WoSHBODS tipped 11 of 12 winners this week and also recorded a Margin MAE of 11.2 points per game and a Totals MAE of 13.8 points per game in what can only be described as a watershed round for the forecaster.

That took its season-long figures to 40 from 61 (66%) for accuracy, 22.0 for Margin MAE, and 18.7 for Totals MAE.

TEAM STATISTICS

We’ve enough data now to make it wortwhile reviewing how WoSHBODS is performing on a team-by-team basis.

Firstly, in the table at right, we look at the MAEs by team for WoSHBODS, and find that Adelaide’s, West Coast’s, and Port Adelaide’s margins have so far been easiest to forecast, and Melbourne’s comfortably the hardest.

Interestingly, the only non-Victorian teams under the average of 22 points per game are Richmond (17.5) and North Melbourne (21.8). That maybe hints at an issue with the way Venue Performance Values are being calculated for Victorian teams, but it’s far too early for us to say anything definitive about that.

At this stage, there doesn’t seem to be any clear relationship between teams’ ladder positions and the precision with which WoSHBODS is forecasting their game margins.

Next we’ll look at the log probability scores associated with each team, based on the probability estimates that WoSHBODS has attached to them in each game.

Here we find that WoSHBODS has done best with Port Adelaide, North Melbourne, Adelaide, and Brisbane, and worst with Geelong, Melbourne, and Carlton (which are also the three teams it’s done worst with on MAE).

On this metric, WoSHBODS seems to perform adequately with Victorian teams, with six of the nine teams whose LPS is below the all-team average coming from that State.

Also, there are only six teams that have generated negative log probability score for WoSHBODS.

WAGERS

All three Funds were profitable this week, the Head-to-Head Fund jumping just over 12c, the Line Fund just under 11c, and the Overs/Unders Fund just under 15c. In aggregate, the Combined Portfolio rose by 11.5c to be up now by 22.6c on the season, that on a +19% ROI and a 1.2 turn.

It would be unreasonable to expect another round as good as that one this season.

TEAM DASHBOARD

The metric rankings currently most highly correlated with the competition ladder (which I order, now that we have teams having played different numbers of games, based firstly on competition points per game played, and then on percentage) are:

• Points Conceded per Game: +0.87

• Scoring Shots Conceded per Game: +0.87

• % of Quarters Won: +0.87

• MoS Win Production Function: +0.87

• Goals Conceded per Game: +0.85

• Q2 Performances: +0.83

• Points Scored per Game: +0.8

• Scoring Shots Generated per Game: +0.76

The metrics least correlated are:

• Own Scoring Shot Conversion: +0.44

• Opponent Scoring Shot Conversion: +0.45

GWS’ numbers in particular look odd for a team sitting 16th on the ladder.

And, lastly, below is the full extended version of the Team Dashboard.

It reveals that:

• there is only one undefeated and one winless teams

• Fremantle are yet to win or draw a Q1 and Collingwood a Q3

• Fremantle have, however, outscored their opponents by almost 4:1 in Q4s

• Richmond have outscored their opponents by over 4.5:1 in Q1s

• North Melbounre have outscored their opponents by over 3.7:1 in Q1s, Q2s, and Q3s

• St Kilda have outscored their opponents by 3:1 in Q1s but been outscored by about 1.5:1 in Q4s

• Collingwood have been outscored by their opponents by 10:1 in Q3s

• Sydney have converted at 56% and allowed their opponents to convert at only 42%, but have generated only 9 Scoring Shots per game while allowing their opponents to generate 13.6.

• Richmond have allowed opponents to convert at only 34%, but have themselves converted at only 47%.

• Fremantle have scored only 7% of their points in Q1s, GWS 8% in Q2s, and Collingwood 8% in Q3s

• Essendon have conceded only 9% of the total they’ve conceded and St Kilda only 10% in Q1s. Fremantle have conceded only 13% in Q4s.

• St Kilda have conceded 40% of all the points they’ve conceded in Q4s, Frematle 41% in Q1s, and Essendon and Richmond 38% in Q2s (and also in Q4s for Essendon).