Matter of Stats

View Original

2020 - Round 21 (Week 3 of the Finals) - MoSHBODS Changes (Is Out Of?) Its Mind

Regular readers might recall that, last week, I explained the intricacies of how the MoS twins come up with Venue Performance Values (VPVs), which are their versions of Home Ground Advantage.

There, I mentioned that teams are assigned a default VPV at venues until such time as they’ve played enough games at that venue for an individualised VPV to be estimated. Last week, the Cats had played only 7 games in the window that MoSSBODS and MoSHBODS consider in their VPV calculations, which resulted in both forecasters tipping an upset Pies win, And we all know how well that went …

Well, this week the Cats have now played enough games at the Gabba for MoSHBODS to grant them an individualised VPV, but not enough games for MoSSBODS to do likewise. Relative to MoSHBODS’ expectations, the Cats have performed extremely well at the Gabba, which means we’ve wound up with MoSSBODS using a default VPV for the Cats and tipping the Lions by 7 points, and MoSHBODS using a result-informed VPV for the Cats and tipping them by 12 points.

If any year was going to highlight and underline edge cases in your computer scripts, this was certainly it.

Again, though, it’s a MoS principle not to change the assumptions made pre-season unless they are demonstrably in error, so we’ll be sticking with those numbers as the official forecasts this week. For next season, however, I’ll probably be moving to smooth the VPV calculations such that, prior to a team playing the required number of games at a venue, they’ll be a blend of the default and a result-informed calculated value, with the latter becoming progressively more heavily weighted as more games are played.

You can see what the other MoS models think in the table at right, which shows that the MoSHBODS contrarian tip is joined by contrarian tips from the RSMP twins in both games (which is itself something of an oddity).

Amongst the Margin Predictors we have a 14-point range in the Power v Tigers game (thanks to MoSHBODS_Marg and the RSMP twins), and a 19-point range in the Lions v Cats game (thanks to the MoS twins).

In both games, eight of the nine Predictors have forecast margins of seven points or less. Were it not for MoSHBODS_Marg, it would be a clean sweep.

On the Leaderboard, MoSHBODS_Marg currently trails Bookie_9, who’s in top spot, by about 14 points. The absolute difference between MoSHBODS_Marg’s and Bookie_9’s forecast margins is about 29 points, so some favourable results might see MoSHBODS_Marg atop the ladder at the end of the round. MoSSBODS_Marg, who trails Bookie_9 by about 26 points, differs in its forecasts by only 2 points.

Amongst the Head-to-Head Probability Predictors the probability estimates span almost 20% points in the Power v Tigers game, and almost 25% points in the Lions v Cats game.

RECENT FINALS HISTORY

The table at right below summarises the results from the last 20 years of men’s AFL Semi Finals based on the final home-and-away ladder positions of the teams, including the results from last week.

The dominance of teams with the double chance is very evident. Collectively, they’ve won over 80% of the Semi Finals they’ve played in, with no single ladder position having a winning percentage below 73%

Next, let’s turn our attention to the Preliminary Finals ahead, where we have 1st vs 3rd, and 2nd vs 4th this year, the 8th instance of the former, and the 7th of the latter.

Teams finishing 2nd have must the better set of outcomes of those two pairings, having won all six of their previous appearances against teams finishing 4th. In contrast, minor premiers have won only four of their seven contests against teams finishing 3rd.

Lastly, taking a whole-of-Finals approach, the table at left reveals how teams have fared across all weeks of the Finals, and shows the still superior records of teams from 1st, 2nd, and 3rd, and the parlous record of teams from 4th.

Between them, the teams finishing Top 3 in the home-and-away season have won all but one of the last 20 Flags.

WAGERS

With Sportsbet (aka, on MoS, BetEasy) making the Power v Tigers game a pick ‘em and therefore choosing not to frame a line market, options were somewhat curtailed this week.

In the end, the only wager was a 2.2% bet on Port Adelaide at $1.90 head-to-head.

Should that wager be successful, the Combined Portfolio will increase in value by 2.2% x 0.9 x 30%, which is 0.6c.

A draw will, instead, see that Portfolio decline by 0.03c, and a Port Adelaide loss will see it decline by about 0.7c.

Finally, here are the details behind the MoS twins’ forecasts