2022 : Simulating the Final Ladder After Round 19

The latest simulations - both Standard and Heretical - suggest that the teams can now be roughly grouped as follows:

  1. Geelong: assured of a Top 8 finish, assured of a Top 4 finish, and 80-85% chances for the Minor Premiership

  2. Melbourne, Brisbane Lions, and Sydney: assured of a Top 8 finish, 60-70% chances for Top 4, and around 2-6% chances for the Minor Premiership

  3. Fremantle: assured of a Top 8 finish, 50-55% chances for Top 4, and around 2-3% chances for the Minor Premiership

  4. Collingwood: virtually assured of a Top 8 finish, 30-35% chances for Top 4, and longshots for the Minor Premiership

  5. Carlton: 90-95% chance of playing Finals, around 15% chance of a Top 4 finish, and extreme longshots for the Minor Premiership

  6. Western Bulldogs: 55-60% chances of playing Finals and extreme longshots for a Top 4 finish

  7. St Kilda and Richmond: 20-25% chance of playing Finals

  8. Gold Coast, and Port Adelaide: 3-4% chance of playing Finals

  9. Hawthorn, Essendon, Adelaide, GWS, West Coast, and North Melbourne: looking to 2023 and draft picks (with North Melbourne still favourites for the Spoon, and Hawthorn still mathematical chances of playing Finals)

(For details about the methodologies I’ve used, see this earlier blog).

LADDER FINISHES

The ladder projections using the Standard methodology (aka “cold” sims) appear on the left, and those using the Heretical methodology (aka “hot” sims) appear on the right.

Looking first just at the Standard Methodology, we see that the range of Expected Wins now runs from 2.8 to 17, which has grown a little from last week’s range of 3.0 to 16.5. We also see that only 10 teams are expected to finish with a win percentage above 50%.

The biggest declines in Expected Win tally came this week for Melbourne (-0.6), while the biggest increase came for Western Bulldogs (+0.7).

In terms of Top 8 chances, there were only two double-digit changes in estimate, including Western Bulldogs’ +21% points and Richmond’s -17% points.

Comparing the results for the Standard and Heretical Methodologies we find:

  • A slightly larger range of Expected Wins from the Heretical approach (2.6 to 17.1 wins)

  • A similar ordering of the teams in terms of Expected Wins with only Brisbane Lions and Melbourne, and North Melbourne and West Coast swapping places

  • Broadly similar estimates for the changes in teams’ Finals chances

TEAM AND POSITION CONCENTRATION

The HHI figures for the most recent simulation replicates appear below, with those from the Standard methodology on the top, and those from the Heretical methodology on the bottom. (For information about the HHI, also see that previous blog linked to earlier).

STANDARD METHODOLOGY

heretical methodology

Another chunky reduction in average uncertainty this week, with decreases or insignificant increases for every team except Melbourne, who have moved from being one of the teams with the least uncertainty for a large portion of the season to now being a team with close to the most uncertainty.

Both methods suggest that teams are, on average, effectively competing for just over4 different ladder positions now, and both have most teams competing for between 4 and 6-and-a-half positions, the exceptions being Adelaide, West Coast, North Melbourne, GWS, and Geelong.

standard methodology

heretical methodology

These tables also imply a large percentage reduction in uncertainty this week, but also suggest that there are now 5 to 6 teams effectively competing for each position 2 through 10 on the ladder.

They also have the average ladder position with effectively just over 4 teams likely to fill it, and with most ladder positions effectively expecting anywhere between 4 and 6=and-a-half teams to be competing for them. The exceptions amongst the ladder positions are 1st, and 15th through 18th.

WINS AND LADDER POSITION

Here’s an updated view of the distribution of team wins and final ladder position, based on the Standard Methodology

If we, again, sum across all the teams, we can see how many wins are most likely for each ladder position, as shown in the chart below.

The draw between Richmond and Fremantle has now brought the scenarios with half-wins more into play and means that 12 wins is now associated with the team finishing 8th in 31% of replicates, 12.5 wins in 15% of replicates, and 13 wins in 41% of replicates.

For the team finishing 4th, 14.5 wins is now the result in 8% of replicates, 15 wins in 69%, 15.5 wins in 11%, and 16 wins in 7%.

These percentages, of course, vary by team, and in the chart below we look at how likely it is that a given team plays Finals if it records a particular number of wins, and how likely that outcome is. In the first chart we use the results from the Standard Methodology, and in the second chart from the Heretical Methodology.

Under both methodologies, Sydney is the only team with a better-than-even chance of making Finals with just 12 wins, while St Kilda and Gold Coast are both worse than 1-in-3 chances with just 12 wins.

Repeating the analysis for finishing in the Top 4.

Under both methodologies, Geelong is the only team with a better than 3-in-20 chance of finishing 4th when it registers only 14 wins for the season, and St Kilda, Carlton, Collingwood, Fremantle, Melbourne, and Western Bulldogs are all less than 5% chances of finishing 4th if they end the season on 14 wins.

LIKELY IMPORTANCE OF PERCENTAGE

Next, we’ll use the Standard simulations to again investigate the estimated likelihood that any given pair of ladder positions be determined after Round 23 based on percentage.

Last week’s results had a dramatic impact on the likelihood that percentages will determine key ladder positions with there now being a less than 1-in-4 chance that 8th and 9th on the final home and away ladder will be separated only on that basis, and only a 1-in-25 chance that 8th and 10th will be decided that way.

There is also now less than a 4-in-9 chance that 4th and 5th will be decided on percentages, and less than a 1-in-4 chance that this will be how the Minor Premiership is decided.

MOST LIKELY SET OF FINALISTS AND TOP 4S

Across the 50,000 replicates from the Standard Methodology, there were 10,836 unique orderings of the 18 teams, and 44 unique orderings of the Top 8.

In the table at right we show the sets of 8 teams that most often finish as the Finalists, in some order.

Three sets of 8 stand out, as you can see, and together appear in over 85% of the replicates. The first of those comprises all of the teams from the current Top 8, and the second swaps in Richmond (10th on the ladder) for Western Bulldogs (8th on the ladder). The third swaps in St Kilda (9th on the ladder) for Western Bulldogs (8th on the ladder).

What about the Top 4, and this time taking order into account.

There remains quite a lot of uncertainty about this, with even the most common Top 4 - which sees Sydney move from their current 6th spot into 3rd - occurring in only about 1-in-25 replicates. The current Top 4 in their current order is only the tenth-most likely final outcome and it appears in only about 1-in-50 replicates.

Altogether there are 32 Top 4 orderings that appeared in at least 1% of replicates, with Carlton cameoing in one, and Collingwood in nine and never higher than 3rd. These 32 orderings - of which there were 593 in total across the entire simulation - together account for only 54% of all replicates.

Geelong finishes as minor premier in all of the orderings, Melbourne and Sydney appear somewhere in all but nine of them, Brisbane Lions in 22, and Fremantle in 18.



Lastly, let’s look at the Top 2s that came out of the simulations.

The current Top 2 of Geelong and Melbourne dominate, that pair in one order or the other appearing in almost 30% of replicates.

The second through fifth most-common orderings all have Geelong in 1st and swap in one of Brisbane Lions, Sydney, Fremantle, or Collingwood for 2nd. Together, they account for over one-half of all replicates.

There are then four more pairings that occurred in at least 1% of replicates, making it 10 in all for which this was the case. Those 10 represent 93% of all replicates.