at 13th from 2017 and how well they went in 2018. And Brisbane, rising to a Prelim before anyone expected.
Get a good fixture, and a good playing group... and its yours.

Anyway, the system is well known, and also is geared to allow the League wriggle room to schedule big games (derbies, rivals, etc) regardless of their own handicapping rules.
Its rules for some...

FMI Assessment 2020 (ves 9).

Time for the FMI machine to give us some 2020 vision, while still walking 'round blind., And once again the FMI assessment relies on two two key planks to determine the fixture balance. First, the determinant on teams' strength in the competition is via the FMI rankings alone. Not simple ladder position, nor number of wins or Supercoach player data.
Those rankings are then applied to the AFL's own fixture, home and away.

A summary of the past FMI assessments of the leagues scheduling is as below, where basically the League has been successful in handicapping the fixture. A 'type' of fairness.

2015: "...all in all, it looks like the AFL have achieved their outcome, so ... well done!"
2016: "the KPI seems to have been met and met well this year. A few exceptions to the rule, but a decent synergy between the objectives and the outcomes."
2017: "So overall, its another win for the AFL in its programme of handicapping its own field."
2018: "the handicappers have basically done their job and the fixture is tilted as per preferred scenarios from HQ".
2019: "an amazing result of the top six teams having the six hardest scheduled, and the bottom six that have the easiest six."

And so to this years assessment, which after 2019's almost flawless outcome, is... a bit of a let down.
The top six from 2019* fill 5 of the top 7 'toughest schedules' for 2020, with the outlier in Brisbane set as the 11th hardest / 8th easiest schedule.
Of the middle 6 from 2019, its a bit more of a scattered outcome. Anything from Essendon having the 4th hardest calendar, down to Port Adelaide with the 3rd easiest schedule.
The bottom 6 fit into the easier end of the scheduled, and only Carlton is misplaced from that cohort with the 8th hardest / 11 easiest schedule

* based on end of H&A season standings

3West Coast5124802.06
9North Melbourne12113400.60
10W Bulldogs/FFC711362000.52
11Brisbane Lions21190-1000.49
13Gold Coast18651-100-0.93
14St Kilda14910100-0.93
16Port Adelaide101131-300-1.26

When its lined to the blocks of finishing positions from 2019 as above, it highlights how out of whack the 2020 fixture looks compared to what the League tries to achieve. This one is a miss on the handicapping measure.

The last measure to be taken from the above is the spread of the tough draw index... the difference between the hardest and easiest.
The 2020 season spread is 5.28
This a decent sort of rating, and possibly an indication of the even-ness of the rankings, post the 2019 season (which really has only 1 outlier in Gold Coast).
The past spread data is as here.
   2020:    5.28
   2019:    8.32
   2018:    5.04
   2017:    8.65
   2016:    8.90
   2015:    8.98
   2014:    9.15

Later - the FMI draw assessment if the FMI split of the draw is applied.

But before that...

Other Assessments

Champion Data

The wall continues to go up and up, as the proponent of AFL data, those that claim the data is "The story behind the game" (as their website says today) are... not telling any stories behind the game.
Well not to you. You don't pay for the data and those stories. You don't get anything.

Yep, not even a tweet about a fixture assessment this year.
The sum total of Champion Data output was published in the controlling footy media (Murdoch, CrocMedia etc), and amounted to... a list.
A list of 'who got the hardest draw and who didn't.
No logic, theory or evidential data to back up the claims.
Nothing. Just the list below.

  1. West Coast
  2. Hawthorn
  3. Gold Coast
  4. Essendon
  5. GWS
  6. Carlton
  7. Richmond
  8. North Melbourne
  9. Geelong
  10. St Kilda
  11. Collingwood
  12. Western Bulldogs
  13. Brisbane Lions
  14. Port Adelaide
  15. Adelaide Crows
  16. Melbourne
  17. Sydney Swans
  18. Fremantle

ABC and HPN.

The HPN team have worked themselves into national broadcaster semi-regulars, and released their assessment quickly enough on the Thursday after the League release of data.

A lot more numbers and explanation, and two methodologies - using both the average wins (in 2019) of the 2020 opponents, and the second using percentage (from 2019).
On the former, the calculation appears to pit the average opponent wins for the 2020 draw against what is considered  'fair draw'. The fair draw? Playing opponents once only... which is fair enough, but seems to miss the home/away balance the league likes with an even number of rounds. Perhaps a better 'fair draw' would be to use the theoretical 34 game season... everyone home and away. That evens the numbers and nullifies H and A influence.

Anyway, read their piece and look at the charts. Its good stuff.

Graft Ratings

Going for the 34 game theoretical season as a yardstick is exactly what the GRAFT Ratings site elects to do, as tweeted out Saturday afternoon after draw release.

Later on Saturday night, an explainer of the numbers was issued. Nice.
And in a nutshell, the Graft system sims the season as dictated by the AFL, and also the omni-draw of 34 games. The difference between the most even structure (34) and the Leagues handicapped 22 game fix, shows who has the worst draw.
Its a clever way to work it out. Kudos Grafty.


It can't be a fixture release without Rohan Connolly's look at how it all plays out.
As per normal, my same critique remains. Using ladder position to split teams doesn't work for mine.
E.G. Hawthorn and Port Adelaide finished with 11 wins and almost 3.3% apart, and are split by 1 point. Melbourne and Gold Coast are also split by 1 point, but in reality it was 2 games and 18.1%. Its... not right.

Anyway, through the power of a little 'share' button on the Footyology post, the RoCo assessment is here. And of interest, there is some correlation to Graft's and the others above.

Matter of Stats

Also part of the furniture in the "can't be a proper look at the season fixture analysis without..." set of nesting occaisionasls is the The Matter of Stats assessment, by the the Doyen of Data analysis, Tony from Matter of Stats. His 'Strength of Schedule' post is the most thorough and amazing of all, done by someone not getting paid or chasing kudos.

Tony applies the most thorough approach to the analysis, looking at team ladder placement, venue performance, and determines not only an overall 'best/worst outcome' from the 2020 fixture, but also he looks at the better/worst fixtures playing at home or away. It might take some time to work through, but it is worth it.

No comments:

Post a comment