2015 - Round 21 : One Of Each

It's taken most of the season, but both the Head-to-Head and Line Funds seem now to have come to the same conclusion: that the TAB Bookmaker has been pricing home teams unattractively this year in both the head-to-head and line markets. Taking the time for wise reflection is often, but not always, the most lucrative course of action.

Evidence for the hypothesis that the Funds have reached this point comes from the fact that they have, collectively, put less of the Overall Portfolio at risk in the last four rounds combined than they did in both Rounds 16 and 17 considered alone.

This week's contribution is a lone Head-to-Head bet of 1.4% on the Roos at $1.95 versus Freo, paired with a lone Line Bet of 2.5% on GWS with 19.5 points start at $1.90 versus Sydney.

(And, by the way, when was the last time a team sitting four wins behind and six positions lower on the competition ladder started as near equal-favourites, albeit when they were playing at home?)

Again then it's a week where little damage can be done or reparations made to the Overall Portfolio, a pair of devastating results knocking just 2c off its value while two best-case results might add just 1.8c.

(It's also a shortened Ready Reckoner for another week, there being seemingly little point in showing the final three games of the round, bereft as they are of wagering activity.)


There is though, as it's said, still "all to play for" in the various MoS Tipping competitions, no less so for the Head-to-Head Tipsters, who this week have turned in a set of tips with slightly below-average levels of disagreement. The all-Tipster Disagreement Index of 21% is the 7th-lowest of the season so far and has been driven up exclusively by Heuristic Tipsters, most impressively by Home Sweet Home whose own Disagreement Index is 51%, and by Silhouette whose Index is 40%.

The levels of dissent amongst MoS' best Head-to-Head Tipsters is very low, however, the Top 6 Tipsters disagreeing about the outcomes in only a single contest, C_Marg and Bookie_9 both plumping for the underdog Roos against Fremantle. In all eight other contests, the Top 6's tips are identical and in support of the favourites.

It's that same Roos v Freo contest that's generated most disagreement amongst the Tipsters more broadly, 10 of them ultimately lining up behind the Roos. Only two other games see more than five Tipsters forming a minority group, eight opting for the Dogs to surprise the Eagles, and five for the Dons to so something similar to the Suns.

In only one game is there unanimity of opinion, that coming in the Crows v Lions game where all 30 Tipsters have opted for a Crows win. I'm a bit surprised that only $7 is being offered about the Lions.


The Margin Predictors have displayed about average levels of disagreement this week, their all-Predictor average Mean Absolute Deviation (MAD) of 5.5 points per Predictor per game the 11th-highest of the season (and also, therefore, the 11th-lowest, this being Round 21).

Combo_NN2 and C_Marg have done most to lift the average MAD this week, their individual MADs of 10.4 and 9.5 points per game, respectively, some 2.5 points per game higher than any other Predictor. Combo_NN2 has achieved its high MAD partly by having the most extreme predictions in three contests, while C_Marg has done similarly by being most extreme in four contests.

Looking across the games we see that it's the Hawks v Port game that has elicited the broadest range of opinions, the MAD in that game coming in at 10.3 points per Predictor and the span of victory margins foreseen for the Hawks spanning 38 points, from a 3-goal to an almost 8-and-a-half goal win.

The only other game with a significantly above-average MAD is the Roos v Freo game where the MAD is 8.9 points per Predictor and the range is 34 points. This is the only game, however, where we find Margin Predictors on either side of zero, 11 ultimately opting for a Roos win, and 8 for a Dockers victory.


It's also a week of near-average disagreement amongst the Head-to-Head Probability Predictors, their all-Tipster average MAD of 4.8% points per game per Predictor the 10th-lowest of the season so far. 

Amongst their ranks it's C_Prob and WinPred doing most to drive the MAD up, C_Prob recording the round's highest MAD for the third week running and WinPred recording the second-highest also for the third week running. C_Prob has achieved its MAD max (sorry) by providing the most extreme probability prediction in eight of the nine contests, WinPred, by comparison, doing likewise in just four.

C_Prob is most at risk of falling back behind the Bookie trio of Predictors if Sydney, Geelong or Fremantle wins.

Four games have significantly above average MADs this week, the highest for the Pies v Tigers game where the all-Predictor average MAD is 8.7% points per Predictor though the predictions for that game span only 19% points from 16% to 35%. The next-highest MAD is associated with the Blues v Dees game where it's 7.9% points per Predictor and where the range is 22% points from 18% to 40%.

The Line Fund algorithm has installed Melbourne (68%) and Richmond (65%) as its hot tips in the line market, though it's also attached significant probabilities to the Western Bulldogs (59%), Fremantle (55%), and GWS (54%). Were it permitted to wager on Away as well as Home teams, the Line Fund would have been far more active this week. I'm not sure whether I'm pleased or disappointed about that.