Well the streak is over - regression to the mean and all that - but, in truth, I'm not all that much discouraged by the weekend's wagering results.
Granted, the Recommended Portfolio shed 2.7c this week, its first loss in six rounds, but Collingwood's victory by 40 points was only a point more than a result that would have landed Investors a $13 wager, Freo failed by just 3 points to cover its 24.5 point spread, and Port Adelaide's 9.17 to the Lions' 9.8 was only a smidge more than a goal shy of turning yet another line wager into a winner. As Ned Kelly (might have) said, "such is life".
The Head-to-Head Fund, with its sole wager on the Dees, was the only Fund to enjoy zero success this weekend, as the Line Fund, its unfortunate wagers on Freo and Port aside, landed 2 from 5, and the Margin Fund secured collects in 3 wagers from 10.
That leaves Investors now up by a tick under 10c on the season.
TIPSTERS AND PREDICTORS
Yet again, favourites predominated amongst the weekend's victors, the Eagles interposing themselves as the sole underdog victors in a room full of the teams more-likely.
That meant the majority of sane Head-to-Head Tipsters landed 8 tips from 9, and did nothing to separate those at the top of the MAFL Leaderboard. Eight Tipsters, consequently, find themselves tied for 1st on 115.5 from 153 (76%). Much further down the Leaderboard, Short Term Memory II and Shadow both tipped the card while Follow the Streak and Easily Impressed I recorded the weekend's low, bagging just 5 from 9.
Amongst the Margin Predictors, Combo_NN2 recorded the week's best mean absolute prediction error (MAPE) with an MAPE of 26.2 points per game, made more impressive by the fact that it was recorded in the face of two blowout victories by the Roos (122 points) and the Cats (101 points). Sometimes it pays off to be the Tipster Most Extreme.
This creditable performance was, however, insufficient to elevate Combo_NN2 on the MAFL Leaderboard, which saw no changes in the all-season ranking of any of the top 8 Predictors.
It's interesting to note that four of the bottom five Margin Predictors now have season-long line betting records sufficient to have turned a profit on line betting at $1.90 across the season to date. Once again we find that lack of success in one domain (predicting victory margins) by a Predictor does not preclude it from wagering success in another.
On head-to-head probability prediction, the weekend's results also precipitated little movement - in fact, none at all, with all Predictors enjoying strongly positive probability scores. ProPred's results were the best of all, though not good enough to move it from 4th on the Leaderboard.
The line fund algorithm, meantime, recorded a mildly negative probability score, its first since Round 15 and only its fifth in the last 10 rounds.
Win_7 and Combo_NN2 both turned in the round's best SuperMargin performances, each selecting the correct bucket in two games, though each doing so in different contests. Combo_NN2's overall performance shaded Win_7's, however, since it was also in error by a single bucket in two more games.
Seven Predictors now have a positive ROI for the season-to-date, foremost amongst them ProPred_3, which sports an ROI just over 20%. If we focus our attention on Predictor performance only when they foretell a home team victory, Combo_NN2 is rated Number 1 with an ROI approaching 50%, which is, of course, great news for Investors. Bookie_9, Investors' other Predictor of Record, has been much less impressive but is, at least, showing an only mildly negative ROI.