2015 - Round 16 : Well That Escalated Quickly ...

/

We're often advised to "Be careful what you wish for", and I'm wondering now about the wisdom of my desire to see more activity from the Funds than we'd seen recently.

The Line Fund's reawoken for a week's slumber to venture just a couple of bets, but at this point of the season that pair represents 10% of the Line Fund and 6% of the entire original Overall Portfolio. The twin line bets have been supplemented by a quartet more from the Head-to-Head Fund, the same number that it made last week, but with more than double the average size. Its four wagers represent almost 21% of the Head-to-Head Fund and a bit over 8% of the Overall Portfolio.  

Port Adelaide and Geelong are the only teams to have attracted the attention of both the Head-to-Head and Line Funds, but the sizes of the wagers and the prices on offer (both being favourites), have meant that the largest upside and downside rests elsewhere - in the hands of the Swans and the Pies where we've large Head-to-Head wagers at prices marginally over $2.

A set of perfect results this week would add about 14.5c to the value of the Overall Portfolio and complete about three-quarters of the task of returning us to profitability. Disaster in every game though would strip about the same amount from the value of the Portfolio and make season-long profitability an unbackable underdog.

We do have five wagerless games this weekend, however, so there will be some time for reflection and recovery.

HEAD-TO-HEAD TIPSTERS

There's only one true short-priced favourite this week, Fremantle at $1.08, so there's a lot for the Head-to-Head Tipsters to potentially disagree about. They've taken this opportunity with alacrity, between them coming up with 14 different sets of nine tips.

The all-Predictor Disagreement Index of 34% is a season-high and, even should each of the more-favoured teams win, the all-Predictor average for the round will be only 6.8 from 9.

Three games in particular have split the group, Gold Coast and Hawthorn attracting a baker's dozen of dissenters, and Collingwood attracting a mathematician's dozen (base 10) of its own. The Bulldogs also have eight supporters, the Lions seven, Adelaide six, and Essendon five, leaving only St Kilda (with two) and Carlton (with one) as the truly unfancied teams of the round.

While Disagreement Indexes are high across the board, the lowest coming in at 27%, there are some genuine extremes including Silhouette's 57% and Short-Term Memory II's 53%, the third- and fourth-highest Indexes we've seen all season. In this company, Home Sweet Home's 39% almost makes it a conformist.

Disagreement amongst the Tipsters at the top of the table is far less than amongst other Tipsters, however. The only game about which the Top 6 disagree is the Gold Coast v GWS game where joint-leader Combo_7, Bookie_3, and BKB are tipping the Giants, and other joint-leader Bookie_9 and the two ENS Tipsters are on the underdog Suns. With Bookie_9 and Combo_7 differing in their selections for this game, barring a draw we'll have an outright leader of the Head-to-Head competition at the end of this round.

MARGIN PREDICTORS

So far this season, the correlation between the combined disagreement levels of the Head-to-Head Tipsters and those of the Margin Predictors has been just +0.20, meaning that rounds where the Disagreement Indexes for the former have been high have not necessarily also been rounds where the Mean Absolute Deviations (MADs) for the latter have been high too.

This week bears that general tendency out, the all-Predictor MAD coming in at just 5.2 points per game per Predictor, which is the fourth-lowest average of the season.

Two games in particular have generated the widest array of margin predictions: the Pies v Eagles game where the range covers 8 goals and the MAD is 9.3 points per Predictor, and the Swans v Hawks game where the range covers just over 5 goals and the MAD is 8.3 points per Predictor. The Freo v Carlton and Melbourne v Brisbane games have also produced wide prediction ranges, both also above 5 goals, but have generated much smaller MADs.

The range in the Fremantle v Carlton game has been defined by Combo_NN2's 56-point prediction and Win_3's 22-point prediction, while in the Melbourne v Brisbane game it's H2H_Adjusted_3's prediction of a 2-point loss by the Dees and Combo_NN1's prediction of a 31-point win that defines the domain.

C_Marg has camped at the extremities of the range in only two games this week, most notably by tipping the Eagles to win by almost 6 goals and Port Adelaide to win by more than 4 goals. It's also virtually as extreme as Combo_NN1 in tipping the Hawks to win by 19 points. All of this bravery/foolhardiness (one to be omitted) has resulted in C_Marg recording the round's largest MAD of 10.0 points per game. 

Combo_NN1 has the next-most divergent set of predictions, its MAD 8.3 points per game, and thereafter follow the two Win Predictors with MADs of around 7.5 points per game. Bookie_9 has the week's most conformist predictions, its 2.3 points per game MAD the 9th-lowest by any Predictor all season and the 4th-lowest if we ignore its own lower MADs from earlier in the year.

As things stand right now, the three Predictors with the best season-long Mean Absolute Errors (MAEs) have the 5th, 1st and 18th lowest average MADs across the season. C_Marg continues to show that there can be predictive success in non-conformity.

PROBABILITY PREDICTORS

We've the fifth-highest level of disagreement amongst the Head-to-Head Probability Predictors this week as even Bookie-LPSO, who has the week's lowest MAD, has recorded a 4.6% points per game result.

C_Prob's MAD of 10.2% points per game is the high-water mark, this only the third occasion that C_Prob has registered a double-digit MAD since the expected early-season recalibration finished around Round 6. C_Prob has generated this MAD on the back of five probability assessments that are more extreme than any other Head-to-Head Probability Predictor, assessing the Pies as only 17% chances, the Swans as 30% chances, Melbourne as 66% chances, Port Adelaide as 75% chances, and St Kilda as 32% chances. If it proves right in the majority of cases about the more likely winners of those five contests, C_Prob will go a long way to closing the gap on the leaderboard to the three Bookie-based Predictors above it.

Pro_Pred, who has the round's next-highest MAD of 7.4% points per game, has also been extreme in a relatively large number of games, it assessing the Cats as 72% chances, the Suns as 67% chances, the Pies as 68% chances, and the Saints as only 18% chances.

The broadest range of opinions has come in the Pies v Eagles, and Swans v Hawks games where the MADs are about 15% points and 12% points, and the ranges about 50% and 40%, both respectively. Opinions are so broad for these two games that, in both, four Predictors have assessed the home teams as favourites, and four have assessed the away teams as favourites.

No team is this week assessed by the Line Fund algorithm as better than 60% chances on line betting, the nearest assessments being 59% for the Cats, Lions and Power, 58% for the Giants and the Blues, and 56% for the Dons. Of those six assessments, Investors care only about those for Geelong and Port Adelaide.