2011 Round 7: No Game Left Behind

After three weeks of only seven games per round (though, thanks to the AFL's game-rationing, it certainly didn't feel as though we were being starved of football) we now enter a nine round period where we'll enjoy the full octet of games each round. 

In this, the 7th Round, we don't start the round early but we do finish it late, with a Monday night game between the Saints and Carlton ending proceedings on what Investors will be hoping will be a profitable note. 

Consistent with the frenetic pace at which both Funds have been identifying allegedly value bets each week, we've this week Head-to-Head wagers in five games, and Line wagers in seven. Three of the Head-to-Head bets, including the two largest, are on favourites, while those on the underdogs, which top and tail the round's action, are relatively constrained - a 3.6% bet on Port at $4.25 and at Home playing the Hawks, and a 5.1% bet on the Saints, also at Home, at $2.65 taking on Carlton. If there's such a thing as a balanced set of wagers, I reckon this is one such set.

The Line Fund is giving start in four of its wagers, and receiving it in the remaining three, including, to my dismay, in a bet of just over 7% on the Suns +49.5 points playing the Lions at their twin home, the Gabba. The TAB Sportsbet bookie, it seems, is now understandably nervous about the Suns' ability to set early in a game, so they're offering about 6 points of additional start to the Suns than their head-to-head price would normally suggest. (In Round 21 of 2010, for example, Richmond were priced, as the Suns are now, at $7 on the head-to-head market, but were only being offered at 43.5 points start.)

In statistical terms, it's as if the bookie's assuming the distribution of the handicap-adjusted margin for this game has a much broader standard deviation than that of a typical game.

Anyway, here's the detail of the wagering:

In four of the games - the interstate clashes - the incremental Venue Experience for the Home team is considerable. For the other games we have the Dogs and Sydney both playing at the mutually unfamiliar Manuka Oval, the Cats facing the Roos and enjoying only a modest Venue Experience advantage courtesy of the decision to play so many of their home games at the G last season, the Suns playing the Lions at the Gabba in only their third home game ever, and the Saints playing Carlton at Docklands, which is a venue at which both teams have considerable recent experience.

Here's what all that wagering might mean for Portfolios:

Wins by Port Adelaide and St Kilda, and merely a modest thumping for the Suns represent this weekend's tallest Ladders, while a Roos win, a Lions thumping of the Suns, and a comfortable Blues victory represent the longest Snakes.

On tipping, the majority-opinion of the Head-to-Head tipsters lies with the favourites in six of the eight clashes, the Tigers and Gold Coast being the only underdogs to enjoy majority support amongst them. In none of the games, however, is there unanimous support for either the favourite of the underdog. Hawthorn and Geelong come closest to a 13-0 vote, with Home Sweet Home the holdout for Port on the basis of the game's venue, and Short-Term Memory I the holdout for the Roos on the basis of their 60-point win against Port last weekend.

As a group, the Margin Predictors disagree with the Head-to-Head tipsters in three games, going instead quite solidly with the favourites in the form of Freo (playing Richmond) and the Lions (playing Gold Coast), and opting for the underdog Saints to topple the Blues.

Measured by the slimness of tipster plurality, it's the Saints v Blues game for which it's hardest to predict the winner; measured by the size of the standard deviation, it's the Gold Coast v Lions game for which it's hardest to predict the margin.

Last week I introduced the idea of using the manhattan distance to measure the disparity between any pair of tipsters' margin predictions. This measure tells us how much ground one tipster could make up on another in terms of MAPE if all results went in its favour. Here's the information for this week:

So, for example, if all results favoured Combo_NN_2 relative to Bookie_3 it could make up 62.1 points, which is almost exactly the amount by which it currently trails Bookie_3 (viz 61.9 points, which you can calculate by multiplying the difference in their MAPEs, 26.98-25.60, by the number games so far this season, 45). Were this to happen, Combo_NN_2 would grab first place on the MAPE ladder.

The gap between the current second- and third-placed Predictors, Combo_NN_2 and Combo_7, is just 2.3 points, yet their margin predictions are 40.1 points different, so there's a far greater opportunity for a change of rankings between these two Predictors.

Next, to the Probability Predictors where we see H2H Unadjusted disagreeing with the TAB Sportsbet bookmaker most notably in the Dogs v Sydney game where H2H has Sydney as favourites, in the Gold Coast v Lions game where H2H rates the Suns a 40% chance when the bookie rates them around 13%, and in the Saints v Carlton game where H2H has the Saints as narrow favourites.

ProPred and WinPred also disagree with the bookies in the Saints v Carlton clash, more dramatically in fact than does H2H. They have the Saints as 60%+ chances.

WinPred also disagrees with the bookies in the Richmond v Fremantle game where it has Richmond as narrow favourites.

Our Line Fund feels particularly confident about the line result in three of the contests, rating the Suns 70% chances, Geelong 66% chances, and Richmond 64% chances. Only in the Port v Hawthorn clash does it rate neither team at least a 55% chance, which is the reason that this is the only contest in which we've no Line Fund wager.