2017 - Round 18 : Hardest Round Ever?

I've been collecting complete TAB Bookmaker data on the line market since 2012, and the average expected victory margin this week - 12.9 points per game - is the lowest it's been for any home and away round during that period.

In fact, as you can see in the table at right, the entire season has had generally low average expected margins by historical standards, a reflection of even the bookmaker's caution about forecasting large victory margins this year.

That might make for some disagreement amongst the Head-to-Head Tipsters, you might think, but that really isn't the case this week bar for some divergence of opinion in the Fremantle v Hawthorn, and Collingwood v West Coast matchups. In only one other game - Richmond v GWS - is there any Tipster other than Home Sweet Home selecting the underdog.

Overall, the level of disagreement has come in at 21%, which is slightly higher than last week's 18% and just 1% point above the all-season average of 20%.

Home Sweet Home is its usual contrarian self though it's joined by a slightly more-contrarian-than-usual MoSSBODS_Marg this week, they the only two Tipsters above the all-Tipster average Index value.

Most importantly, MoSSBODS_Marg has tipped differently from the two RSMP Tipsters in two games, giving it a chance to grab the outright lead on the Leaderboard should both Richmond and Collingwood win.

The Margin Predictors' average mean absolute deviation (MAD) this week is just 3.5 points per Predictor per game, well down on last week's 4.0 and also below the all-season average of 4.2.

Highest MAD of the week belongs to MoSSBODS_Marg's (7.9 points per game), the first time its been in that position since Round 5. Things didn't go well for it that time as it finished the round with a mean absolute error (MAE) of 32.5 points per game, which was the worst of all the Predictors.

The second-highest MAD is C_Marg's (7.2 points per game) which, while high, is a full point below its all-season average. Third-highest is MoSHBODS_Marg's 4.9 points per game.

MoSHBODS_Marg now leads the RSMP Predictors by about 15 points, and will extend that lead if:

  • Essendon win by 13 points or less (6 point gain)
  • Port Adelaide win by 13 points or more (4 point gain)
  • Sydney win by 23 points or less (8 point gain)
  • Richmond win or draw (8 point gain)
  • Collingwood win by 8 points or more (14 point gain)

From a game-by-game perspective, it's the Collingwood v West Coast game that has the highest MAD (5.7 points per Predictor), mostly because of the contrarian forecasts by MoSSBODS_Marg and MoSHBODS_Marg. The other above-average MADs are in the Adelaide v Geelong game (mostly attributable to C_Marg and MoSSBODS_Marg forecasts), and in the Richmond v GWS game (mostly attributable to C_Marg, MoSHBODS_Marg, and MoSSBODS_Marg forecasts).

For the Head-to-Head Probability Predictors it's the same three games that have elicited the greatest divergence of opinions, most memorably in the Collingwood v West Coast game where the probability estimates are book-ended by C_Prob's 29% (it's not a fan of the Pies at Docklands) and MoSSBODS_Prob's 64%. 

MoSSBODS_Prob has the round's highest MAD (7.4% points per game), ahead of C_Prob (7.0% points per game). 

The all-Predictor MAD is 4.5% points per game per Predictor, which is the highest it's been since Round 9.


Five head-to-head and five line bets define the weekend's wagering so far, these 10 wagers representing 5.6% of the original Overall Portfolio, eclipsing last week's 5.5% and setting a new highest mark for a single round since Round 11.

The larger bets are on Collingwood, which consequently controls over one-third of the downside and almost one-third of the upside this week, a win by them capable of adding 2c to the value of the Overall Portfolio, and a loss by more than a goal capable of subtracting the same amount.

The two other games with significant swings are the Richmond v GWS game (best +1.2% / worst -1.1%), and the Essendon v Kangaroos game (best +1.4% / worst -0.8%).

In total, 5.6% of the Overall Portfolio is at risk, and a best-case set of results is capable of adding just over 6% to that Portfolio.


Here are the MoS twins' views on team and total scoring for the week.

Commentary to follow.