As things turned out my rounding policy was irrelevant. What mattered instead was that the Swans' grubbered kick- on-the-siren clattered into the right-hand goal post rather than slipping by it, keeping Sydney's victory margin in the 1 to 9 point range, and that the Roos held their nerve in the final term to topple the Crows and land our head-to-head bet at $3.05.

Those two results proved to be the only cheery spots on an otherwise bleak weekend for Investors. Not that we weren't without other promising opportunities. Collingwood, for a time, looked good things to cover the 15.5 point spread they were offering, though in the end were probably lucky to emerge victors at all, and Fremantle led the Dons by 5 points at the final change only to spend the last term watching the Dons rack up 10 scoring shots to 1.

The net result was that the Head-to-Head Fund rose 2.2c, the Line Fund fell 5c, and the Margin Fund rose 8.8c leaving Overall Portfolios down by one-tenth of a cent on the weekend. As things now stand, the only Fund in profit is the Head-to-Head Fund, and even it is up only by 0.1c.

While the Roos' victory was welcomed by Investors it wasn't by any of the Head-to-Head Tipsters bar H2H_Unadj_10 who has emerged as the sole leader on head-to-head tipping courtesy of having selected the Roos to win by a point. It's now on 83 from 108 (77%), a tip ahead of 12 other Tipsters on 82. 

Head-to-Head Tipsters overall averaged just 3.4 from 6 for the round, the best of them (of which there were many) snagging 4, and the worst, Follow The Streak, scraping up just 2.

Margin predictions were, by this season's standards, quite errant this week, with the all-Predictor average coming in at 35.23 points per game, the 4th-highest average for a single round this season. Best was Combo_NN2's performance of 33.19 points per game and worst was Win_7's 37.42 points per game. Bookie_3 (37.33), Bookie_9 (36.41) and Combo_7 (35.54) all had relatively poor rounds, sufficiently poor in Bookie_3's case to allow Combo_NN2 to snatch 3rd place on the Leaderboard. 

Still, eleven Margin Predictors have season-long sub-30 MAPEs.

None of the Margin Predictors selected the correct bucket on SuperMargin wagering in other than the Sydney v Geelong game, where 10 of them managed this feat, though many of them came within a bucket in the Collingwood v West Coast and Fremantle v Essendon games. In the three other games, successful SuperMargin wagers paid out at $41 and at $17 (twice), so it's perhaps not surprising that none of the Margin Predictors was close to predicting those outcomes.

On line betting, as a group Margin Predictors averaged 3.5 from 6, with Combo_NN1 and Combo_NN2 sharing Best On Ground status, each snaring 5 from 6. Also, we still find ourselves in the amazing position of having every Margin Predictor sporting line betting accuracy levels above 50%.

In head-to-head probability prediction the recent trend of the TAB Bookmaker outscoring MAFL's own Predictors was reversed this week as the TAB Bookmaker recorded the round's only negative probability score and only his third such score this season. ProPred did best amongst the MAFL Head-to-Head Probability Predictors and now sits 3rd on the ladder just a fraction of a point behind H2H_Adj.

The Line Fund algorithm had one of its best rounds of the season in turning in just a very slightly negative probability score for the round.