2016 Round 20 : Overs/Unders Update

/

The Overs/Unders Fund has made just three wagers this week, the fewest it's made since Round 13 when only six games were on offer, and only the second time it's made so few bets in a 9-game round. With experience comes wisdom (or is that uncertainty?).

For the two unders bets, MoSSBODS' assessment is that the total will be more than 2 goals smaller than the TAB's Total, while for the lone overs bet its assessment is that the total will be just over 1 goal higher.

On such smallish margins do our wagering fates rest this weekend.

Combined with the seven head-to-head and line wagers, the three overs/unders bets take the total amount at risk this week to just 5.8% of the original Overall Portfolio, which is the lowest proportion at risk since Round 15.

The good news about that is that, no matter how badly the round goes, the Overall Portfolio will end Round 20 in profit for the season.

MoSSBODS COMPARISON TO THE TAB

The relative lack of overs/unders wagering activity this week reflects a greater similarity of opinion between MoSSBODS and the TAB bookmaker. In the six contests in which no wager has been made, the absolute difference between MoSSBODS' predicted total and the TAB bookmaker's is no higher than 7 points.

That said, MoSSBODS continues to predict slightly lower totals than the TAB bookmaker. Its all-game average score is 173 points, while the TAB's is 178 points.

This week, Geelong is the team for which MoSSBODS' projected score is most different from the TAB's, MoSSBODS predicting the Cats to rack up 107 points against the Dons and the TAB foreseeing a much-larger 125.5 points.

The only other teams for which the absolute differences exceed a goal are Hawthorn (98 MoSSBODS, 107.5 TAB), the Kangaroos (78 MoSSBODS, 86.5 TAB), and St Kilda (81 MoSSBODS and 88.5 TAB).

Looking lastly at game totals, both MoSSBODS and the TAB have the Crows v Lions game as the weekend's highest-scoring game, and have the Dockers v Eagles game as its lowest-scoring.

The latest comparative performance data for MoSSBODS and the TAB is as follows:

Home Team Scores

  • MoSSBODS: 19.8 MAE; +5 Ave Error (Actual - Predicted)
  • TAB: 17.7 MAE; +1 Ave Error 

Away Team Scores

  • MoSSBODS: 19.9 MAE; +1 Ave Error
  • TAB: 20.1 MAE; -4 Ave Error 

Aggregate Scores

  • MoSSBODS: 27.2 MAE; +7 Ave Error
  • TAB: 25.5 MAE; -3 Ave Error 

Game Margins

  • MoSSBODS: 29.2 MAE; +4 Ave Error
  • TAB: 28.6 MAE; +5 Ave Error 

There are some early signs that the switch in methodology has, as hoped, improved MoSSBODS' ability to predict totals in that it has produced a lower MAE than the TAB in 2 of the 3 rounds during which the new methodology has been used. Prior to the introduction of the new methodology MoSSBODS had done this only 4 times in 16 rounds.

During that same period, MoSSBODS has produced lower MAEs for Home Team scores in 1 of the 3 rounds, and lower MAEs for Away Team scores also in 1 of the 3 rounds. These compare with 6 from 16 (Home scores) and 9 from 16 (Away scores) records under the older methodology, suggesting there's no compelling evidence that the change has affected MoSSBODS' ability to predict these aspects of a game.