REFORM: THE FIXTURE

We have posted before on this blog about the need for the AFL to reform its business. To ensure the strength of the competition throughout its ranks, to retain the mix of clubs, its history and traditions, and to retain the involvement and participation of the broader football community. This means creating a system where all teams have a reasonable expectation of success within a not too distant time frame, and all fans and supporters are embraced, and the voices of a few do not over-ride the whole.

Accordingly, the proposal below is our offering to the debate on reform to the AFL competition. We welcome your comments.

NOTE: the below was originally published in 5 individual posts, links here:
   ONE  TWO  THREE  FOUR  FIVE



   1. THE PROPOSITION

There are many reasons for change inside the AFL - draft tampering, tanking, rules around rookie-listing, and so on. The one that gathers the most popular support though, is for ‘making the fixture [1] fair’. In the context of most football followers, that means every team playing the other, either once in a 17 round structure or twice, home and away, in a 34 round monster season.

We at the FootyMaths Institute think the AFL ‘draw [2]’ should be structured to meet that end, as best as possible with an 18 team competition, AND it should also be free from the other fixes imposed by the AFL. It is clear that the fixture is rigged to generate revenue for clubs and the league, to drive up membership of select clubs, and relegates others to scratching out existences.

We believe that the playing schedule should not only be made fair in terms of balancing out teams playing home and away, but also fair in terms of ‘opportunity’. The opportunity to meet popular clubs home and away each week, as well as the opportunity for free to air television coverage to boost brand awareness and supporter base.

We believe that the schedule should also be removed from any notion of ‘soft draws’ and preferential treatment to some clubs, at the expense of others. We also believe that the competition should also not convert to any tiered structure, involving promotion/relegation or otherwise.

Our proposal below is clear, transparent and removes the gerrymanders of the current fixture, but still leaves open the AFL to schedule blockbusters into appropriate dates (itself a formm of gerrymandering of the draw, but we consider this proposal a first step to balance). It instills an essence of randomness to each seasons draw, and it does so in a simple, structured and untainted way.

We also believe the other structures in the AFL should remain the same;
     - a season based on 22 rounds,
     - with the current top 8 finals system, and
     - the draft employed in its current state.

Also, the AFL is now the pre-eminent sports competition in Australia. It dominates the media in all states, bar NSW and Queensland. Its televised product is the most expensive annual sporting product in Australia. It is the only truly national competition with multiple teams in each state (barring Tasmania and the Territories).

It is our belief that the AFL is such a sporting juggernaut within Australia, and is now so powerful and financially secure, that it can now finally fling-off the crutches it has used to hold itself up and set itself free – namely that it is time to ditch the gerrymandered fixture.


SYSTEMS THAT WON’T WORK:
From our assessments, we believe the below are unworkable systems the AFL should avoid:
A- A season where each team plays others only once, for a 17 round season [3].
Playing teams once a season would not be preferred as per our previous blog post, 'A Very Good Year?'.
B- Fitting 18 teams into a home and away structure involving 34 rounds [3].
This would result in too many games, too much wear and tear on players and playing surfaces. Clubs would need larger lists to provide more support to the playing stocks to cover fatigue and injury.
C- Conferencing the 18 clubs into 2 groups of 9 teams [4].
Possibly a workable system, though 9 teams means 8 or 16 rounds per season in each conference. And there is also a bye each week for each conference. And as there is a bye, to completely play out each team against the other (home and away) it turns into an 18 week season. All of which is shorter than we have now, or if you play teams only once each.
You could ‘solve the bye’ problem by having inter-conference games, but there is an issue around “who plays who and where” that returns the ‘draw’ back to a ‘fixture’. 
D- Conferencing or separating teams based on fixed criteria, such as geography, fan base, commercial power etc [5].
We believe that any conferencing of teams into tiered divisions should be avoided. Perceptions of premier conferences, either via a promotion/relegation system, or via pooling the powerhouse clubs together.
Those type of structures will only reinforce and entrench imbalance.
Equally, conferencing on geographical location should also be avoided in a competition so heavily Melbourne-focused as the AFL is. Splitting teams this way will create one Melbourne-centred conference and the others regionally-based, and travel requirements could be unbalanced.

