2015 - Round 4 : Not A Lot To Lose

/

Neither the Head-to-Head nor the Line Fund algorithm think much of the TAB prices on offer this week with the result that they've put only 2.6% of the entire Portfolio at risk across the nine games, the smallest proportion wagered in a home-and-away round since Round 11 of 2013 when byes meant that only six games were played. To find a smaller proportion at risk in a round with a full quota of nine matches you'd need to go back to Round 7 of that same year.

Coincidentally, that round also had the Lions playing the Eagles, and the Saints playing the Blues at home as part of it. Then, the Eagles prevailed 108-82 and the Saints 77-68, though the Saints were playing at a "home" ground some 2,600 km nearer to where they regularly sleep. Home ground status can be such a fluid concept in the AFL.

Back to the wagering. The Head-to-Head Fund has, like the most conservative Jeopardy player of all time, risked just 3% of its original bank, eking it out over three different wagers, two of which will be wholly irrelevant when history reviews this year's MoS performance. Of substance is only the Fund's 2.3% wager on Fremantle at home to Sydney, paying $1.65.

Two Line Fund wagers complete the week's spartan tilt at TAB-baiting, one on the out-of-form Lions, at home, bolstered by 16.5 points start playing the Eagles, and the other on last week's first-up winners in the Cats, also at home, but with only a 7.5 points handicap (or a little over half-a-Gazza) playing the Roos.

No single game represents greater than a 0.8c upside or downside, and the entire collection only spans a range of outcomes from a loss of 2.6c to a gain of 2.2c. If Investors are to claw back the losses of the season to date, substantial progress won't be made this weekend.

With so little activity of financial significance this week, I'll be most interested in the progress of the various Tipsters and Predictors.

HEAD-TO-HEAD TIPSTERS

Combo_NN2 enjoys just a one tip lead on the Head-to-Head Tipster Leaderboard, pursued most closely by the Bookie posse of BKB, B3 and B9, plus Combo_7. Combo_NN2's lead over that group will extend to two tips if the Gold Coast upset GWS, and will shrink to zero if the more-favoured outcome transpires. From amongst the pack a further tip behind, Combo_NN1 and C_Marg both have chances to climb the ladder, with Combo_ NN1 opting for GWS over Gold Coast, and Geelong over the Roos, and with C_Marg tipping Port Adelaide to defeat Hawthorn, and the Dogs to defeat the Crows. 

Looking more broadly across the vista of Tipsters and games we find that only in the round's first game, Friday night's Richmond v Melbourne clash, is there unanimity of opinion. In four other games we find no more than token support for the underdogs, with Essendon favoured 28-2 over the Pies, GWS favoured by the same margin over the Suns, Adelaide favoured by the same margin too over the Dogs, and Home Sweet Home the lone supporter of the Lions over the Eagles.

In the remaining contests there's 21-9 support for the Blues over the Saints, 22-8 support for Freo over the Swans and, in the only game really differentiating the herd, 17-13 support for the Roos over the Cats.

Overall, the level of disagreement amongst the Head-to-Head Tipsters is about the same this week as last, the Overall Disagreement Index coming in at 19% this week compared to last week's 18%. Home Sweet Home, Easily Impressed I, Consult The Ladder, and C_Marg have done most to lift the levels of disagreement this week, a not entirely unfamiliar role for most of them.

MoS' Official Head-to-Head Tipster, Win_3, has opted for two upsets this week: Port Adelaide over Hawthorn, and Geelong over the Kangaroos.

MARGIN PREDICTORS

Combo_NN1, after a disappointing 2014 season in which it finished 2nd-last, and an only mildly above average 2013 where it placed 8th, currently sits atop the Margin Predictor Leaderboard. It enjoys a slim 7-point lead over RSMP_Weighted, looking to relive its glorious 2013 rookie season in which it defied all-comers with a 26.4 MAE.

This week, the average difference between their margin predictions is about 8 points per game, the largest for the Cats v Roos game where Combo_NN1 has the Cats as 12-point favourites and RSMP_Weighted has them as 4-point underdogs.  

