2011 Round 24 Results: Good for the Competition; Bad for Investors
/If you're an Investor, I can only hope that you read and independently acted on the Simulations blog, in particular this entry from 2 weeks ago in which I noted the value in the Suns for the Spoon at $3.75, because you certainly didn't get much value from your Portfolio investment this weekend.
Our collective fate was largely sealed by the Pies' lacklustre performance on Friday night when, after leading early by 24 to 5, they switched off, registering just 9 more scoring shots in the game to the Cats' 34, thereby leaving the cash from our head-to-head and line bets firmly in the pocket of the TAB bookmaker.
The Dogs' failure - by one-half a point no less - to cover the spread, and the Blues' poor second-half against the Saints didn't help matters. These two additional losses, offset though they were by the Eagles' thumping win over the Crows and by Sydney's solid victory over the Lions, were enough to drop Portfolios by another 8c across the round.
That's four straight rounds of losses - an unusual sequence for this time of year, which has traditionally been profitable for MAFL - and leaves us down by 22.5%. It's hard to see us making up that deficit over the remaining 9 games of the season, especially since we'll almost certainly be wagering on home team favourites if we're wagering at all.
So, who's to blame? (Aside from me.)
If we apportion blame to teams firstly when they've failed to win on cue, then the Dogs, Dons, Tigers, Suns and Lions deserve the largest dollop based on their head-to-head performances. Combined these five teams represent the entirety of the losses that the Head-to-Head Fund has suffered so far this season.
Only four teams have generated a profit for Investors on head-to-head wagering when they've been wagered on: Collingwood, Hawthorn, the Roos and West Coast.
On line betting, it's the Dons, Tigers and Dogs which are also major contributors to the losses, and it's Collingwood and Hawthorn which have in this form of wagering too made significant positive contributions. The Cats have also helped (though not this week).
Allocating blame to teams instead when they've failed to lose on command, it's now Carlton, West Coast, Melbourne, Fremantle and Sydney that deserve our opprobium for their head-to-head performances, and the Dogs, Adelaide and Essendon that merit our applause.
As far as line betting is concerned, only Geelong is responsible for more than 10c worth of losses, while the Dogs and the Saints have each added 10c or more to the Line Fund's price.
Onto tipping then, where we find that victories by only 5 of the 8 favourites dragged the average performance down this week. At just 4.8 from 8 it was only the 19th-highest average for the season. The round's best score of 6 went to Combo_NN_1, which was enough to move it into a share of the lead with BKB and Bookie_9 on 141.5 from 187 (76%), one tip ahead of Combo_NN_2 on 140.5 (75%).
The end of the home-and-away season signals the end of tipping by the Heuristic Tipsters. Collectively - BKB aside - they've not had an outstanding year, the best amongst them being Consult The Ladder which tipped 134.5 correct results during the season, a 72% result. In other words, the most effective heuristic this season apart from tipping the favourite was to tip the team higher up on the ladder in each contest.
Home Sweet Home fared worst amongst the Heuristics, recording only a 103.5 from 187 (55%) performance, further evidence of the decline of the home ground advantage and part explanation of the poor performance of the Head-to-Head Fund this season.
Notwithstanding the unremarkable tipping performances of the Heuristic Tipsters and the lower-than-average winning rate for home teams, the simple expedient of wagering only on Heuristics' home team victory predictions would have yielded a profit for 5 of the Tipsters, and commencing such a strategy in Round 6 would have nudged one more of them, BKB as it happens, into black ink.
Ride Your Luck would have produced the greatest ROI - just over 10% had its home team predictions been wagered on commencing in Round 6. Even an impatient punter, wagering on Ride Your Luck's home team predictions from the start of the season, would have enjoyed a 7% ROI.
Whilst we bid farewell to the Heuristic Tipsters for another year, the remaining Head-to-Head Tipsters will continue prognosticating throughout the Finals.
So too will the Margin Predictors, which this week saw their MAPEs particularly savaged by the surprisingly large victory margins of the Cats and the Eagles. The all-Predictor average for the Round of 34.45 points per game was the 15th-highest of the season.
Combo_NN_1 performed best, recording a 30.08 points per game result, almost 3.5 points per game better than any other Predictor from the top 5 on the table. Combo_7 was next best on 33.54 points per game, ahead of Combo_NN_2 on 34.72 points per game and Bookie_9 with 36.39 points per game. Bookie_3, which fared worst of all amongst the leading Predictors with a 37.64 points per game result, nonetheless continues to head the MAPE Leaderboard and remains the only Predictor with a season-long sub-30 MAPE. It's now on 29.73 points per game for the season, ahead of Combo_7 on 30.37 points per game and Bookie_9 on 30.92 points per game.
The average line betting performance was a surprisingly high 5.4 from 8 this week, the 2nd-highest average for a round this season, once again underscoring the general lack of an association between margin prediction and line betting prowess. In fact, the season-long correlation between the line betting and MAPE performances of the Margin Predictors now stands at +0.16, meaning that better line betting performances are, weakly, associated with poorer (ie higher) MAPEs. As well, the Predictors with the two best line betting performances this season are ranked 1st and last on the MAPE Leaderboard.
We also find that stronger line betting performances are associated with poorer head-to-head tipping results: the correlation of these two measures is -0.30. So, as curious as it seems, it continues to be the case that an ability to correctly identify winning teams and to accurately assess victory margins is insufficient to produce line betting success. Exactly why this is the case is something I want to explore in future blogs.
It is the case, however, that predictors that do well on head-to-head tipping also tend to do well on margin prediction - the correlation here is -0.74 - so the universe is not entirely perverse.
Finally then, to the Probability Predictors.
It was a much better week for the Head-to-Head Probability Predictors as they all recorded comfortably positive probability scores. Best amongst them was the TAB bookmaker, who continues to lead all-comers.
The Line Fund algorithm did not fare as well, hurt especially by the failure of the Pies to cover the 20.5 point spread they were offering the Cats, a result that the algorithm assessed as having only a 24% probability of occurring.