2015 - Round 16 Results: Worse, Much Worse

I think it's probably about time to call this year as a loss-maker for MoS.

One of either the Pies or the Swans needed to win on Saturday for the Portfolio to return a profit for the week and then, when they both failed, Port Adelaide needed to win and cover the spread on Sunday afternoon to limit the damage. Port didn't, the Head-to-Head Fund dropped 18c, the Line Fund shed 0.5c, leaving the Overall Portfolio down by another 7.6c and finishing the week down by 27c in all. The only satisfying thing about the final result was the symmetry it bought to the two Funds' fortunes, with each now down by 27c too.

TIPS AND PREDICTIONS

Eight favourites, many of them narrow ones, grabbed the competition points this week, but the unusually high levels of disagreement amongst the Head-to-Head Tipsters meant that this didn't translate into an exceptionally high all-Tipster average. It was, in fact, just 6.0 from 9, the lowest average for four weeks.

Best were Combo_7's, Bookie_3's, BKB's and Combo_NN2's hauls of 8 from 9, and worst Home Sweet Home's and Short Term Memory II's 4 from 9.

Combo_7's equal league-leading performance was enough to keep it one tip ahead of the field on the MoS Leaderboard, a position it now holds with a 97 from 134 (72%) record. Bookie_3 and Bookie_9 are now tied for second on 96, and BKB holds outright fourth on 95.

This week's all-Predictor Mean Absolute Error (MAE) of 20.5 points per game per Predictor was the lowest for three weeks and the second-lowest of the season - and that with the unexpectedly high 89-point demolition of the Swans by the Hawks. C_Marg returned the round's lowest MAE of just 15.5 points per game for Round 16, a result so low that it dragged its season-long MAE back under 30. Nine other Predictors returned sub-20 MAEs for the week as well, and seven of them also now sport sub-30 season MAEs.

C_Marg's stunning MAE saw it return to the top of the MoS Leaderboard, a position it surrendered only last week to RSMP_Weighted. In terms of ladder positions though, Combo_NN1 was the big mover, its 16.5 MAE lifting it four places from 15th into 11th.

Five Predictors still have season-long line betting records of better than 52.6%, the level of accuracy required to make money line betting at $1.90 prices, but only C_Marg, at 61%, is comfortably above that level.

C_Prob, narrowly, turned in the week's best probability score amongst the Head-to-Head Probability Predictors. Bookie-RE was next best, then Bookie-LPSO and Bookie-OE. The latter Predictor, however, still retains the lead overall, although it's a very narrow one now to Bookie-RE.

The Line Fund algorithm returned a positive probability score this week, the first time it's done so since Round 11. Regrettably, its accuracy was mostly reserved for those games where it assessed the away team as the more likely victor on line betting, so Investors were unable to capitalise on its canny calibration. Still a glimmer of hope perhaps.