Friday, October 31, 2014

The Fairness of the AFL Fixture

In yesterday’s post I rated the difficulty of each AFL team’s draw in 2015. Each year there is a fair bit of variation in the difficulty of the draw between teams; for example, in 2015 I consider the club with the easiest draw, West Coast, to have a goal a game advantage over the club with the hardest draw, Port Adelaide.

Over at the FootyMaths Institute it has been proposed that a fairer 2015 fixture would have each team play opponents of roughly the same level throughout the year. To achieve this it has been suggested splitting the teams into three conferences of roughly equal strength as follows, with each team playing the teams in their own conference twice and every other team once.

Conference 1: Teams ranked 1, 6, 7, 12, 13, 18.
Conference 2: 2, 5, 8, 11, 14, 17.
Conference 3: 3, 4, 9, 10, 15, 16.

That seems about as fair as you can get to me. (Ladder positions can be misleading of course, but I doubt the AFL is going to do a fixture based on a Power Rankings system!) A comment I made was that the conferences need only be notional; that is, teams could be split into conferences for the purposes of the fixture, but the ladder is calculated according to the same method as it always was.

Yesterday I wondered how far the actual AFL fixture deviated from this ‘fair’ system. To compare the two, let’s see how each team’s draw would be rated under the FootyMaths system. I am going to assume that the FootyMaths fixture would result in no net home ground advantage to any team over the course of the year. This probably will not strictly hold because of things like Geelong playing in Geelong against Melbourne clubs, but unless you get savaged like St. Kilda did for 2015, net home ground advantage is generally the least important component of a club’s fixture.

Here are the results:

Overall
Effect of not playing own team
Effect of teams played twice
Sydney
78.0
33.8
44.2
Hawthorn
68.0
34.6
33.4
Port Adelaide
41.8
21.5
20.3
North Melbourne
28.8
9.2
19.6
Adelaide
24.2
16.7
7.5
Fremantle
21.9
15.6
6.3
Essendon
18.2
3.9
14.3
West Coast
17.5
13.4
4.1
Richmond
11.8
6.5
5.3
Geelong
4.8
7.1
-2.2
Carlton
1.3
-4.6
5.8
Gold Coast
-13.7
-12.0
-1.6
Collingwood
-26.5
-12.7
-13.8
Western Bulldogs
-42.7
-20.8
-21.9
GWS
-51.4
-21.0
-30.3
Brisbane
-54.2
-22.4
-31.7
Melbourne
-57.0
-27.9
-29.1
St. Kilda
-70.9
-40.7
-30.3

What the - ? This seems to be more uneven than the current AFL fixture! Note that the easiness of each’s team fixture is essentially determined by their ranking. Lower-ranked teams are now hit by a double whammy: not only are they the only team that do not get to play themselves when each team plays each other once, they are also the only team in their ‘conference’ that do not get to play themselves twice either.
We seem to have hit on something crucial here about what constitutes a ‘fair’ fixture. On one hand you could argue that the AFL should compensate for lower-ranked teams not being able to play themselves and give them easier teams to play to make up for it (although they seem to have overcompensated on this point). On the other hand you could argue that a ‘fair’ system should abstract from this effect.
Note here that a ‘fair’ fixture could result in teams having draws of varying difficulty. Perhaps Sydney and Hawthorn might have the easiest draws under the FootyMaths system because they do not have to play themselves, but this does not necessarily mean that the fixture is not ‘fair’.
What happens if we take out the effect of not having to play your own team? (I have also taken out the net home ground advantage from the AFL fixture as well.)
FMI fixture
AFL fixture
Gold Coast
10.4
-8.4
North Melbourne
10.4
-19.7
Essendon
10.4
-0.8
Carlton
10.4
19.7
Sydney
10.4
-63.8
St. Kilda
10.4
93.2
Hawthorn
-1.1
-96.2
Port Adelaide
-1.1
-101.4
Collingwood
-1.1
27.7
Richmond
-1.1
-17.9
Western Bulldogs
-1.1
77.0
Melbourne
-1.1
107.5
GWS
-9.3
72.5
Fremantle
-9.3
-38.1
Geelong
-9.3
-88.7
Adelaide
-9.3
-24.2
West Coast
-9.3
27.8
Brisbane
-9.3
33.9
Standard deviation
8.3
63.6

