2015 - Round 2 Results: Margin MAE-hem

If you've had a poor round's wagering and tipping, I think you're due an appalling pun in your blog post title.

MAE-hem it was though, as just four of nine favourites grabbed the competition points this weekend, and the victory margins in five games exceeded 40 points.

Worse, as far as Investors were concerned, was that neither of the anointed home teams did what they were asked head-to-head, and that only two of the six enwagered home teams emerged victorious on Line betting. As a result, the Head-to-Head Fund shed just under 5c and the Line Fund just under 3c, shaving about 3.5c off the Portfolio price to leave it down by a little under 2.5c on the season. The relative caution of the two Funds, it turns out, was well-advised.


You know it's been an odd season, based on a review of recent years, when you find Heuristic Tipsters occupying most of the upper rungs on the Head-to-Head Tipping Ladder.

This week, 10 of them bagged the round-high five from nine leaving them tied on 11 from 18 with Combo_NN2, all of them one tip ahead of Head-to-Head tipping royalty, BKB.

All told, the Head-to-Head Tipsters averaged just over 4 correct tips from 9 this week, down from 6.1 from 9 in Round 1.

The Margin Predictors also struggled with the round's unexpected results as only Combo_NN1, which now leads all Margin Predictors, recorded a sub-40 MAE. It sits in 1st position almost 50 points clear of Combo_NN2 having been nearest the pin in fully one-third of all contests. At this point in the season I'll admit to feeling a certain pride that the two Combo_NNs and the two RSMP Predictors, all MoS alumni, fill positions 1 through 4 on the Leaderboard, and that C_Marg sits in 6th. In my empirical, data-driven heart I know that this is unlikely to last, especially with Bookie_9 looming.

Overall this week the Margin Predictors recorded an average MAE of 44.2 points per game per Predictor, which is quite a lot worse than last week's average of 24.4 points. MAE-hem indeed.

I'm also inordinately pleased that C_Prob sits atop the Probability Predictors' Leaderboard, albeit with a negative Log Probability Score for the season so far. All three of the more directly TAB Bookmaker-based Probability Predictors are filling C_Prob's mirrors, however.

Less pleased I am, however (to channel Yoda) with the Line Fund algorithm's Log Probability Score of -0.1326 bits per game. Small wonder then that the Line Fund finds itself down by 2c with a 6 and 7 record.