Matter of Stats

View Original

2015 - Round 18 : A Recalibration?

After two weeks of fervent and, ultimately, futile and fruitless forays into the wagering markets, this week's lone line bet comes as a welcome pause. Though Investors won't be recouping any significant portion of their losses this week, some time for the Fund algorithms to reset and readjust to this season's rhythms is almost certainly not a bad thing.  

That solo line wager is on Port Adelaide who are giving the Saints just over four goals start in their game against them at the Adelaide Oval on Sunday.

The Head-to-Head Fund's complete inactivity is due, largely, to the paucity of opportunities afforded it, six home teams being ruled out from any possibility of a bet by the Fund because their price was under $1.50. That left the Fund only to weigh up wagers on a Carlton side at $6, a Gold Coast side at $7.50, and - probably the only genuine possibility at all - an Essendon side at $2.90. To be honest, I'm not at all sad it passed up on all three of those options.

A single bet makes the usual full Ready Reckoner seem overkill, so I've instead provided a Ready Reckoner Lite, which appears at right. It tells a very simple story: Port wins by 26 points or more and the Overall Portfolio rises in value by 2.7c; any other result and it falls by 3c.

HEAD-TO-HEAD TIPSTERS

When all nine of the favourites are priced at $1.42 and under, it's reasonable to expect a high level of agreement amongst the Head-to-Head Tipsters, and that's exactly what we see this week.

In fact, the 12% all-Tipster Disagreement Index is the second-lowest of the season, larger only than the Index for Round 14. But for the Heuristic Tipsters, the Index would have been lower still, with Easily Impressed II (41%), Home Sweet Home (38%), and Shadow (31%) in particular driving up this week's combined number.

Just two games see minority support groups bigger than a pair, the Port Adelaide v St Kilda game encouraging a band of five Tipsters to opt for an upset Saints win, and the Sydney v Adelaide game doing likewise for a band of seven Crows contrarians. None of the Tipsters at the top of the MoS Leaderboard have tipped against any of the favourites, so there'll be no change in their ordering at round's end.

MARGIN PREDICTORS

Levels of favouritism in head-to-head markets have no bearing on the extent to which the Margin Predictors can disagree about margins of victory, which is why the Margin Predictors have this week collectively been able to produce the highest average all-Predictor Mean Absolute Deviation (MAD) since Round 10, and the second-highest MAD of the season.

Combo_NN2 has done most to drive this average up, its MAD of 13.8 points per game almost 4 points per game higher than any other Predictor. It is Predictor Most Extreme (PME) in five contests, tipping the Hawks by 43, Pies by 40, Eagles by 23, Power by 39, and Dockers by 57.

Next highest MAD belongs to ProPred_7 at 9.7 points per game (it's PME in 4 games), and then to C_Marg at 9.5 points per game (it's PME in 2). Low MAD of 2.3 points per game belongs to Combo_7, making it now six times that it's attained this position this season. During the course of the entire season, Combo_7's MAD has never exceeded 4.2 points per game in any single round. It truly is - as it was designed to be - a broad indicator of MoS Margin Predictor opinion.

At the game level, it's the round's first and last contests that are sparking the broadest range of opinions, Friday's Hawks v Tigers game producing a 13.0 points per Predictor MAD, and Sunday afternoon's Dockers v Giants game an 11.5 points per Predictor MAD. In sharp contrast, the Predictors are much more aligned about the likely victory margin in the Swans v Crows game, the MAD coming in at just 2.2 points per Predictor, and the entirety of the predictions spanning just a 13 point range from a 10.5 to a 23.4 point Swans win.

Thinking about the current MoS Leaderboard, the predictions that are most likely to alter the relative positions of the top 3 are:

  • Geelong v Brisbane: C_Marg +59 / RSMP_Weighted +50 / ENS_Greedy + 49
  • Carlton v Kangaroos: -45 / -24 / -36
  • Gold Coast v West Coast: -41 / -34 / -46

PROBABILITY PREDICTORS

The Head-to-Head Probability Predictors, in terms of disagreement levels, have steered a course this week somewhere between the low levels of disagreement amongst the Head-to-Head Tipsters and the high levels amongst the Margin Predictors.

They've returned an all-Predictor average MAD of 4.7% points per game per Predictor, which is lower than it's been for the past two rounds, but higher than the four rounds prior to that.

No single Predictor has demonstrated truly extreme behaviour - gone MAD mad, if you like - with ProPred's 6.1% points per game the highest MAD recorded. C_Prob, at 5.7% points per game, has the next-highest MAD. Low MAD, very narrowly, belongs to Bookie-OE, fractionally ahead of Bookie-RE.

Only one game has elicited a MAD above 7% points per Predictor, this the Hawks v Tigers game where it's 9.6% per Predictor and where the predictions span a 24% point range from 63% to 87%. The Blues v Roos game has a MAD of just 1.1% points per Predictor, and a range of just 5% points spanning 11% to 16%.

If C_Prob is to make further advances on the MoS Leaderboard it most needs Adelaide, Collingwood and Geelong to prevail since these are the games where its probability assessments (from a log probability score perspective) are most different from the three Bookie family predictors. 

The Line Fund algorithm feels that the Gold Coast are receiving nowhere near enough start and, accordingly, assesses their opponents, West Coast, as 69% chances to win in the line market. Other teams it believes enjoy greater than 55% prospects are GWS (62%), Brisbane Lions (59%), Melbourne (58%), and Port Adelaide (57%).