2016 Round 23 : Wagering and Predictions Update

A quick errata to begin with.

Today I discovered that I had the Roos v Giants, and Suns v Power games in the opposite order to the TAB, the upshot of which was that I transposed the prices and lines for those two markets.

They weren't all that different, but different enough to:

  • Slightly reduce the line bet on the Roos to 1.6% (who are actually only receiving 12.5 points start, but are priced at $1.95)
  • Trigger head-to-head and line bets on the Suns

The final wagers and prices are then as follows.

Apologies to anyone who was inconvenienced by this. Fortunately, the changes are small or additive.

The updated ready reckoner for the nine head-to-head and line bets appears below and now shows a total at risk of 7% of the original Overall Portfolio and a maximum upside of 9.5%..

Also slightly affected are the Margin Predictors other than C_Marg and the three MoSSBODS-based Predictors, and all three Bookie-based Probability Predictors. The updated forecasts appear below.

For the Roos v Giants game, the adjustments for the Margin Predictors are all in the 1 to 3 point range and for the Probability Predictors around 1% point. For the Suns v Power game the changes are slightly larger, the margin predictions moving by between 2 and 4 points, and the probability assessments by 4 to 5% points.


MoSSBODS finds itself broadly agreeing with the TAB Bookmaker this week, finding just three games where its views about the final total score differ sufficiently to warrant a wager.

Two of those bets are on the under, and the third on the over.

They bring the total proportion of the Overall Portfolio at risk this week to just over 8%.

A direct comparison with the TAB reveals that MoSSBODS once again tends to predict smaller totals than the TAB, but in most cases by less than 10 points.

At the team level, only four of the differences are in double-digits, Geelong, Carlton, Hawthorn and Fremantle all projected to score between 10 and 12 points less by MoSSBODS than by the TAB. For two of those teams - Hawthorn and Geelong - that difference is enough to see the TAB Bookmaker projecting a score of over 100 for them, which means that he has six teams recording such a total to MoSSBODS' four.

In terms of expected aggregate score, the ordering of the nine games is very similar for MoSSBODS and the TAB, both tipping the Saints v Lions game to be the highest-scoring game of the round, and both tipping the Dockers v Dogs to be the lowest-scoring game of the round.

Across the nine games, MoSSBODS is projecting an average total score of 172 points to the TAB's 178 points.

Last week, MoSSBODS continued its improved aggregate prediction performance by recording a better mean absolute error (MAE) than the TAB once again. It also recorded a lower MAE for away team scores, though the TAB was more accurate on game margins and home team scores.

The updated all-season records are now as follows:

Home Team Scores

  • MoSSBODS: 19.0 MAE; +5 Average Error (Actual - Predicted)
  • TAB: 17.2 MAE; +1 Average Error

Away Team Scores

  • MoSSBODS: 20.3 MAE; 0 Average Error
  • TAB: 20.5 MAE; -5 Average Error

Aggregate Scores

  • MoSSBODS: 26.5 MAE; +5 Average Error
  • TAB: 25.1 MAE; -4 Average Error

Game Margins

  • MoSSBODS: 29.4 MAE; +5 Average Error
  • TAB: 28.7 MAE; +6 Average Error