2020 : Simulating the Finals before Week 1

Most years, by this time in the season I would have already, along with simulating the final home-and-away ladder, have also simulated the Finals. But this year I’ve just felt that there have been too many unknown, unknowns to add into the sources of variability.

Today then we take a first look at a MoSHBODS-based simulation of the nine games of the Finals.

METHODOLOGY

Broadly speaking, the simulations of the Finals will use the same methodology that was used in the simulations of the home-and-away season, including the manner in which home ground advantage is estimated.

In previous years, I know there have been debates about the extent to which home ground advantage applies in the Finals series, or at least in the Grand Final. You could probably make an even stronger case this year that home ground advantage will be further attenuated, but I’ve had reasonable success in previous years by simply assuming that Finals operate much like regular home-and-away games so I’ll be sticking to that methodology for now. If nothing else, it’ll be interesting to assess its success or failure come the end of the season.

View fullsize

Now, to implement the approach, we do need to make some assumptions about where the Semi-finals and Preliminary Finals will be played, contingent on whose playing in them. The assumptions used for the simulations are summarised in the table at right.

It sees every game of the Finals series being played at either the Adelaide Oval or the Gabba, which confers something of an advantage in the simulations to both Port Adelaide and Brisbane Lions, given the methodology for determining venue effects as described above.

It’s conceivable, I suppose, that one or two other venues might be used, but the current quarantine restrictions would seem to limit the possibilities.

There are also some interesting questions to ponder, such as where the First Semi-Final might be played if Geelong lose the First Qualifying Final, but I expect the impact of that assumption on the simulation results will be fairly small.

And, speaking of results …

RESULTS

View fullsize

In the first chart we simply look at the estimated probabilities for each of the 8 Finalists going out in a particular week of the Finals.

We see that the Top 4 teams provide the Grand Finalists in all but a small number of replicates, with the Bulldogs the only team from outside the Top 4 registering a probability above 5% of making the final game of the season.

We also see that Port Adelaide and Brisbane Lions have roughly equal chances of winning the Flag of around 3 in 10, and Richmond and Geelong also have roughly equal chances of around 1 in 5 to 1 in 7.

Drilling down a level on the 50,000 replicates we can analyse the paths to elimination followed by each team.

View fullsize

We see, for example, that Port Adelaide:

• goes in in a Semi-Final to Collingwood in about 5% of replicates and to West Coast in another roughly 5%

• goes out in a Preliminary Final to Western Bulldogs in about 3% of replicates, to St Kilda in about 3% of replicates, to Richmond in just over 10% of replicates, and to Brisbane Lions in a bit over 15% of replicates

• loses in the Grand Final in about 25% of replicates

• wins the Grand Final in just over 30% of replicates

Lastly, let’s look at what the replicates reveal about possible Grand Final matchups.

View fullsize

‘We see that a Port Adelaide v Geelong Grand Final is the most common pairing, accounting for almost 25% of all replicates. Port Adelaide wins about 60-65% of the time.

Next most-common is a Port Adelaide v Brisbane Lions Grand Final, which appears in just over 17% of replicates.

Richmond v Brisbane Lions, and Geelong v Brisbane Lions Grand Finals account for about 13% of replicates each.

Port Adelaide v Richmond accounts for about another 10%, and Richmond v Geelong about 8%.