2016 Round 18 : Overs/Unders Update

In keeping with this week's theme of "balance", the Overs/Unders Fund has made four bets this week, two on the unders and two on the overs.

Three of the four wagers are on games overlooked by the Head-to-Head and Line Funds, with the result that the Funds, together, have risked money on seven of the nine contests. In total, they've put in play 7.4% of the original Overall Portfolio.


These latest overs/unders wagers have come about based on the considered, digital opinions of the MoSSBODS System, which this week predicts lower totals than the TAB in seven games (despite the recent tweak to its biases that should tend to increase its estimates) and higher totals in only two.

The thresholds applied to MoSSBODS' outputs relative to the TAB totals offered meant that five of its suggestions for "unders" wagers were ignored, but both of its suggestions for "overs" wagers were accepted.

At a team level. MoSSBODS predicts lower scores than the TAB for 10 teams this week, and higher scores for eight, though in only two instances does the difference reach double digits in absolute terms. MoSSBODS has Port Adelaide scoring 99 points to the TAB's 83, and has Sydney scoring 94 points to the TAB's 109.

Both MoSSBODS and the TAB have West Coast, Essendon and Hawthorn scoring 100 points or more this week, but the TAB alone adds Sydney to that list, and MoSSBODS alone adds GWS.

MoSSBODS' and the TAB's opinions about the round's likely high-scoring contest also differs, with MoSSBODS selecting the Power v Giants game, and the TAB selecting the Dons v Lions game (which MoSSBODS has as the second-highest scoring). Both agree, however, that the Swans v Blues game is most likely to be the low-scoring game, though their estimates of the total differ by almost four goals.

Though it is, of course, erroneous to draw firm conclusions from a small sample, it's interesting to note that the first week of operation of the new MoSSBODS total and team score prediction methodology resulted in MoSSBODS registering a smaller mean absolute error (MAE) than the TAB on game totals and on home and away scores, averaged (it did worse on Home teams but much better on Away teams).

The latest full season comparison is now as follows:

Home Team Scores

  • MoSSBODS: MAE 19.8 / Ave Error +7 (Actual - Predicted)
  • TAB: MAE 17.8 / Ave Error +2

Away Team Scores

  • MoSSBODS: MAE 18.9 / Ave Error +2
  • TAB: MAE 19.4 / Ave Error (4)

Total Scores

  • MoSSBODS: MAE 25.7 / Ave Error +9
  • TAB: MAE 24.2 / Ave Error (2)


  • MoSSBODS: MAE 29.0 / Ave Error +5
  • TAB: MAE 28.3 / Ave Error +6