Some games can be strange affairs for me now. As well as posting tips and wagers here on MoS, I also participate in the FMITL and in all three Monash Competitions, using different algorithms in each to exploit the particular scoring mechanics employed. That leaves me with divided - sometimes trisected - loyalties.
Take this week's Richmond v Collingwood game as a good example of my confusion. Investors, me included, needed the Tigers to win to land our head-to-head bet, but my Monash interests were best served by a narrow Pies win. My FMITL fortunes required only that the Pies win, the size of that win being unrelated to the reward. In the end I found myself weighing up the merits of a draw, which would have meant a small wagering loss, a reasonable Monash score in two of the three competitions, and a no result in FMITL. It's my own doing, I know, but it does complicate things.
Anyway, in the end, Investors got the result they wanted in a Tigers win, that win the sole successful wager by the Head-to-Head Fund this weekend, which nevertheless managed to reverse the trend of recent weeks and creep up by 0.6c to now be down by just under 8c on the season.
The Line Fund also showed pleasing form, landing two bets from three to climb by 2c, leaving it down by just over 9c on the season. Combined and Portfolio-weighted, the two Funds increased Investor fortunes by 1.4c on the week. The Overall Portfolio is now down by 8.7c on the season.
TIPS AND PREDICTIONS
Seven of the nine favourites won this weekend and that, combined with the generally low levels of disagreement amongst them, meant that the average result for Head-to-Head Tipsters was 6.8 from 9, the second-highest average for a full round this season.
A handful of Tipsters foresaw one of the upset results, the ProPred Tipsters, Easily Impressed I and Short Term Memory I all bagging 8 correct tips from 9. Worst were the 5 from 9s recorded by the WinPred Tipsters and Silhouette.
BKB, Bookie_3 and Combo_7 all scored 7 this week and remain joint-leaders, now on 43 from 63 (68%).
Surprising blowout victories by the Giants, Hawks and Eagles, and the upset Lions victory, all served to elevate Mean Absolute Error (MAE) performances this week, the all-Predictor average finishing at 36.5 points per game, the third-highest average MAE this season.
Best was Combo_NN2's 28.8 points per game, but this was sufficient only to drag it back into the middle of the MoS Leaderboard. That Leaderboard is now headed by RSMP_Weighted, which sits less than 2 points ahead of RSMP_Simple and just over 3 points ahead of a resurgent C_Marg. Combo_7, MoS' Official Margin Predictor, is in fourth position, less than 9 points from the lead.
Only four Predictors have profitable line betting records: C_Marg, Bookie_3, Combo_NN1 and Combo_NN2, the first two on that list still predicting line betting winners at an astonishing 60% rate of accuracy.
Bookie-OE returned the week's best log probability score amongst the Head-to-Head Probability Predictors, its score of +0.29 bits per game allowing it to retain top spot on the MoS Leaderboard. The two other Bookie Predictors also scored more than +0.20 bits per game, as did C_Prob, these solid performances lifting the all-Predictor average to +0.21 bits per game for the week, the second-highest average this season.
The Line Fund algorithm performed poorly for the round as a whole, its score of -0.15 bits per game its third-worst score this season. Still, it performed well-enough in the games in which it recommended wagers, so Investors were spared any serious consequences of its continued miscalibration.