2016 Round 5 : Bordering on Surgical

Again it's the Overs/Unders Fund that's most active this week, but I've no issue with that given its five from seven performance last weekend. For their parts, the Head-to-Head and Line Funds have chosen to battle on only three fronts and, in the Line Fund's case, with relatively small weaponry.

Those Head-to-Head and Line wagers are all on moderate underdogs, priced in a narrow $2.35 to $2.45 range and receiving between 8.5 and 10.5 points start.

Combined, the 11 wagers represent just under 6% of the initial Overall Portfolio and promise a maximum upside of a bit under 7%. The best possible result in a single game would be a St Kilda win in a game where fewer than 188 points were scored.


The Head-to-Head Tipsters have recorded their third-lowest (and, for that matter, third-highest) overall Disagreement Index this week, that figure of 21% driven higher, as usual, by the joint competition-leading Home Sweet Home.

Broadest disagreement is in evidence in the Melbourne v Richmond game where the Tigers enjoy the favour of only a single Tipster more than do the Dees. There's a little less disagreement, but not a trivial amount, in the St Kilda v GWS clash too, where the Giants have garnered the support of seven of the 11 Tipsters.

The Margin Predictors are at their disagreement nadir for the season, their overall mean absolute deviation (MAD) just 4.2 points per game per Predictor.

It's the Fremantle v Carlton game where the MAD is highest, but only because the Predictors can't agree on by how much the Dockers will win. Forecasts span a 27-point range from 18 to 45 points and have an associated MAD of 7.8 points per Predictor.

Perhaps the most interesting game, however, is the one with the lowest MAD, the Melbourne v Richmond matchup, where we have the Predictors arrayed narrowly either side of a drawn result - Bookie_3, in fact, being unable to choose which side to array itself on.

ENS_Linear_MoSS has the round's largest MAD, which, at 7.5 points per game, is a little over 0.5 points per game higher than the next-largest MAD, ENS_Greedy_MoSS' 6.9 points per game. The smallest MADs of just 2.1 points per game belong to ENS_Linear and ENS_Greedy.

Levels of disagreement are also relatively low amongst the Head-to-Head Probability Predictors this week, their all-Predictor MAD of 3.6% points per game per Predictor the second-lowest of the season, higher only than the 3.4% points in Round 3.

Three games have above-average MADs, the Collingwood v Essendon game having the highest MAD of all at 6.2% points per Predictor. The Sydney v West Coast game shows the lowest levels of forecast variability, all five Predictor's probability assessments of the Swans' chances falling within a 5% point range from 54% to 59%.


MoSSBODS has eight of the nine home teams actually enjoying a net positive home ground advantage this week, seven of them sized at two Scoring Shots or greater.

Despite that, it has only five home teams emerging victorious, and only the Dogs and Dockers doing so comfortably.

On the Overs/Unders front it has the round generating about 1,571 points, or about 175 points per game, which is almost 10 points below the season-to-date average of 184 points per game.

(NB The following text and table have been updated since the original version to correct errors in the reported TAB Overs/Unders details.)

Comparing those Overs/Unders projections, and the individual team score projections that make them up, with the same projections by the TAB reveals that MoSSBODS foresees lower Total Scores in eight of the nine contests, and lower individual scores for 15 of the 18 teams.

The biggest difference in projected Total Scores is for the Saints v Giants game, which MoSSBODS sees as a game likely to generate only 164 points, while the TAB has it as the second-highest scoring game of the round with an expected tally of 187.5 points.

For just four teams does the difference in individual score projections exceed two goals, with MoSSBODS having the Dockers scoring 89 points and the TAB having them score 110.5, MoSSBODS having the Giants rack up 85 points and the TAB having them score 98, MoSSBODS having the Dees tally 78 points and the TAB having them score 90, and with MoSSBODS tipping the Pies to score just 86 points and the TAB expecting 99.

So far this season the respective performances of MoSSBODS and the TAB in forecasting game scores can be summarised as follows:

  • MoSSBODS MAEs: Home Teams 24.5 / Away Teams 19.6 / Combined 22.1 / Totals 32.0
  • TAB MAEs: Home Teams 22.0 / Away Teams 22.3 / Combined 22.1 / Totals 29.9

My suspicion at this point of the season is that MoSSBODS has Net Venue Effects for home teams that are, on average, too small. So far home teams have scored about 14 points more on average than MoSSBODS has expected (though, to be fair, they've scored over 9 points more than the TAB has expected too). Away teams, in contrast, have scored only 1.6 points, on average, more than expected.