Matter of Stats

View Original

2016 Round 10 : Overs/Unders Update

Investors were already hoping that rain might boost the Lions' and Dons' chances in upsetting or at least staying close to their respective opponents, and today's news of six Unders wagers for the weekend will only reinforce that pluvial desire.

Those six wagers come with a wide range of Totals ranging from a low of 180.5 for the Carlton v Geelong game to highs of 200.5 in both the Brisbane Lions v Hawthorn, and West Coast v Gold Coast games.

Placing that Unders bet on the Lions v Hawks game was a curious affair. My first attempt to place it when the price was still showing as $1.90 was rejected by the TAB, apparently because they had reached their "limit" on this particular wager. I was invited to "try again later", which I did a few minutes afterwards only to be informed that the price for my bet had been adjusted to $1.85. At the same time as I placed this bet I noticed that the line market price for the Lions had also fallen 5c from $1.80 to $1.75. A couple of minutes after that all markets on the game had been suspended. Odd.

(They've since reopened with substantial adjustments: the Lions are now +42.5 on the Line Market and $7.00 Head-to-Head.)

Anyway, the six Unders wagers by the Overs/Unders Fund represents the largest single-round activity for that Fund since the seven wagers it made in Round 4 and takes the total Overall Funds at risk to 7.8% for the week. That's roughly the same amount as has been risked over the past five weeks.

MoSSBODS COMPARISON TO THE TAB

Comparing MoSSBODS with the TAB this week yields a familiar narrative, MoSSBODS, for example, convinced that the round will produce, on average, 174 points per game and the TAB convinced the average will be 14 points per game higher.

The differences are greatest in the six games in which the Overs/Unders Fund has wagered, with MoSSBODS predicting aggregates 15 to 24 points lower for those games. In the three other games the Fund hasn't wagered, but MoSSBODS is still projecting aggregates slightly lower than the TAB's projections.

MoSSBODS disagrees with the TAB not just about game totals, but also about which of the games will be the highest- and lowest-scoring. MoSSBODS has the Adelaide v GWS matchup as the round's high-scoring game, but the TAB has the West Coast v Gold Coast game, and MoSSBODS has the Carlton v Geelong contest as the low-scoring game, while the TAB opts for the Essendon v Richmond game.

Looking at individual team scores, MoSSBODS has 15 of the 18 teams scoring fewer points than does the TAB, especially:

  • Hawthorn: MoSSBODS 103, TAB 125
  • Carlton: MoSSBODS 52, TAB 71
  • Melbourne: MoSSBODS 83, TAB 99
  • Richmond: MoSSBODS 95, TAB 109
  • West Coast: MoSSBODS 129, TAB 142

Only the Lions, Dons and Giants are expected to score more points by MoSSBODS than they're expected to by the TAB.

So far this season, MoSSBODS' performance compared to the TAB, measured by mean absolute error (MAE) per game, has been:

  • Home Team Scores: MoSSBODS 22.3, TAB 19.4
  • Away Team Scores: MoSSBODS 19.7, TAB 20.8
  • Either Team Score: MoSSBODS 21.0, TAB 20.1
  • Total Scores: MoSSBODS 30.5, TAB 28.1
  • Game Margins: MoSSBODS 30.0, TAB 30.5

Treating each of the nine rounds separately MoSSBODS has produced a smaller MAE than the TAB:

  • Twice for Home Team Scores
  • Six times for Away Team Scores
  • Three times for Either Team Score
  • Twice for Total Scores
  • Five times for Game Margins

That's a fairly consistent story then: MoSSBODS for Away Team Scores and Game Margins; the TAB for Home Team Scores and Total Scores. There will be tinkering in the off-season, doubtless ...