Now that seems better. Basically the difference between the hardest ‘conference’ and the easiest one is less than a point a game. For the AFL fixture the difference between the hardest draw and the easiest one is about two goals a game.
Personally I like the FootyMaths proposal; it seems to me about as fair as you can get with each team playing 22 matches. It might be even better than having a 17 round season.

Thursday, October 30, 2014

Who Has The Easiest AFL Draw in 2015?

The 2015 AFL fixture was released today, which means it is time to do my annual assessment of the difficulty of each team’s draw. This is done by summing up, over every game a club plays, the easiness of their opponent as based on my AFL Power Rankings at the end of this season, while adjusting for home ground advantage.

The rating for each team’s draw can be broken down into three components:

-          The effect of not playing your own team: For example, Hawthorn’s draw is made easier by the simple fact that they don’t have to play Hawthorn.
 
-          Net home ground advantage: This is simply the net effect of the adjustments for home ground advantage across the season. If you want to do well on this component don’t play your ‘home’ games interstate! Geelong is rated highly because I give it home ground advantage when playing Melbourne teams in Geelong.
 
-          The effect of which teams your team plays twice: This is the collective strength of the opponents that each teams plays twice, with a higher rating for this component meaning that you have easier opponents in your return bouts.

Which clubs should send the AFL a fruit basket? The easiness of each AFL team’s draw in 2015, ranked from easiest to hardest, is shown in the table below:


Overall
Effect of not playing own team
Net home ground advantage
Effect of teams played twice
West Coast
66.5
13.4
12.0
41.2
Melbourne
57.7
-27.9
6.0
79.6
Western Bulldogs
53.4
-20.8
18.0
56.2
GWS
30.4
-21.0
0.0
51.4
Carlton
16.5
-4.6
6.0
15.1
Adelaide
9.3
16.7
0.0
-7.4
Fremantle
-7.0
15.6
0.0
-22.6
Richmond
-10.9
6.5
-6.0
-11.4
Brisbane
-11.0
-22.4
0.0
11.5
Essendon
-11.0
3.9
-18.0
3.1
Sydney
-14.2
33.8
-18.0
-30.0
Collingwood
-15.6
-12.7
-18.0
15.1
St. Kilda
-18.2
-40.7
-30.0
52.5
North Melbourne
-19.4
9.2
-18.0
-10.6
Gold Coast
-20.5
-12.0
12.0
-20.4
Hawthorn
-27.1
34.6
0.0
-61.6
Geelong
-32.6
7.1
42.0
-81.7
Port Adelaide
-46.5
21.5
12.0
-79.9


Four teams in particular should be happy with their fixtures: West Coast, Melbourne, the Western Bulldogs, and Greater Western Sydney. The AFL has tended in recent years to give teams that finish near the bottom of the ladder an easy run. This season’s last-placed team, St. Kilda, does well in terms of which teams they play twice, but has been savaged in terms of home ground advantage.
At the other end, top-ranking teams Hawthorn, Geelong and Port Adelaide have relatively tough fixtures in terms of which clubs they play twice. The difficulties of Hawthorn’s and Port Adelaide’s draws are tempered by them not having to play themselves, while Geelong as mentioned above do well out of home ground advantage.
One unusual feature of the 2015 draw ratings is that only six teams have a ‘positive’ balance while two-thirds of the competition has a ‘negative’ balance. The ratings are fixed so that they sum to zero, so it suggests that the majority ‘suffer’ in return for a few teams having really good fixtures.
On that point, I would like to compare the ‘fairness’ of the 2015 fixture with the ‘fair fixture’ proposal over at the FootyMaths Institute. I might look at that topic in a future post.