Matter of Stats

View Original

2019 : Simulating the Final Ladder After Round 19

The results of the latest 50,000 simulations appear below.

(If you’re curious about the methodology used to create them, you can start here.)

RESULTS

The Western Bulldogs, Port Adelaide, and Hawthorn still have double-digit chances of making the final 8, but there’s a slightly better than 50% chance that the 8 we have now will be, in some order or another, the 8 we have come the end of Round 23, according to the latest simulations.

In the race for the Top 4, five teams have roughly 50% chances or better, and Collingwood has now only a 1 in 6 chance.

Geelong are now estimated to be 70% chances for the Minor Premiership, and the Brisbane Lions 21% chances.

With so few rounds to go, changes in the estimated number of Expected Wins were all fairly modest, and none greater than 0.7 wins in absolute size.

In terms of estimated probability of making the Finals, the largest moves were:

BIGGEST LOSERS

  • Adelaide (down 13% points, making it 23% points in two weeks)

BIGGEST GAINERS

  • Western Bulldogs (up 11% points)

Collingwood this week saw a further 25% point decrease in its Top 4 prospects, making it a 50% decline in just two weeks, while West Coast, Richmond, and GWS all enjoyed 9% point increases in their chances of a Top 4 finish.

DETAILED LADDER FINISH ESTIMATES

The detailed view of each team’s estimated probability of finishing in each of the 18 possible ladder positions appears below. Blank cells represent ladder finishes that did not occur even once in the 50,000 simulations, while cells showing a value of 0 represent estimated probabilities below 0.05%.

The narrowing of the range of possible ladder finishes for most teams is apparent this week, though there remains some fuzziness for the teams currently in 9th to 14th, which we’ll return to in a moment.

WINS REQUIRED FOR TOP 8 AND TOP 4

GWS’ chances of a Top 8 berth should they finish 11 and 11 for the season increased significantly this week to about 70%, which is about 20% points better than the estimated chances for any other team when registering the same final tally of wins and losses.

Most teams will need at least 12 wins to be estimated 75% chances or better of playing Finals.

The estimated probability of the team in 8th spot finishing with just 10 wins rose again this week to 1.6%, or about 1 in 60.

Fourteen wins gives the seven teams now capable of registering that many a very wide range of estimated chances of finishing Top 4. Geelong’s chances are roughly 60%, GWS’ just under 50%, Brisbane Lions’, West Coast’s, Richmond’s and Collingwood’s roughly 20 to 30%, and Essendon’s about 10%

With fifteen wins, Geelong and GWS are virtual certainties for a Top 4 finish, Brisbane Lions and West Coast about 90% chances, Richmond and Collingwood about 80% chances, and Essendon about 65% chances.

The most-likely tally of wins for teams in various final home-and-away ladder positions are now:

  • 1st: 17 wins

  • 2nd: 16 wins

  • 3rd and 4th: 15 wins

  • 5th: 14 wins

  • 6th and 7th: 13 wins

  • 8th: 11 or 12 wins

  • 9th: 11 wins

  • 10th, 11th and 12th: 10 wins

  • 13th and 14th: 9 wins

  • 15th: 8 wins

  • 16th: 7 wins

  • 17th: 6 wins

  • 18th: 3 wins

TOP 2s AND TOP 4s

This week, again, we’ll look at what the simulations are suggesting are the most-likely combinations of teams finishing in key positions.

For Top 2s, we have the situation as shown at right, which still sees a Geelong-Brisbane Lions 1-2 finish as, fairly comfortably, the most likely, it occurring in about 1 simulation in 3. A Geelong-West Coast finish is the next most-likely, occurring in about 1-in-5 simulations, with Brisbane-Geelong and Geelong-Richmond finishes following, each occurring in about 1-in-8 simulations.

No other pairing has an estimated probability above 5%.

Across all 10 of the combinations shown here, five different teams appear at least once: Geelong, Brisbane Lions, West Coast, Richmond, and GWS. Combined, the 10 combinations represent 96% of all of the simulations.

Looking next at Top 4s, we see that even the most-likely combination appears in only 7% of simulations, that being for a Geelong-Brisbane Lions-West Coast-GWS finish. It’s just ahead of a Geelong-Brisbane Lions-West Coast-Richmond finish, which occurred in 6% of simulations.

Three other orderings appeared in 4.5% or more of the simulations.

Across all 10 of the combinations shown here, five different teams appear at least once: Geelong, Brisbane Lions, West Coast, GWS, and Richmond. Combined, the 10 combinations represent only 42% of all of the simulations.

In terms of the Top 8, there were 8,215 different orderings this week across the 50,000 simulations, none of them occurring more than 430 times (0.9%). In other words, even the current most-likely Top 8 ordering is about a 115/1 shot.

If we ignore order, we find 74 unique combinations of teams in the Top 8s across the simulations, the most commonly occurring set of being the eight teams currently filling the Top 8 spots on the ladder, namely Adelaide, Brisbane Lions, Collingwood, Essendon, Geelong, GWS, Richmond, and West Coast, which turned up in just over 1 simulation in 2 (50.6%).

LIKELIHOOD OF PERCENTAGE DETERMINING POSITIONS

This week’s results served to make it slightly less likely that some key positions on the final home-and-away ladder will be determined by percentage because the teams in those positions finish tied on competition points.

In the latest simulations, 4th and 5th are separated by percentage in about 47% of the simulations (down 4% points), and 8th and 9th are separated by percentage in about 42% of the simulations (unchanged). As well, 8th and 10th are separated by percentage in about 17% of the simulations, and 8th from 11th in about 5%.

Also, in just over 1 simulation in 3, 1st and 2nd are separated only by percentage.

TEAM AND POSITION CONCENTRATION

One way of measuring how much uncertainty there is in the competition is to use the Gini measure of concentration commonly used in economics for measuring income inequality to quantify the spread of each team's estimated final ladder positions across the 50,000 simulation replicates, and to quantify the concentration in the probabilities across all the teams vying for any given ladder position.

This week saw the uncertainty about the final ordering of the teams decrease quite significantly again, with the Gini coefficients rising to around 0.75.

At the team level, we saw reductions in uncertainty for all but Carlton and St Kilda, the biggest reductions being for Collingwood and Essendon.

St Kilda, Fremantle, North Melbourne, Western Bulldogs, Port Adelaide, Hawthorn, and Richmond now have the most uncertainty, while Gold Coast, Geelong, Melbourne, Carlton, and the Brisbane Lions have the most certainty about their eventual ladder finishes.

If we adopt a ladder position viewpoint instead, we see that 1st and 18th remain the positions with the narrowest range of likely occupants.

All 18 positions saw increases in certainty, the largest for positions 6th, 7th and 8th. The positions with most uncertainty are 9th through 14th.

Roughly speaking, we have slightly more certainty this year after 19 rounds than we did last year at the same point about the final ordering of the teams at the end of the home-and-away season (the Gini coefficient then was about 0.74).

GAME IMPORTANCE

Here are the assessments of importance for the remaining 36 games of the home-and-away season. (See this blog for details about how these are calculated.)

Adelaide is now involved in all of the top 4, and Port Adelaide in both the 5th and the 6th. The Western Bulldogs are involved in 4 of the Top 10.

To finish we’ll again look at the importance of the remaining games in terms of their estimated likely impact on the Top 4 (using an equivalent methodology to that we used for assessing the likely impact on the Top 8).

Here, we find Richmond are involved in the Top 2 games, and in 4 of the Top 7, while GWS is playing in the 3rd, 4th and 5th most important games.