2017 - Round 23 : Here Already?

/

It's genuinely hard for me to believe that we're at the end of the home-and-away season already with the equivalent of only two regular rounds worth of games to be played.

More unbelievable is the fact that MoSHBODS has spawned such effective forecasters in its first year, MoSHBODS_Marg and MoSHBODS_Prob outforecasting the TAB forecasters in all three MoS contests. It is though, of course, only the first year, and almost any forecaster can do well over a couple of hundred forecasts. Repeating the performance in 2018 would give it much more credibility.

Anyway, to Round 23 then where the TAB bookmaker sees a few potential blowouts and, as a result, has set an average handicap of almost 21 points per game, which is the highest since Round 19. That average includes three games with handicaps of 6 goals or more.

In only one of the past six seasons, however, has the final round seen a lower average handicap. That was the 2013 season where the average was just one-tenth of a point lower.

The Head-to-Head Tipsters have responded with only slightly above-average levels of disagreement, the bulk of it coming from Consult The Ladder, Home Sweet Home, and C_Marg.

Home Sweet Home has gone contrarian in three games, Consult The Ladder and C_Marg in only two. Combined, they represent 7 of the 12 tips for underdogs this week.

Four of those underdog selections belong to Collingwood, a victory by whom would see MoSSBODS_Marg regain top spot on the Leaderboard. A Pies loss would see the two RSMP Tipsters claim that lead.

The Margin Predictors have a collective mean absolute deviation (MAD) of 3.6 points per Predictor per game this week, which is slightly above last week's figure but below the all-season average. C_Marg has the highest MAD (6.0 points per game), ahead of MoSSBODS_Marg (4.6) and MoSHBODS_Marg (4.4). This season has been a fascinating case study in how moderate levels of divergence such as we've seen from MoSHBODS_Marg and MoSSBODS_Marg can be very beneficial in a Tipster, while extreme levels such as we've seen from C_Marg can be catastrophic. 

As evidence of that claim, across the season so far, C_Marg has the highest mean MAD but also the highest (ie worst) MAE, while MoSHBODS_Marg lies 3rd in terms of average MAD but sits comfortably atop the Leaderboard on MAE.

On which topic, MoSHBODS_Marg leads the two RSMP Predictors on that Leaderboard by about 29 points, and will see that lead change depending on the outcomes of each game roughly as follows:

  • Increase by 4 points over the RSMP Predictors if the Western Bulldogs win by 7 points or more
  • Increase by 8 points over the RSMP Predictors if Collingwood win by 3 points or more
  • Increase by 10 points over the RSMP Predictors if the Brisbane Lions win by 9 points or more
  • Increase by 11 points over the RSMP Predictors if Sydney win by 49 points or less
  • Increase by 1 point over the RSMP Predictors if Geelong win by 3 points or more
  • Increase by 11 points over the RSMP Predictors if Port Adelaide win by 43 points or less
  • Increase by 2 points over the RSMP Predictors if Essendon win by 40 points or more
  • Increase by 4 points over the RSMP Predictors if Richmond win by 10 points or less
  • Increase by 5 points over the RSMP Predictors if Adelaide win by 18 points or more

Relative to recent weeks, there's quite a lot of potential movement across those nine games (56 points in all).

Looking from a game-by-game perspective, only three games carry a MAD of greater than 4 points per Predictor: Collingwood v Melbourne (5.2), Sydney v Carlton (5.2) and Brisbane Lions v Kangaroos (4.2). Even in those games the forecasts only span a 2 or 3 goal range.

The Head-to-Head Probability Predictors have an Overall MAD for the round of 3.2% points per Predictor per game, which is the lowest since Round 19 and a full 1% point below the all-season average. C_Prob has the largest MAD, ahead of MoSSBODS_Prob and MoSHBODS_Prob.

There is though an unusually high MAD for the Pies v Dees game where the probability estimates for the Pies start at 42% and range up to 60%. The MAD for that game is 7.8% points per Predictor.

MoSHBODS_Prob has the Pies at 55% in that game and a loss by them will therefore be beneficial to the bookie-based Predictors whose estimates are all around the 42% mark. That's the only game where there is a significant difference between MoSHBODS_Prob's and the bookmaker's estimates.

WAGERS

This week sees three head-to-head and four line wagers totalling about 2.8% of original Combined Portfolio funds.

That's a higher proportion than last week but still quite subdued by the standards of this season.

About two-thirds of the total upside and downside for the round belongs to Collingwood. A victory by them would add almost 2c to the value of the Overall Portfolio, while a loss by 9 points or more would subtract about the same amount.

In total, the right result in the five games we've wagers on would lift the value of the Overall Portfolio by 2.7%.

MoSSBODS AND MoSHBODS

Here are the MoS twins' views on team and total scoring for the week.

Commentary to follow.