2013 : Round 1A - Wagers & Tips (A Few Wagers, Lots of Tips)

It puzzles me still why the AFL would choose to start the season not with a bang, but with a whimper, but whimper it is as just two games are scheduled for the first week of the first round of the season.

And it's not as if either game could credibly wear the tag "blockbluster". Friday night sees Adelaide take on Essendon at Football Park, while Saturday promises a Fremantle v West Coast Derby at Subiaco. Interesting? Sure. But season-opening? I don't think so.

Anyway, with the newfound freedom to wager from Round 1, both the Line and Margin Funds have voted with their - well ours, really - cash.

(This week, I've opted to show the information for all 9 games, despite just 2 being in-play. I've done this to give everyone a chance to get used to the new formats, charts and so on. Please note that the wagering and tipping details for the remaining 7 games of the Round are very likely to change prior to next week.)

So, we've no Head-to-Head bets, two Line bets, and two Margin bets, though both Margin bets are on the same team and for the same points range.

Recall that, under the new arrangements for 2013, the Head-to-Head and Line Funds remain relatively shy even when they do bet for the first few rounds of the season, whilst the Margin Fund seeks to trade on its historical superiority in the early parts of seasons by wagering more from Rounds 1 to 11.

Still that means we see one-fortieth of both the Line and Margin Funds at risk this "weekend". Here's the ups and the downs of that:

In the Adelaide v Fremantle game, our best result is a Crows win by 16 to 19 points, which would land all three wagers and would lift the Recommended Portfolio by 1.5c. In the Freo v Eagles clash our wishes are more diffuse: we just want Freo to win, draw or lose by less than a goal. That's worth about another 0.6c to us.

RESULT AND MARGIN TIPS

I've made a few changes in relation Head-to-Head and Margin Predictors this season, as well as changes to the Probability Predictors, as you'll detect in the image below. 

Firstly, if we look at the head-to-head tips, amongst the Head-to-Head Predictors opinion is disproportionately and algorithmically swayed by the opinions of Consult The Ladder (CTL), so we see relatively high levels of agreement across all 9 games (although I again remind readers that anything pertaining to the last 7 games of the round is subject to revision once the TAB markets are confirmed and/or released). The new bar chart on the right depicts the high levels of agreement by virtue of the large swathes of green or red for each contest, signifying near agreement by the Predictors about the eventual winners.

The '% Disagreement' figure at the foot of the predictions of each Tipster provides a measure of how different from the average Tipster are each individual Tipster's predictions. Formally it's how likely that a tip selected for a game at random will be different from the tip of another Head-to-Head Predictor for the same game, also chosen at random. On this metric we can see that Home Sweet Home (HSH) is the most contrarian of all the Predictors, given current predictions.

In the next block, which pertains to the Margin Predictors, we find even more changes relative to last season.

Firstly, we've 3 new predictors: 

• LPSO, which is based on a Log Probability Score Optimised predictor of Home team victory probability, and
• Two Really Simple Margin Predictors (RSMPs), whose historical performances rival anything I've created previously. The first is an evenly-weighted ensemble of RSMPs, while the second is an historically-optimal weighting of those same RSMPs (although a number carry a 0% weighting.

There's also a new, diagrammatic summary of the Predictors' opinions, rendered as boxplots, which show the dispersion of underlying opinion. The boxplot for each game comprises the following: 

• "Whiskers" that extend to encompass the entirety of the range of predictions
• A shaded box that spans the range of the middle half of the predictions
• A grey bar, which marks the average of the predictions
• A red bar, which marks the median 

The main purpose of the boxplots is to depict the broadness or narrowness of opinion in each game, a depiction that is supported numerically by the standard deviation figures recorded on the right of the relevant section. This week - for now, at least - it's the GWS v Sydney clash that is eliciting the greatest variability in opinion across the Predictors. Both of this weeks' clashes are, in contrast, associated with relative uniformity of opinion.

The numbers at the base of this section of the table record the average absolute difference between a Predictor's margin prediction and those of its colleagues. Here too, a smaller number connotes greater congruence, which means that P7 is the Predictor most aligned with the average position of all Predictors.

Lastly, we come to the Probability Predictors, amongst which we'll review the opinions of the Line Fund first of all. The handicaps on offer this week have proven sufficient to draw the Line Fund into two wagers, one on the favourites, Adelaide, and the other on the narrow underdogs, Fremantle. For the rest of the round, at this point the Line Fund is tending to favour the Home teams.

The remainder of this part of the chart is dedicated to the predictions of various transformations of the TAB Bookmaker head-to-head prices, ProPred, WinPred and the Head-to-Head algorithm (and its adjusted companion). The stripe chart on the right of this portion of the table shows the opinions of the three Bookmaker-based Predictors, WinPred, ProPred, and the Head-to-Head Predictor, unadjusted. Here, the more clustered the stripes, the more similar the underlying predictions and we see moderate levels of disagreement amongst the Predictors for the two games of this week.

The Average and Range figures recorded at the far right of the chart provide information about the average of the six main Predictors and the range, in percentage point terms, between their lowest and highest probability estimates for the Home team.

Finally, the numbers at the foot of this section of the chart record the average absolute percentage point difference between a Tipster's margin predictions and the average of those of its colleagues. These show that ProPred is most attuned with the will of the collective Margin Predictors.

That's it then. Any comments, positive or negative, about the new format for the Wagers & Tips section would be very much appreciated, as would any comments about MAFL in general.

Here's to a busy, lucrative, quiet first week.