2012 Round 4 Results : A Win But Not a Profit

Investors' last-minute wager on the Giants was the only thing standing between them and a small profit this weekend, though to call that wager out simply because it was the last one placed, is to do it an injustice. Certainly there were other wagers that teased for far less time.

As it was, a Crows victory by between 20 and 29 points was a legitimate possibility even quite late into the game, so Investors had a much longer ride on their $21 longshot than I imagine most were expecting.

It was down to the Hawks yet again, this week in partnership with the Eagles, to deliver a result in the requisite range. Their loss by 5 points, which secured our SuperMargin wager at $7.50, was enough to all but cover the losses on the remaining seven wagers, the final result being a 1.25% loss for the Margin Fund and a 0.25% loss for the overall Portfolio. That leaves the Portfolio now priced at 98c.

Wagering aside, generally this week's results were much what was expected, the favourites losing in only two games, the more startling of these losses being the Blues' loss to the Dons on Saturday. With seven favourites winning, BKB along with a number of other Head-to-Head Tipsters finished with 7 from 9 for the week, which helped lift the average tipping performance to a creditable 6.4 from 9. So predictable has the season been so far, however, that this average is only the third highest - or, if you prefer, the second lowest - of the season.

The Head-to-Head Leaderboard now has 13 tipsters sharing the lead, each on 29 from 36 or about 81%. 

Seven of the games this week ended with 6 goals or fewer separating the teams, and the remaining two produced margins of just 46 and 65 points, which made for a jolly week of margin predicting.

The average absolute prediction error across all Margin Predictors was a season-low 27.32 points per game. Best performed was HU3 on 22.14 points per game and worst was CN2 on 32.48 points per game. B9 sits atop the leaderboard with a season-long MAPE of 29.29 points per game, a fraction ahead of B3 on 29.39 points per game. These are the only two Margin Predictors with sub-30 MAPEs.

Accurate margin prediction translated into accurate line betting market prediction for many of the Margin Predictors, with the average performance coming in at 6.1 from 9 driven up by 8 from 9 scores by all 4 of the Head-to-Head based Predictors. This second successive week of better-than-chance line betting prediction has lifted all the Margin Predictors other than CN2 up to or above the 50% mark for the season.

Head-to-head probability predictions were also accurate, the aggregate probability scores for all the Head-to-Head Probability Predictors except WinPred finishing positive for the fourth successive week. Season-long probability scores remain absurdly high, highest of all for H2H_Adj, which leads the pack with an average score of 0.345 per game.

Our Line Fund algorithm did not fare so well this week, returning its worst aggregate probability score for the season - not an ideal fact to be reporting the week before the Fund that is based on this algorithm goes live. To date, the Line Fund has correctly predicted the winner on the line market in just 15 of the 36 contests, which makes me glad that I've waited this long before taking any notice of its opinions.

To finish, here's the update on the SuperMargin performance of all the Margin Predictors:

Looking firstly at their performance had we followed their predictions in every game, it's HU3, HU10 and HA3 with the stand-out results. They've each predicted the correct bucket in 6 of the 36 games and been within a bucket of landing another 6 or 7. Despite these stellar numbers, wagering on them would have only produced an ROI of +27%. Adopting the strategy of wagering also on the buckets on either side of their preferred bucket would have instead resulted in losses, ably demonstrating the in-practice folly of this in-theory interesting idea.

Turning instead to the results if we wagered only when a Predictor selected a bucket implying a draw or a Home team win, it's the same three Predictors that lead out. They've each landed 4 of their bets of this type with an aggregate ROI of +26%.

Were we to follow the bucket-either-side approach instead, by far the best result is that of W7 with a +21% ROI from 9 collects.