OUR PROPOSAL -SUMMARY: 
The most effective way to create a balanced draw is to employ a 3 conference system of 6 teams in each group.

The conferences are selected by the previous seasons complete ladder, with each conference seeded for balance. Each conference has two of the top 6 teams, two of the middle 6 teams, and two of the bottom 6 teams in it.

Conference teams play within each conference home-and-away (10 games), and also 12 inter-conference games, for a 22 weeks season. The only ‘fixing’ of the draw is to ensure each Melbourne team travels approx 5 times per year and retains 11 home games per year.

Finalists are selected from the conference winners, and they are guaranteed finals places and double chance slots. The remainder of the 8 is filled by awarding ‘wild cards’ to the next best 5 teams, and continues as per the current final eight system.

At the end of each season, the conferences are combined into a complete 18 team ladder (as per now) and this ladder is used for
     - re-drawing the following seasons conferences and,
     - allocating the draft order for the next draft.
It has no influence on who is allocated to finals positions.


[1]  We use the term ‘fixture’ to denote that the games are fixed by the AFL into its uneven, unbalanced state.
[2]  The term ‘draw’ is used as a more equitable result where teams play each other on an equal footing.
[3]  An 18 team competition means each team plays 17 other teams. Play them once = 17 game season. Play home and away = a 34 game season. 
[4]  Probably the most popular format being discussed, but it retains a few holes that most commentators haven't thought through (our opinion).
[5]  Again, another popular proposal, possibly mooted by those thinking of mirroring US conference systems. The key problem with this is the very Melbourne-centric nature of the AFL... an inherent imbalance in travel requirements that is not known in US systems.



   2. THE DETAIL

In our proposal, we suggest splitting the 18 teams into 3 conferences, based on ladder position where teams are seeded. Each conference contains 2 of the previous seasons top teams, two mid-table teams and two of the bottom teams. It also includes season-on-season re-drawing of the conferences to ensure ongoing balance.
And for arguments sake, why not name the conferences after three of footballs most famous names from the three key footballing states; Barassi, Farmer and Robran conferences.


SPLIT INTO CONFERENCES
In separating the teams into conferences, the previous seasons ladder is used, and a seeding system is employed to balance the split. The seeding we have chosen [1] will see the following teams in each group.

     Barassi Conference:      1st, 6th, 7th, 12th, 13th and 18th.

     Farmer Conference:       2nd, 5th, 8th, 11th, 14th and 17th.

     Robran Conference:       3rd, 4th, 9th, 10th, 15th and 16th.

This seeding selection provides each group with an evenly selected number - two from the top 6 teams, two from the mid-table 6 teams and two from the bottom 6 teams. Also, adding each of the above ladder positions produces the same number: 57.


So for the 2013 season, the ladder from the 2012 season would be used to split the teams into the conferences as at right.

In some instances, conferences will be made up of teams spread all over Australia, while others will be more localized.
To overcome any sense of bias in travel, the inter-conference games can be arranged so that those teams that don’t travel so much in their conference will play more 'on the road' games to compensate and balance out travel commitments.

As before, the only ‘fixing’ of this draw is to ensure teams play 11 times at home, and at least 5 times interstate.


In the instance above, the following is immediately detectable:

     - there are two interstate teams in one conference, and three in the other two conferences.
     - the two Adelaide teams are drawn together, but the two West Australian teams are not together.
     - similarly, the two Sydney and Queensland teams are also not drawn together, and
     - the scope for Melbourne 'blockbusters' is disrupted as Collingwood won't have 2 games against Essendon, Richmond, Geelong and Hawthorn.
     - There is still 2 matches for Collingwood against Carlton and St Kilda, and we propose that scheduling of the draw remain flexible enough that the Anzac Day game can still be reserved for Collingwood and Essendon.

This will alarm some AFL folk, as it has taken away some of the current mandated blockbusters, but as before, we think the AFL is now in a position to no longer lean on those crutches, and to let the scheduling process become just marginally more free.


