2016 Round 8 : Roos Snooze and Lose?

It's another measured week of wagering in Round 8, four head-to-head wagers paired with four line wagers on the same teams (and the overs/unders markets still yet to be posted).

Those eight wagers span pretty much the entire wagering landscape, since they include a head-to-head bet on the Dons at $10 and a line bet with 52.5 points start, a head-to-head bet on the Giants at $1.05 and a line bet with a -56.5 point handicap.

It's those Dons bets that tantalise with their combined promise of riches, however, as a win by the Dons would add 9.5c to the value of the Overall Portfolio, and even a draw would deliver a healthy 4.5c filip.

Those Dons bets represent only the second-largest potential loss for the Portfolio though, the Lions capable of imposing a larger, 2c loss should they fail to defeat the Pies.

In total the eight wagers represent 6.2% of the original Overall Portfolio and offer a maximum upside of a bit under 14%.

(It'd be remiss of me not to acknowledge that the inspiration for this week's blog title comes from a recent episode of the Dad and Mog Footy Pod. Definitely worth a listen, especially if you're a Tigers fan seeking solace and empathy.)


The Head-to-Head Tipsters - Home Sweet Home aside, naturally - are almost of a single mind this week, with only the Lions v Pies contest eliciting any serious disagreement. In that game the collective view has the Pies winning, but the vote is only 6 to 5.

The all-Tipster Disagreement Index for the week is 18%, the third-lowest for the season, and is once again boosted by the contrarian Home Sweet Home, whose early-season successes have rapidly faded to leave it now seven tips adrift of the competition leader, MoSSBODS_Marg.

Much higher levels of disagreement are on display amongst the MoS' Margin Predictors this week, their all-Predictor mean absolute deviation (MAD) of 6.2 points per game per Predictor a season-high figure. Elevating that average have been the two ensemble predictors based on MoSSBODS_Marg, both of them so deviant that their MADs are double-digits.

ENS_Linear_MoSS has the most extreme margin prediction in four of this weeks' games, Bookie_3 in three games, and Bookie_9, RSMP_Simple and ENS_Greedy_MoSS in two games each.

Looking across the games, none stands out as having an exceptionally high MAD, though the Hawthorn v Fremantle game almost achieves double-digit status and the GWS v Gold Coast game has a MAD of over 8 points per Predictor. In contrast, the MADs for both the West Coast v St Kilda (2.8 points per Predictor) and Carlton v Port Adelaide (2.5 points) games are very low.

So far this season, a game's MAD has been only a weak indicator of whether or not the set of predictions for that game is likely to span the actual result, as indicated by the chart below.

Games where the range of predictions include the actual final margin are those where the points for the game sit on either side on the horizontal zero line (for example, the 2nd game). Just 13 games so far this season have been of that type, and the average MAD for them has been 5.4 points per Predictor.

There have also been 31 games where the Predictors have all underestimated the home team's ability (so the points all sit above the zero line). The average MAD for these games has been 4.9 points per Predictor.

In the remaining 19 games the Predictors have all overestimated the home team's ability (so the points all sit below the zero line) and the average MAD has been 4.4 points per Predictor.

One final interesting aspect of the margin predictions this week is that the all-Predictor mean absolute margin across the nine games - the average forecast margin of victory, if you prefer - is 24.1 points per game. That's the highest it's been all season, eclipsing the 20.4 points per game from Round 3, and only the third time this season that the average has been higher than 20 points per game. That doesn't augur well for impartial spectators this weekend.

Disagreement amongst the Head-to-Head Probability Predictors is slightly below average this week, their all-Predictor MAD of 3.7% points per Predictor per game the 5th-highest - and thus 4th-lowest MAD - of the season. 

As is now customary, it's MoSSBODS_Prob and C_Prob registering the highest MADs, and the three bookmaker-derived Predictors registering the lowest MADs, MoSSBODS_Prob being this week most deviant in terms of its probability assessments for Adelaide, Hawthorn, Brisbane Lions, and Melbourne. 

Almost as a direct consequence, it's the four games involving those teams where the MADs are highest, especially in the Hawthorn v Fremantle game where the all-Predictor MAD is 7.6% points per Predictor.


MoSSBODS this week has eight of the nine home teams enjoying a net benefit from playing at home, though seven of those eight are assessed as gaining that advantage mostly or entirely because their opponents are flying and not busing to the game.

Essendon is the only home team assessed as being disadvantaged by the venue at which they are playing (viz Docklands), but their relative deficiencies in offensive and defensive abilities, according to MoSSBODS Ratings, make that additional disadvantage fairly moot. MoSSBODS has the Dons as one of seven teams predicted to score 75 points or fewer this weekend, and their opponents, the Roos, as one of four teams predicted to score over 100 points.

MoSSBODS also has five games generating fewer than 175 points, and none generating more than about 190 points.