Simulating the Finalists for 2015 After Round 15

According to the latest round of simulations using updated ChiPS Ratings seven teams' Finals' chances were materially affected by last weekend's results:

  • Geelong (Down 25%; 47% to 22%)
  • Adelaide (Down 17%;  50% to 33%)
  • Collingwood (Down 7%; 80% to 73%)
  • The Kangaroos (Up 16%, 43% to 59%)
  • GWS (Up 14%; 19% to 33%)
  • The Western Bulldogs (Up 13%; 58% to 71%)
  • Richmond (Up 6%; 83% to 89%)

All of the other teams with genuine Finals aspirations emerged from the weekend with much the same prospects as they entered, Fremantle, West Coast, Hawthorn and Sydney all remaining near-certainties for a place in September, and Port Adelaide still about 4/1 chances. Essendon, St Kilda, and even Melbourne, also continue to make appearances in the Final 8, but only in the more exotic simulation results.

The simulation inputs used to produce these and the following estimates of team fates are not all that different from what they were last week. Their details, in the form of probabilities, appear in the table at right. (I actually use margin predictions in the simulations; these are the probabilistic equivalents of those margins.)

Hawthorn still has the highest expected number of wins across the remainder of the season (5.7 games), West Coast second-highest (5.2), and the Roos third-highest (4.7). Carlton has fewest (1.3 games).

The manhattan chart below depicts the profiles of every team's simulated ladder position across the 100,000 replications. Many teams still have reasonable chances of finishing in any of at least three or four ladder positions, this uncertainty a reflection of the closeness of the competition, especially amongst the second tier of eight or so teams after the big 4 (more on which in a minute), the distance the competition still has to run, and the inherent uncertainty of the sport of Australian Rules.

Viewing these same simulation outputs instead as a heatmap, the competitive hierarchy of the competition can be seen very clearly with four teams vying (mostly) for the top 4 spots, eight more competing for positions 5 through 12, three quibbling over positions 13 through 15, and three more eyeing a place in positions 16 to 18.

We can further explore the competitive landscape for that second tier of eight teams by interrogating the simulation results to ask, specifically, how does Team A's chance of making the Finals vary dependent on whether or not Team B makes the Finals?

The results of that interrogation appear in the table below.

The first row, for example, relates to Adelaide's chances of making the Finals and tells us that:

  • Adelaide made the Finals in 33% of all replications (the number on the far left)
  • Adelaide made the Finals only 29% of the time when Collingwood did, but 45% of the time when Collingwood didn't. 
  • Similarly, Adelaide made the Finals only 20% of the time when Geelong did, but 37% of the time when Geelong didn't. 

The remaining columns for Adelaide can be read in the same way and the differences between Adelaide's "unconditional" probability of making the 8 (ie 33%) and the "conditional" probability of making the 8 when some other team does or doesn't make the Finals used as a measure of how dependent Adelaide's entry into the Finals is on that other team's ultimate fate.

Reviewing the table in that light, we can say that:

  • Adelaide's chances of playing in the Finals are lowest if Port Adelaide makes the 8 and highest if Richmond misses
  • Collingwood's chances are lowest if Geelong make the Finals, highest if Richmond doesn't or Port Adelaide doesn't
  • Geelong's chances are lowest if GWS make the Finals, or if Adelaide does, and highest if Collingwood miss the 8
  • GWS's chances are lowest if Geelong make the Finals, or if Port Adelaide does, and highest if Richmond miss the 8, Collingwood miss the 8, the Kangaroos miss the 8, or the Western Bulldogs miss the 8
  • The Kangaroos' chances are lowest if Port Adelaide make the Finals, or if Adelaide does, or if GWS does, and highest if Richmond miss the 8
  • Port Adelaide's chances are lowest if Adelaide make the Finals, or if GWS does, and highest if the Western Bulldogs miss the 8
  • Richmond's chances are lowest if Adelaide make the Finalsand highest if Collingwood miss the 8, or of the Kangaroos do
  • The Western Bulldogs' chances are lowest if Port Adelaide make the Finalsand highest if Collingwood miss the 8

Of all the teams, Richmond's fate is least dependent on other teams' outcomes, the absolute difference between their highest and lowest conditional probabilities being just 13% points. In absolute terms, Adelaide's fate is most dependent on other teams, the difference between their highest and lowest conditional probabilities coming in at 38% points. In relative terms, however, Geelong's and Port Adelaide's fates are also highly dependent on the results for other teams, their probability of making the Finals altering by a factor of over 3 depending on whether some specific team does or doesn't make the Finals.


Once again we'll use the latest TAB prices as an indicator of the extent to which our simulations are consistent or otherwise with prevailing expert opinion in the relevant wagering markets.

This week the comparison with those prices suggests that we rate the Hawks' and West Coast's chances for the Minor Premiership significantly higher than does the TAB Bookmaker, and that we also still rate the Dogs' chances for a spot in the Finals far higher than does the TAB Bookmaker.

Note that, as I did last week, I have shaded green those TAB prices that represent an edge of 5% or more, and shaded grey those TAB prices that represent a smaller, but positive (though not one I'd seek to exploit) wagering expectation.

Also note that I'm still providing estimated probabilities for the Spoon market despite the fact that the TAB is not fielding in this market at this stage. It interesting to note that the Dons have non-zero estimated probabilities for a spot in the 8 and for the Spoon.


The most common pair of teams finishing in the first two places in the simulations were Hawthorn and West Coast, they appearing in that exact order is just over 15% of replications and in the reverse order in just over 14% more.

West Coast and Fremantle was another roughly equally-common pairing, this cropping up about 14.5% of the time in this order and another 9% of the time in the reverse ordering.

Fremantle / Hawthorn was also prominent in the Top 2s, and these two teams fill the first two spots in the most common Top 4 of Fremantle / Hawthorn / West Coast / Sydney. This quartet, in that order, appeared in just over 8% of replications, which is fractionally more often than we saw West Coast / Fremantle / Hawthorn / Sydney in that order.

It's still the case that all 10 of the most-common Top 4 finishes include some ordering of those same four teams.

Finally, let's review the most-common Top 8s from the simulations, with the same caution of last week that none of them occurred in more than about 1 simulation in 800, so none of them might reasonably be called "likely" in absolute terms.