[1]  An alternate split would be to have the groups split into every third team, i.e. 1, 4, 7, 10, 13, 16 and 2, 5, 8, 11, 14, 15, 17 and 3, 6, 9, 12, 15, 18.
Further, you could ballot the teams into groups of 6 (top six, mid six and bottom six) and then draw two from each pot into conferences. This would be eminently televisable, image Bruce welcoming viewers to the "Melbourne Bitter Conference Draw, live from Jupiters Casino on the Gold Coast" 


CONFERENCES AND THE DRAW
Teams are drawn to play each other in their own conference twice, home and away. This is 10 rounds of the year (6 teams = 1 team plays 5 opponents, home and away = 10 games).
They also play other teams in other conferences once each (each team plays 12 other teams once each = 12 games).

This results in a 22 game per team season, just as we had in 2012, and it can be stretched out to 23 rounds (with byes) if the AFL desire it.

A simple 22 round draw is as per right (click to expand) based on the above split for 2013.
It has been worked to include the 10 games played inside conferences (with the yellow bar along side), and the 12 inter-conference games (with the green along side).

It is an example only so please don't worry if any team has too many or not enough home games, travel etc. That issue can be solved by folk who have more time to work through that issue (and also get paid for it).

The key point if you work through it is that each team plays inside its conference twice and outside once.
Further, if AFL HQ wish to move about the games to fit particular days of the year, that can be done as well.
Given the Anzac Day 'tradition', and the desire of clubs to play a particular team on the opening round (...etc, etc), it is easily adaptable for those games to get shuffled about.

Equally, we have book-ended the inter-conference games in the middle of the season, with intra-conference games either side for reasons of simplicity. Again, it is at the AFL's discretion to determine when and where games are played.


END OF REGULAR SEASON
It is worth remembering that the conferences above will change year on year, so the mix of teams in conferences, as well as opportunity for blockbusters will alter year on year.
So while some teams wont get the big matches some years, the wheel will turn to produce those chances in later seasons.

We will cover how the Finalists are selected in our next post.
A later post will then cover how the Draft and these conferences work together, and also the mechanics to the season-on-season changes to the conferences.


Under our proposed system, the power of the scheduling and allocation of games is partly held by the AFL but are mostly divested, as the ladder from the previous season sets the conferences and the number of games to be played.

If the AFL and the clubs themselves are thinking of what is best for football, they will also need to divest themselves of self-interest as well to fully embrace this concept.
For the good of the game.



   3. EFFECT ON THE FINALS                      

We propose no changes to the structure of the finals system, nor its execution. The current finals 8 system is retained. Only the selection criteria for teams is modified, because simply an 8 team system is not wholly divisible by 3. Therefore a new process is employed.

Under our system, the top team from each of the 3 conferences goes through to the finals automatically, and take three of the four double chance spots.
The conference winner with the better record is slotted into the first qualifying final game, and the lesser two performed conference winners play each other in the second qualifying final.

To complete the remaining 5 finals slots, ‘wild cards' are used. These are allocated too the best performed of the 9 teams that finish in 2nd, 3rd and 4th in each conference.
And the best of those 9 is rewarded with the last double chance slot, and is scheduled to play the first qualifying final (against the best of the conference winners).

The next 4 best teams then fill the two elimination finals match slots, based on season merit.

This completes the allocation of teams into the finals. There is a reasonable chance that the best of the 2nd to 4th finished teams could be scheduled to play their own conference winners in the first week of the finals. We don’t recommend altering that, as by merit both teams deserve those places.

From the allocation of teams onward, the finals system does not change.


SCENARIOS
We suggest that the three conference winners alone gain automatic selection as reward for gaining that title. It helps give those conference championships value beyond a simple award.

Our modelling has also shown occasions where a 2nd place team could have a better record than a conference winner. In that case, the second placed team performed better, but could still earn a double chance finals spot (if they have the 4th best overall record), or at least an elimination finals spot.

In allocating wild card spots, we suggest taking the results from the best 9 non-conference winners... that is, positions 2, 3 and 4 of each group. We have chosen this as our selection criteria, as in our testing and modeling of this system, as there is a chance a deserving team misses out.
We have seen occasions in our modelling where a team finishing 4th in a conference misses out on a wild card slot even though it has a better record than another team that gains a wild card.


