Another pretty good week for WoSHBODS, with a Margin MAE just above the season average at 19.6 points per game, a Totals MAE of 21.4 points per game, a fifth consecutive positive LPS, and a 7.5 from 9 result on accuracy, all of which has led to the table below.

We can break down that overall Margin MAE of 19.1 points per game into team-by-team contributions, which we do in the table at right and which shows that WoSHBODS has been particularly precise in game margins for matches involving the Dogs, Pies, Dees, Blues, Tigers, and Hawks, and particularly imprecise in game margins for matches involving the Giants, Crows, Swans, and Saints.

Looking solely at the four expansion teams, we have Hawthorn and Essendon still well under the all-team average, Port Adelaide at just over the average, and Sydney much higher than the average.

Altogether, nine teams have MAEs below the all-team average, and nine have MAEs above the all-team average.

We can also break down the overall Margin MAE based on the size of the eventual victory, which we do in the table at left and which shows that, still, it has been the larger victories where the forecasts have been least accurate, and the narrower victories where they have been most accurate.

We have MAEs of under 15 points per game in matches won by under 4 goals, and MAEs of over 19 points per game in matches won by 4 goals or more.

What’s also true is that WoSHBODS has tended to be more accurate in games that were won by relatively large margins. It is 26.5 from 44 (60%) in games won by under 3 goals, and 39.5 from 46 (86%) in games won by 3 goals or more.

The Log Probability Scores can also be analysed by team, and the results of doing this are shown in the table at right where we assign one half of the LPS for a game to each of the two teams.

Here we see that WoSHBODS’ probability estimates have resulted in strong LPS returns from the Dees, Swans, Dogs, and Dockers, and weak and negative returns from the Eagles, Blues, and Tigers.

Overall, there have been positive LPS contributions from 15 teams, and negative contributions from only three.

The Ranking on Team Metrics table appears below and continues to show the relative unimportance of Own Scoring Shot Conversion this season, with the Top 8 teams on the ladder now including five of the bottom seven teams on that metric. We also see that Opponent Scoring Shot Conversion still isn’t a great deal more important. What has mattered in the AFLW this season is the ability to generate Scoring Shots, and prevent opponents from doing the same.

There are reasonably high correlations between ladder position and ranking on all other metrics, with only Q1 Performances showing a correlation below +0.78.

(I must admit that I am pretty happy about how well the WoSHBODS Win Production Function has performed in its first year)

Lastly, here is the full Team Dashboard, which shows that:

• Essendon, who sit 10th, have scored more goals than three of the teams that will play Finals

• Gold Coast are in 9th but have an 88 percentage

• Sydney have scored just under half their points in Q1s

• Carlton have scored over 40% of their points in Q1s

• Fremantle have conceded over 60% of their points in Q2s and Q4s

• Brisbane Lions have conceded over 70% of their points in Q1s and Q3s

• Sydney have won only 4 quarters all season (and none in Q3), which is seven fewer than any other team

• Sydney have percentages of 63 in Q1 and 23 to 33 in the other three quarters

• Brisbane Lions and Melbourne have won 29 of their 40 quarters, and North Melbourne have won 26

• Melbourne have more than doubled their opponents’ scores in every quarter

• Brisbane Lions have outscored their opponents in Q2s to such an extent that their percentage is 700

• Adelaide, Brisbane Lions, Geelong, North Melbourne, Melbourne, and Richmond are the only teams to have outscored their opponents, in aggregate, in all four quarters

• GWS, Hawthorn, Port Adelaide, Sydney, and West Coast are the only teams to have been outscored by their opponents, in aggregate, in all four quarters