2011 Round 20 Results: Winning the One (and, More Importantly, Five) Percenters

Yet another good week for Investors, whatever your metric, sees Investor Portfolios at 99.3c thanks to a 5.7c gain by the Head-to-Head Fund and a 7.3c gain from the Line Fund, its sixth profitable round in a row.

Across these six rounds the Line Fund has racked up an impressive 17 and 6 record and added over 45c to the Fund's value. During this same period, the Head-to-Head Fund has also made net gains, despite setbacks in Rounds 18 and 19, adding just under 2c to its value while landing 18 successful wagers from 26.

The week's favourites did all that they could to continue the unbroken string of favourite victories that stretched back to Round 17, but the week's two narrowest favourites both lost by the narrowest of margins. First to lose were the Swans on Saturday night, pipped by a point when Adam Goode's after-the-siren set shot drifted left for a behind. Then, on Sunday, the Crows (queue cliche) staged a dramatic comeback to defeat the Lions by 5 points.

That of course meant BKB scored only 6 from 8 but many other Head-to-Head Tipsters managed to tip the card, so many in fact that the all-Tipster average was 6.9, the 2nd-highest average for any round so far this season. At the top of the Leaderboard, Bookie_9 scored 7 while Combo_NN_1 scored just 6, which has allowed Bookie_9 to take outright leadership on 118.5 from 155 (77%), one tip ahead on Combo_NN_1 on 117.5 (76%) and a further tip ahead of BKB on 116.5 (75%).

Another indicator of the relative predictability of results this season is the fact that Consult The Ladder lies in 8th spot with a 72% record, meaning that the team higher on the ladder has won 72% on all contests this season.

Absolute Prediction Errors took another beating this week as a result of huge victories by the Cats (150 points) and the Pies (138 points), and large victories by the Blues (76 points) and the Eagles (57 points). These victory margins helped to push the all-Predictor absolute prediction error to 39.64 points per game for the round, the 2nd-highest this season.

Bookie_3 negotiated the trouble best, registering a 30.05 points per game performance. Next best was Combo_7, almost 5.5 points per game worse on 35.49, then Combo_NN_2 about half a point further back on 35.84 points per game. Bookie_3's strong performance helped it extend its overall lead over Combo_7, which now stands at 0.81 points per game (28.42 versus 29.33). Bookie_9 lies in third place on 29.91 points per game, the only other sub-30 season-long record.

The Margin Predictors were also poor judges of line betting this week, collectively averaging just 3.6 from 8 and, at best, tipping only 4 from 8. Correlations between season-long line betting predictive accuracy and proficiency as measured by other metrics remain low. Predictors' head-to-head predictive accuracy correlates -0.22 with their line betting accuracy, their MAPE correlates -0.18, and their RMSE correlates -0.23 (the latter two of which are at correlated in the "right" direction since you'd hope that lower MAPE and lower RMSE would mean higher line betting predictive accuracy). It remains the case that a talent for accurately predicting game victors or victory margins is no guarantee of line betting success.

Probability Scores for all Probability Predictors were strongly positive this week - even that for the Line Fund algorithm. Best amongst the Head-to-Head Probability Predictors were H2H Adjusted and H2H Unadjusted, both of which closed the gap a little to the TAB Sportsbet bookmaker.