To help in the understanding of the finals allocations, an example of one of our data modeling results is attached.

Working through a season, we could devise the three conference tables as at right.

For those wondering, the match scores were determined by the following excel calculation:
=RANDBETWEEN(x,y)
where
x = a teams highest score from 2012, less 10%y = a teams lowest score from 2012, plus 10%

This calculation was done to ensure there is weighting to the teams to reflect their strength, but also randomise the results to a degree, creating occasional upsets.

For example: In a game between Hawthorn and Sydney, the scores were determined thus:
   HAWTHORN: a random number between 162 and 76
   SYDNEY: a random number between 124 and 77.
So in that case, there is a good overlap of scoring windows, creating win chances for both teams.

The conferences are therefore tabled separately, and conference winners and finals spots determined. In this instance, Hawthorn, Adelaide and Collingwood are conference winners and gain automatic finals spots (above the solid line).

The other nine teams listed below the solid line and above the dotted line are the teams in the running for the wild cards:
- Richmond, Geelong and Fremantle.
- North Melbourne, West Coast and Essendon.
- St Kilda, Sydney and Carlton.
Based on merit, Collingwood with the best record play the best performed non-conference winning team (Richmond) in the First Qualifying Final.
The other two conference winners, Adelaide and Hawthorn, then enter the Second Qualifying Final.

Elimination Finals are filled by the next 4 teams with the best records that finished 2nd to 4th in each conference.
In this modelled example, the Elimination Finals are between St Kilda and Sydney, and North Melbourne and Geelong.

The Finals Series then play out as per the normal system.


   4. EFFECTS ON 'OUT OF SEASON' EVENTS

Before we detail how our system of conferences influences the draft, the pre-season competition and the following seasons conferences, a point on three conference models.

As we have seen in the previous post on the finals, a three conference system is difficult to fit into any normal linear system, or an even numbered team system.

Accordingly, we consider the best approach to determining the (lineal) draft order, and the (lineal) re-ordering of conferences that we described in the original post, is to merge the three conferences back into a full 18 team ladder, based on merit.

So as a preamble to the below, the merging of the three conferences at right (from our worked example listed in the previous post) into a single, merit-based ladder, produces the table as below.




From this table, the draft is easily set and the next seasons conferences can be re-drawn.


EFFECT ON THE NATIONAL DRAFT
To arrange the draft order, the AFL need only revert to the re-combined ladder (a single 18 team ladder as above).
Under this modelled example, GWS get first pick, followed by the Gold Coast, Melbourne etc.

We propose no further changes to the draft system.


EFFECT ON THE PRE-SEASON COMPETITON
The model we propose fits inside a regular 22 round season. We see no need to alter the format of the pre-season competition as it stands today.


EFFECT ON FUTURE CONFERENCE SELECTION
As per our opening statement (Principles), the only path to a fair conference system, is by altering the make-up of the conferences season-on-season.

Only by re-drawing the conferences each season can you remove any perceptions of imbalance, or powerful / weak conferences. With the competition geographically compromised by 10 teams (more than half of the whole) in the one market, Victoria, re-configuring the conferences annually distributes the Victorian teams, as well as varying each teams travel requirements.

As per the second post (Detail), the conference selection method we have elected to use is a seeded system to attempt to balance them. As per that system, the following teams fall in each group.

     Barassi Conference:      1st, 6th, 7th, 12th, 13th and 18th.

     Farmer Conference:       2nd, 5th, 8th, 11th, 14th and 17th.

     Robran Conference:       3rd, 4th, 9th, 10th, 15th and 16th.

Which from the worked example, in the combined ladder above, yields the 2014 conferences as below.


Again, this seeding selection provides each group with two teams from the top 6 teams, two from the mid-table 6 teams and two from the bottom 6 teams.

