2018 - Round 26 : Overs/Unders Update

As we venture into the world of score prediction for the second-last time this season, we find near-perfect harmony amongst our four forecasters.

Across the four teams and two games, the differences in score forecasts - team and game totals - nowhere exceed 4 points, and there is total agreement about which will most-likely be the low-scoring team and low- and high-scoring games, and only minor debate about which will be the likely high-scoring team.

(Note that I'm still using "Centrebet" to refer to what are now BetEasy forecasts.)


  • All: West Coast v Melbourne (169 to 172)


  • All: Richmond v Collingwood (160 to 164)


  • MoSHBODS: Richmond and Melbourne (88)

  • MoSSBODS, TAB & Centrebet: Richmond (87 to 90.5)


  • All: Collingwood (72.5 to 76)


With such high levels of agreement this week, there’s no thought of a wager by MoSSBODS. Would that it might have had that same instinct a little more often this season.

Here’s the detail.

Across the two games, the estimated overlay is no greater than 3.3 points in either.


MoSSBODS and MoSHBODS both finished on the right side of TAB and Centrebet totals in only one of the two games last week.

That left MoSSBODS' with a season-long 49% record against the TAB and against Centrebet, and left MoSHBODS with a 47% record against the TAB and a 49% record against Centrebet.

MoSSBODS’ lone wager was an ill-fated overs bet with the TAB. That dropped its overall strike rate on overs bets across the season to 37% and left its strike rate on unders bets at 56%.

On the mean absolute error (MAE) metric, the TAB and Centrebet registered joint-lowest MAEs this week on game margins, away team scores, and game totals, while the TAB recorded the lowest MAE, alone, on home team scores.

That left Centrebet still leading on the season-long view for home team scores (MAE of 17.0) and game margins (25.9), and the TAB still leading on away team scores (17.4) and game totals (22.6).

It’s interesting to note from a review of the Errors section of the table above that:

  • All four forecasters over-estimated home team and away team scores, but home team scores more than away team scores, which meant that all four over-estimated game margins from the home team’s point of view.

  • MoSHBODS over-estimated home team scores least (by 0.8 points, on average), while the TAB over-estimated them most (by 2.3 points).

  • MoSHBODS over-estimated away team scores least (by 0.2 points, on average), while Centrebet over-estimated them most (by 1.8 points).

  • MoSHBODS over-estimated game margins from the home team perspective most (by 0.6 points, on average), while MoSSBODS over-estimated them least (by 0.4 points), fractionally better than Centrebet.

  • All four forecasters also over-estimated total scores, MoSHBODS by the least (2.9 points, on average), and the TAB by the most (4.0 points), fractionally worse than Centrebet.

  • All of which goes to show that, given the general dominance of the two bookmakers on MAE across all four forecasts types, being less biased is different from being less variable.