2015 - Round 17 Results: No Better; No Bettor

When I originally conceived the idea to separate the more technical of MoS blog posts from those that were more applied by establishing a Statistical Analysis journal for the former and a Wagers & Tips journal for the latter, the purpose was to allow visitors to choose the types of posts they preferred to read. What I now appreciate after this season is that the demarcation also serves to shield the credibility of those more technical posts from the stark reality of what happens when I apply their conclusions.

It wasn't until Sunday this week that MoS Investors enjoyed their first return, a late, cruel goal from Fremantle having deferred their gratification from the previous evening. That Sunday collection was from a win by the Dogs on line betting and was joined late on Sunday by the second and final collect of the weekend, which was from a head-to-head bet on the Eagles.

Those two collects were sprinkled lightly through a base of seven losses, five in the head-to-head market and two in the line market. In total, the round dropped the Head-to-Head Fund by about another 14c and the Line Fund by about another 5.5c, which together stripped another 9c off the Overall Portfolio. It now stands - well, totters really - down 36c on the season.


The Head-to-Head Tipster results were very mixed this week, three Tipsters (C_Marg, Combo_NN2 and CTL) scoring 8, and one Tipster (Home Sweet Home) scoring 3. The all-Tipster average of 5.7 tips was the lowest since Round 12, and the 6th-lowest all season. 

Combo_7's 5 from 9 performance was enough to see it head the MoS Leaderboard on 102 from 143 (71%), just half a tip ahead of Bookie_3 (who got his half-tip from his draw prediction this week), and a whole tip ahead of BKB and Bookie_9. 

Consult The Ladder's 8 from 9 score this week has left it just 6 tips from the lead, which is fairly remarkable when you realise that it's an Heuristic Tipster whose sole rule is "tip the team that's higher on the ladder".

The Margin Predictors' performances split naturally into two groups this week, one group - those at the top of the MoS Leaderboard - registering Mean Absolute Errors (MAEs) between 26.5 and 35.5 points, and the other group registering MAEs between 37.4 and 42.3 points.

C_Marg recorded the week's best MAE of 26.5 points per game, which left it over 50 points clear of the field at round's end. RSMP_Weighted, whose MAE was over 5 points a game worse than C_Marg's, nevertheless retained 2nd spot, while ENS_Greedy slipped into 3rd.

The all-Predictor average for the round of 35.7 points per game per Predictor was the highest since Round 12 and the 6th-highest of the season. This relatively poor margin prediction translated into poor line betting performances too, leaving only four Predictors now with a season-long profitable record: C_Marg, RSMP_Simple, Bookie_3, and Combo_NN2.

Head-to-Head probability prediction was also sub-par, the all-Predictor average log probability score of +0.16 bits per Predictor per game the lowest since Round 12 and the 8th-lowest all season. C_Prob registered the week's highest score, and lifted itself into 3rd position on the MoS Leaderboard as a result, nudging Bookie-LPSO into 4th. At the top of that Leaderboard, Bookie-RE has a slender lead over Bookie-OE. If only the Head-to-Head Fund has been using C_Prob for its decision-making this season rather than the H2H algorithm.

The Line Fund algorithm returned a marginally negative log probability score for the week, its 6th in the past 8 weeks.

I'm not at all sorry that Round 17 is over.