2013 : Round 9 - Results

Until the final game of Round 9 you could describe the games from which our best wagering performances had come using a single characteristic: they were the ones on which Investors had no wagers.

We Investors have no legitimate reason to feel aggrieved by the results of Friday's or any of Saturday's games since our wagers finished well out of the reckoning, having generally established their futility fairly early on in the respective contests, but we were objectively "unlucky" on Sunday, with the Hawks needing just one more goal to land our SuperMargin wager on them and with the Roos just needing to lop only the last 30 seconds off their game against the Crows to hand them a victory and us another successful SuperMargin wager, this one requiring them to win by between 1 and 9 points.

Fremantle were the sole bright spot on an otherwise bleak weekend. Their win by 90 points allowed the Line Fund to finish the weekend 1 and 2, dropping 2.8c to leave the Fund down by almost 18c on the season. That performance from the Line Fund, coupled with the Head to Head Fund's loss in its only wager, and a 0 from 10 record for the Margin Fund, combined to strip another 3.5c worth of value off the Recommended Portfolio. It's now down 12.8c for the season. 

TIPSTERS AND PREDICTORS

It was Head-to-Head Tipsters towards the bottom of the MAFL Leaderboard that fared best on tipping this week, with Consult The Ladder and Silhouette doing best of all in selecting 8 winners from the 9 games, and Easily Impressed I and II, Short-Term Memory II and Ride Your Luck all bagging 7. Most other Tipsters landed 6, missing the underdog wins by the Swans, Dogs and Crows, leaving the top of the Leaderboard unchanged with Combo_NN2 still leading by 3 clear tips. Home Sweet Home had another sad round as just 2 home teams grabbed the competition points this week, which means that home teams have now been only 50:50 propositions so far this season.

ProPred_3 turned in the round's best margin prediction performance with a 27.45 point mean absolute prediction error (MAPE) per game, which was enough to move it into 5th spot on the leaderboard. The RSMP-based Predictors still hold the first two places on that leaderboard and still have season-long MAPEs of under 27 points per game. Two of the Bookie-based Predictors now loom threateningly in third and fourth places.

The H2H-based Predictors recorded the four worst performances of the round with MAPEs of around 33 points per game. They now occupy four of the bottom five positions on the leaderboard, though they all have season MAPEs below 29, which would, in other seasons, be quite creditable figures.

There was no change in the order of the Head-to-Head Probability Predictors this week, with all seven Predictors recording positive probability scores for the round, the seventh time that this has happened this season. Best was ProPred's +0.25 bits per game, and worst were the H2H Predictors' +0.14 bits per game.

Despite recording just 1 win from 3 from its wagering efforts, the Line Fund algorithm had a relatively successful week overall, returning a net positive probability score for it across the 9 games.

SUPERMARGIN PERFORMANCES

Three of the Margin Predictors selected the correct bucket in two games this week - WinPred_3, WinPred_7 and Bookie_3 - though the WinPred Predictors' overall performances were all the more impressive as they also included two predictions that were in error by a single bucket.

Combo_NN2 did not select the correct bucket in any game, but was in error by a single bucket in four contests (including two of the games on which we had wagers). It's now been correct 12 times and in error by a single bucket in 17 other games. That total of 29 games for being within a bucket of the correct answer is the best amongst all Margin Predictors. 

Combo_7 is shaping up as the season's hard-luck SuperMargin performer. It was in error by a single bucket in two games this weekend, which made it 21 games for the season where it's been out by this amount. On only four occasions, however, has it selected the correct bucket.