The relatively low level of disagreement between these two Predictors is mirrored more broadly across all Predictors, the all-Predictor MAD of 5.0 points per game a season low, eclipsing the 5.3 points per game from Round 1.

This historic low is, in part, a reflection of the high levels of agreement about the likely winners in many contests. In five games all 19 Margin Predictors share a common opinion about who wins, and in two more just a single Predictor dares to differ.

The maximum range of predicted margins for any single contest is 38 points for the Lions v Eagles game, and that range has been almost doubled solely by C_Marg's outlying prediction in that game. A better indication of the high levels of agreement in every game is the fact that the highest MAD for any game is 6.8 points per Predictor for that same Lions v Eagles encounter.

Switching perspective back to a Predictor-by-Predictor view, we find that C_Marg has the round's highest MAD at 8.8 points per game, which is a little higher than Combo_NN2's 8.5 points per game. Last week, MADs of this magnitude wouldn't have placed them in the top half-dozen Predictors.

Low MAD for the week belongs to Bookie_9, making it three out of four weeks now that it's carried this tag. Combo_7, MoS' Official Margin Predictor, has a MAD of just 2.8 points per game, which places it 3rd-lowest.

PROBABILITY PREDICTORS

It's a dogfight - as it usually is - amongst the three most directly Bookie-based Predictors at the top of the Probability Predictor Leaderboard. Bookie-LPSO, who finished 2nd last season to MoS' Official Head-to-Head Probability Predictor, Bookie-OE, currently heads the Leaderboard. Bookie-RE lies in 2nd, Bookie-OE in 3rd, and C_Prob sits 4th, just behind the Bookie trio as was the case at the end of season 2014.

Amongst the Head-to-Head Probability Predictors it's the Port Adelaide v Hawthorn game that causing the greatest division in terms of the more likely victor, five leaning towards the Hawks and three towards the Power.

In probability terms, however, it's the Dogs v Crows game that carries the largest range of predictions (35% points), narrowly ahead of the Lions v Eagles game (34% points), and it's the Dockers v Swans game that's associated with the largest MAD (9.2% points per Predictor).

Overall, the level of disagreement amongst the Head-to-Head Probability Predictors, as measured by the average MAD of 6.0%, is as low this week as it was last week making them joint season lows.

Only three Predictors have MADs above that all-Predictor average, C_Prob chief amongst them with a MAD of 10.2% points per game. WinPred, at 7.9%, and ProPred, at 7.7%, sit second- and third-highest.

The Line Fund algorithm rates only two teams, the Lions and the Cats, as better than 60% chances on line betting and, both being home teams, has authorised the Line Fund to wager on those assessments. The two away teams rated most-likely to prevail on line betting are the Dees (59% chances), and the Swans (56% chances).

DISAGREEMENT HISTORY

As readers are, by now, perhaps beginning to gain some familiarity with the Tipsters and Predictors in the MoS 2015 stable, I though it timely to provide an updated summary of their relative contrarian natures by providing a table of round-by-round Disagreement Indexes and MADs to the end of Round 4.

For the Head-to-Head Tipsters you can see that it's from amongst the Heuristic Tipster Group - those from BKB down - that the more significant levels of disagreement tend to emerge. Home Sweet Home is especially notable for its willingness to tip the untippable. From outside the Heuristic Tipster Group, C_Marg is the Tipster most likely to surprise.

Bookie_9 and Combo_7 tend to produce the lowest MADs each week, and ENS_Greedy and ENS_Linear are demonstrating only mildly contrarian behaviour, if any. C_Marg, after a slow start to the season, is now reasserting its more typical divergent behaviour, while for Combo_NN1, Combo_NN2, and the Win and the H2H families of Predictors, the level of divergence is more a round-to-round proposition.

Amongst the Head-to-Head Probability Predictors, it's the three directly Bookie-based Predictors that show greatest adherence to the norms, while all other Predictors show varying levels of divergence each week, C_Prob and WinPred arguably the most-reliably divergent.