For Investors, there's very little action and no joy at all to report for the round just completed. Our lone, hopeful wager on the Dees survived contact with reality for about a quarter before disappearing in a flurry of second-quarter Magpie goals and behinds. Over the final three quarters of Monday's game the Dees were outscored 16.14 to 4.4.
Still, the wager was small, and the Head-to-Head Fund is only weighted 30% in the Recommended Portfolio, so the net result was just a 0.4c loss.
TIPSTERS AND PREDICTORS
Of more interest was the ongoing battle amongst the various families of Tipsters and Predictors.
Five of the six favourites were victorious over the extended weekend, so it's little surprise that the round's best-performed Head-to-Head Tipsters were those that aligned themselves with Bookie Knows Best. Most of the leading Tipsters though found reason to disagree with BKB on at least one tip, and so finished with just four correct predictions.
As a result, Combo_NN2 remains the leader - by two tips - from seven pursuing Tipsters. At the other end of the ladder, Home Sweet Home had yet another horrid round, bagging just 2 from 6 to leave it floundering at under 50% for the year - probably not such a bad round then for the Head-to-Head Fund to be circumspect.
The RSMP Margin Predictors continue to lead their peers, with RSMP Weighted atop the table and now 0.38 points per game ahead of RSMP Simple. To help understand the practical significance of the differences in the performances of the Margin Predictors I've added a new column to the Leaderboard this week, labelled "Gap", which records the gap, in points, between the best Predictor and the Predictor in question.
So, for example, in order for RSMP Simple to overtake the leading Predictor, RSMP Weighted, it needs to make margin predictions that are, in total and in absolute terms, more than 36.1 points closer to the actual game margins than RSMP Weighted.
To give you a feel for this "Gap" measure, consider the fact that the difference between the best and worst performances this week was about 16 points per game, or about 96 points in total: Combo_NN2 recorded a mean absolute prediction error (MAPE) of 21.96 points per game, while ProPred_7 racked up an MAPE of 36.84 points per game. Or, as another means of contextualising the "Gap" measure, note that the average absolute difference between the margin predictions of RSMP Weighted and Bookie_3 has been about 7.2 points per game this season. Based on that, you might claim that RSMP Weighted is currently about 5.5 games ahead of Bookie_3.
RSMP Weighted and RSMP Simple continue to be above-average predictors of line results too, each of them selecting the winner is about 59% of games. No other Predictor has a better than 55% record, even if you adopt a contrarian approach to their predictions (ie predict the team they're tipping to lose on line betting).
Amongst the Head-to-Head Probability Predictors there's been little movement on the Leaderboard except that WinPred has now dragged itself ahead of the H2H Predictors. The Risk-Equalising variant of the TAB Bookmaker prices continues to lead all-comers. All of the Head-to-Head Probability Predictors recorded positive probability scores for the week.
The Line Fund algorithm recorded a very marginally negative probability score for the round leaving it with a slightly negative score for the season to date.
Had you decided to wager in the SuperMargin market in every game this season based on the opinions of a particular Margin Predictor, your best choices would have been to base your activity on either of the ProPred family, each of which is showing ROIs of around 30%.
Combo_NN2 would have been your next best selection, followed by the WinPred family. Bookie_9 however, most notably for Investors, wouldn't have been in your top 10.
This week, Combo_NN2 selected the correct bucket in two games and was in error by a single bucket in one more. Unfortunately for Investors, this prescience was exercised in games with away team victors, and so did not attract the Margin Fund's attention.