Every so often I need a round like Round 16 to confirm my belief - or remind me of the obvious fact, you decide which - that no approach to wagering, bar its avoidance, is immune from outrageous fortune.

Based on my experience from this weekend I'm offering the management of TAB Sportsbet a suggestion for a new form of wagering, the reverse-multi, a wager that pays out only in the unlikely event that every leg of a punter's multi is unsuccessful. This weekend, we'd have landed our first reverse-multi.

How lucrative might that have been? Precisely estimating the odds of what we achieved is difficult, but let's see if we can come up with something reasonable. According to the logic of a line market, the five unsuccessful line bets were all 50:50 propositions. However, in the game where we had Richmond as the subject of a line bet, we also had them as a SuperMargin wager, failing to cover the spread but winning by 30 to 39 points. So, let's adjust the probability of losing both the bets in this game to 45%. In the two other games, where we had head-to-head bets, let's use the standard MAFL approach of estimating the probability using the wagered-on team's price on the sum of the two prices. We have then as the probability of landing our reverse-multi 0.5 x 3/4.4 x 0.45 x 0.5 x 0.5 x 0.5 x 1.55/4, which is 0.0074 or about 1 chance in 134. 

So, we just landed a 130/1 shot. Why don't I feel lucky right now?

All the weekend's bleeding cost us 17.5c, which leaves us now down by just over 19c on the season. Another round like that and 2012 will be MAFL's last year.

Quickly drawing a veil over the week's wagering then, let's move on to head-to-head tipping where we find the Head-to-Head Tipsters with a modal (ie most common) score of 6, an average score of 5.9, a best score of 7 - for Combo_NN1, Consult The Ladder and Easily Impressed II - and a worst score of 1, for Home Sweet Home. Yes, that's right: only one home team was successful this weekend, and that was the Roos in the first game of the round.

All the front-runners on the Tipping leaderboard snagged 6, so there's been no change at the head of the table where H2H_Unadj_10 still leads by one tip, now on 104 from 135 (77%).

Best amongst the Margin Predictors this week was Bookie_3 with an MAPE of 32.95 points per game, over 7.5 points per game better than the all-Predictor average for the week of 40.51 points per game. Worst was ProPred_7's 45.46 points per game. Bookie_3's strong performance allowed it to close the gap to Combo_7, who now leads by only 0.05 points per game. Bookie_9 remains in 3rd position, just 0.01 points per game further back.

The Margin Predictors were also on the wrong side of the line markets with most of their predictions this week, their average of 1.7 correct predictions from 9 a season-low performance by a considerable margin. They also missed out on most of the SuperMargin results, except in the Lions v Saints game where 7 of them selected the correct bucket. Overall, the broad pattern of season-long SuperMargin profitability across the Predictors changed little this week, save for the fact that Combo_NN_2 has returned to red ink.

It was also an unhappy round for all of the Head-to-Head Probability Predictors except the TAB Bookmaker. His were the only predictions to earn net positive probability scores. WinPred's performance, although net negative, was sufficiently superior to ProPred's for it to assume 4th spot on the ladder.

The Line Fund algorithm also did poorly, turning in its second-worst net score for a round of the season, better only than its result in Round 11.