In this season, there are
   - two conferences that have 2 interstate teams in them, and one that has four,
   - the two Adelaide teams and the two Sydney teams are now not pooled together,
   - the Queensland teams and the West Australians are pooled together,
   - the Melbourne 'blockbusters' look to be based on Collingwood having 2 games against Carlton and Geelong, though Hawthorn v Richmond should also draw a crowd.
   - The one Collingwood v Essendon game can still be reserved for Anzac Day.


The seasons draw then falls out of this with teams playing 10 games inside each conference and 12 outside the conference, as we described previously.

As before, our proposal brings transparency to the draw, and removes the gerrymanders of the current fixture. It still leaves open the AFL to schedule blockbusters into appropriate dates, as well as provide some randomness to each seasons draw.



   5. CLOSING COMMENT

It is interesting that the AFL was at one stage considering options such as conferences when introducing the new development teams (as per Fox Sports: AFL considering conference system to accommodate an 18-team competition) but never developed the idea further. Perhaps the most telling part in that stalling of development was from AFL chief operating officer Gillon McLachlan, who said
"We understand that our football community likes tradition, they like the way it's working at the moment"
We at the FootyMaths Institute also like tradition too... the football tradition of all teams getting to play all others home and away, and the Australian tradition of equality and a 'fair go' for all, just to name two examples.

We also know that the above footballing tradition cannot be achieved neatly with an 18 team competition, but our proposed model is a good start to meeting that, while giving all teams a 'fair go' at representation in the draw.

We also credit the footballing public with intelligence to be able to wrap their heads around a conference model, and in particular the one as we have proposed. One that creates and re-creates itself over seasons, ensuring that each conference is balanced, and the draw is not rigged by 'faceless' AFL men, but determined by the finished order from the previous year.
Open.
Honest.
Transparent.



Other football fans and personalities have also proffered options on how to redress the draw. We have also studied those concepts, and find most are lacking.
Lets put them into our classroom for study and grading.


The key proponents to date have been.

1. Matthew Clarke and Michaelangelo Rucci
The rather misleading headline "The Advertiser's AFL reporter Michelangelo Rucci calls for the AFL to make the draw equal for all" suggests it is Rucci making the call for the draw to be 'equal for all', but if you read the article it is really Matthew Clarke with the concept.

The Clarke concept calls for geographical conference splits (Northern, Western and Victorian), and also sets all the most heavily supported Melbourne clubs into one conference.
This concept is just so wrong, so unfair, on so many levels. Most of which we have detailed above.
...and oh yes, it was published by a News Ltd affiliate. Says more than enough.

  F  



2. Mick Malthouse
Another News Limited affiliate provides us with the next idea, this time from a doyen of the game; "The AFL premiership should be split into three divisions, writes Collingwood coach Mick Malthouse".

Mick, who at the time was in his last year at Collingwood, suggested a 'three division AFL season', according to the copy editor who devised the headline. If you think of divisions in terms of the old VFA system (or English FA, etc), then Micks proposal is not that at all. Just more misleading editing at News Limited.

At its core of Micks message was a call to 'stop the floggings' that were happening. This would be achieved by top teams playing bottom teams only once per year. This is effectively what the AFL are doing in 2013, along with the maintenance of the gerrymandered fixture.

What Mick, and the AFL, have failed to recognise is that other influences are fueling blow-outs... such as we described in relation to club spending. The AFL draw Mick Malthouse proposal does nothing to address that, and will continue (and possibly entrench) the divide between haves and have nots.

  E  



3. Ed Wyatt
The Back Page Lead and Ed Wyatt discussed concepts for conferences in 2010 ("Can AFL conferences really work?"). To Ed's credit, he throws up some formats, and also throws them out too, but ultimately there is no firm proposal tabled.

Ed also is one of the few that acknowledges that running a 'Big 6' conference wont be acceptable to the other clubs, and that the geographic skew that the AFL finds itself in preclude using location as a determinant that US leagues do. Ed also tips cold water on a two conference system, without any more of an argument than "nine is a strange, uneven number to have in your conference."
Fair enough!

  C  



4. Jason Feldman via Mark Stevens
Mark Stevens in the News Limited press writes of another approach, this time proposed by Jason Feldman... "a keen student of the game" ("Forget the blockbusters AFL, just fix the fixture").

