2014 - Round 5: A Strategic Retreat

/

Next week the Line Fund doubles its bet size, and the week after that the Head-to-Head Fund does the same, so there's a sense of pausing in anticipation of bigger battles to come in this round's relatively subdued wagering.

It's the Margin Fund that's especially quiet this week, venturing just four bets across two games. As I write this - early evening on Wednesday - I'm actually unable to place two of these wagers since, for reasons unfathomable to me, the TAB Bookmaker has decided against framing a SuperMargin market for the Essendon v St Kilda contest. This despite the fact that he's perfectly willing to frame head-to-head, line, Winning Margin and a host of other markets for the same game. It's really quite odd, but I find it hard to believe it's merely an oversight.

Anyway, for the purposes of the blog as it now stands I've assumed that we'll secure a price of $7.00 for the two wagers we desire on Essendon. I'll update the blog once the market has been framed and we're set.

(UPDATE Thursday late PM: The market's finally up and we have now secured both bets at $7.00. I still can't fathom why it took so long for this market to be posted.)

The Margin Fund's other wagers are on the Adelaide v GWS game where it's left us with the dreaded Chasm of Despair by insisting we wager on the Crows to win by 20 to 29 or by 40 to 49 points.

Three wagers comprise the entirety of the Head-to-Head Fund's repertoire, one of them essentially a rounding error of a bet on Sydney at $2.15 and the two others not much larger: a 0.8% bet on West Coast, and a 0.6% bet on Melbourne.

That leaves only the activity of the Line Fund to discuss, which this week is the Fund most active. It's made five wagers, four on teams receiving start and just one on a team offering start.

Combined, these 12 wagers represent only 4.6% of the Fund, a much smaller proportion than last week's 7.3% and than the 6.0% of the week before. 

Nonetheless they still represent a combined upside of 6.3% and a combined downside of 4.6%, so it's not as if there's nothing of substance at stake.

In only two games - Collingwood v Kangaroos and Geelong v Hawthorn - do we have no stake at all, and in the remaining games our minimum downside is 0.5c. A worst-case single game result for us would be an Eagles loss by 3 points or more, which would see 0.9c lopped off the value of the Recommended Portfolio.

Upsides this week range from lows of 0.6c, which could accrue from favourable results for the Lions, Swans or the Dogs, to a high of 1.5%, which we'll enjoy only if Adelaide wins by 40 to 49 points.

HEAD-TO-HEAD TIPSTERS

This week we find moderate levels of disagreement amongst the Head-to-Head Tipsters, most of it in the Collingwood v Kangaroos, Sydney v Fremantle, and West Coast v Port Adelaide contests. In each of those games there's at least 30% support for the minority selection. Note that this week I've added two rows to the Head-to-Head Tipping table that convert the combined Tipster data into percentages. In the first new row, labelled "Raw % Home Team", the percentage calculation is straightforward and each Tipster counts equally, whereas in the second new row each Tipster's prediction carries a weighting based on its predictive accuracy across the season to date. So, for example, because on average slightly better-performed Tipsters have chosen Collingwood to win (viz Bookie_9, Bookie_3 and Combo_7), its Weighted percentage support is marginally higher than its Raw percentage support.

In three games there's only a single dissenting Tipster, though this week we can blame Home Sweet Home for only one of them. In the Essendon v St Kilda game, it's Silhouette that's the lone Saints supporter, and in the Western Bulldogs v Carlton game it's Combo_NN_2 that's the solo Blues fan.

Home Sweet Home has the week's largest Disagreement Index, though Silhouette's not far behind it.

MARGIN PREDICTORS

In six contests the Margin Predictors' average predicted margin is less than 2 goals, narrowest of all in the Collingwood v Kangaroo contest where they favour the Roos to win by just 0.3 of a behind, and in the West Coast v Port Adelaide game where they favour the Eagles to win by only a single behind more.

The Sydney v Fremantle game, where the average predicted margin is only a further point larger still, the range between the highest and lowest predicted margins is the smallest of all the contests at just 12 points.

The rows labelled Range, SD and Weighted Mean are new for this table this week. Range and SD (Standard Deviation) have fairly self evident definitions, and the Weighted Mean row, similar to the Weighted % row in the Head-to-Head Tipster table above, is based on weighting each Predictor's opinion using its relative Mean Absolute Prediction Error (MAPE) for the season so far. Each Predictor's margin prediction is weighted by its MAPE for the season to date relative to the all-Predictor MAPE.

Largest in terms of average, range and SD is the Adelaide v GWS game where the Predictors collectively think that the Crows will win by about 8 goals, though their opinions span a range of about that same size.

Bookie_3 and ProPred_7 are the Predictors with the highest Disagreement Indices this week, relegating C-Marg into equal 3rd with Win_3.

PROBABILITY PREDICTORS

Six of the games in Round 5 see the Head-to-Head Probability Predictors on the same side of 50%, as reflected in the row labelled "Combined", one of three new rows in the table below.

The other new rows are the ones labelled Range and SD, which have the obvious definitions (though note that the SD figures are here in percentage point terms).

It's the Collingwood v Kangaroos matchup that's best divided the Predictors, with four Predictors finishing up in each camp. At 19%, however, the range of opinions in relation to this game are far narrower than those in the Melbourne v Gold Coast matchup, where C-Prob has distanced itself considerably from the pack by rating the Dees as 57% chances, a full 18% points higher than the assessment of any other Predictor.

That prediction, combined with some other outlying ratings in other games by C-Prob, has been enough to make it the Predictor with the highest Disagreement Index (here, the Mean Absolute Deviation) this week.

According to the Line Fund algorithm, St Kilda are almost 80% chances to go down to Essendon by 44 points or fewer, thereby securing victory for their line bettors. Sydney, Hawthorn and West Coast are the only other teams whose line betting chances are assessed as being 60% or higher.

ChiPS PREDICTIONS

In only one game does ChiPS' margin prediction have a different sign to the difference in its Team Ratings. That game is the Melbourne v Gold Coast game where the roughly 16 Rating Point advantage that the Suns enjoy is more than offset by the combined effects of Melbourne's superior recent form, their Home Ground Advantage, and the fact that the contest is an Interstate one.

(Melbourne's superiority of form in comparison to Gold Coast's is an interesting assessment by ChiPS and not one I'd defend loudly in front of a knowledgeable audience. It rates Melbourne's loss to GWS by 32 points and its win over Carlton by 23 points, collectively, more highly than it rates Gold Coast's win over Brisbane by 53 points and loss to Hawthorn by 99 points. You could argue that one for days ...)

ChiPS' only margin predictions larger than 15 points in absolute size come in the Essendon v St Kilda game where the Dons are assessed as having superior class and form, and about a 3 point Home Ground Advantage, and in the Adelaide v GWS game where the Crows enjoy superior class and form, and benefit from the Interstate nature of the clash. The Crows also suffer a minor deduction because of their relative inexperience at Adelaide Oval, but still come out assessed as likely 40 point winners.

Based on ChiPS' probability assessments, two teams are worthy of head-to-head wagers this week: Brisbane Lions and Melbourne.