The Feldman approach goes closest to ours, suggesting three conferences, a reduced focus on blockbusters and altering the conferences.
But the detail shows that the conferences are altered only slightly, and that there remains gerrymandered home state derbies almost every year, and the Melbourne teams move about the most. Nonetheless, there is a great degree of thought to balance in this system.

There is also a possible error in the draft allocation where they have published "Draft picks 1-8 to be based on records of non-finalists". With an 8 team finals system there should be 10 teams as 'non-finalists'.
Also, this is difficult to grasp, and not explained or developed further: "Draft picks 9-16 based on positions after finals."

Oh, yes... that is all that Jason Feldman is credited as by News Limited... "a keen student of the game". A quick google search may show him also to be a Match day Statistician for Champion Data, but we haven't confirmed that. If so, then 'way to disclose your sources, News Ltd'!

Overall, a good system.

  B  




There are other proposals about the interwebs (for example, The Roar website has 5 alone), but in most cases that we have studied, they revert to type of having conferences
   - arbitrarily determined, usually to the benefit of the big clubs, or
   - determined by geography.

Most also consider taking the top two teams from the conferences into the finals direct, even though there may be a case where a second placed team in one conference has a poorer record than a third placed team in another conference.
Most don't even consider how the top 6 teams are allocated the 4 'double chance slots' in the finals, nor do they address the draft.


The primary aspect that all other proposals gloss over is the notion of being 'fair' and giving every team a fair go. As per our 'principle' statement, we consider 'fair' to be more than just adjusting the playing schedule for an even number of games (home and away).

Ultimately, we see 'fair' also in terms of ‘opportunity’. The opportunity to meet popular clubs home and away each week, as well as the opportunity for free to air television coverage to boost brand awareness, club finances and supporter base, revenue, etc.
Our conference model delivers a fair go, a draw that is open, honest and transparent without the fingerprints of the gerrymandered 'fixture' we have today.

That is the end game the AFL should be playing, because working toward that creates the overall evenness on match day, it brings more people through the gates, more broadcast outlets tuned to your product.
Not positioning the AFL to that end will see further disparity, leading to disillusionment and ultimately a diminished league.

3 comments:

  1. Would a lottery draft work in the AFL, I ask because then free agent compensation could be 3 "balls" etc?

    ReplyDelete
  2. Just had a read, I do appreciate the work put in (and also how long ago it was written), but are you aware that this system is almost the exact same as the current system, only the double ups are from a set format rather than the current theoretical formula? The way your finals transfers work would give the same result every time as just stacking every team up on one ladder at the end of the year UNLESS one of the conference 2nds has a better overall record than a conference 1st, where an aggregate ladder would favour the better record.

    ReplyDelete
  3. Thank Harry.
    Yes it was written a while ago, back early 2013. If I recall correctly, I think the handicap system was loosely introduced the same season. And yes it is basically the same in that the double ups are set, but the distribution is different.
    The above has a distribution locked in by previous ladder position, while the AFL current rule is more flexible and allows for 2x derbies etc 'as a given'. The above make no guarantees.
    The current AFL system has more flexibility, as per the Stevo tweet in this link (http://footymaths.blogspot.com.au/2014/11/fixture-correct-weight.html).

    The other key difference is the AFL system sets the 2x games within blocks of 6. SO the top 6 teams get a harder schedule, and you could say are handicapped, and the bottom 6 get a softer schedule. They get theoretical boost.
    The post has each team playing 2x games against teams spread thru the table, so it has balance.
    It also has transparency because immediately after the Grand Final, you know who your team plays twice next year, the ladder tells you.

    Also, since writing, I can see this system would easily work without splitting into conferences, though I feel splitting helps understand your 2x matchups as well as creates an extra challenge to finish top.

    Finally, yes in most occasions the top 8 of a combined ladder would be the finalists regardless of conferences. But there could be occasions where a conference could be collectively poor and only 1 or 2 go to the finals based on wins, so I've allowed for the option of giving finals spots to better teams in other conferences. But yes most times its the same as a top 8.Its a minimal change to increase balance.

    Thanks for reading!

    ReplyDelete