MoSHBODS' Forecasting Record in Grand Finals

MoSHBODS has tipped the Tigers to win by about 16 points on Saturday, but Giants supporters shouldn’t be too concerned about that, because MoSHBODS’ record in Grand Finals since 2000 isn’t particularly impressive, as you can see from the table below.

2000 to 2018

Giving it half-credit for the Pies v Saints draw in 2010, it’s only tipped the winner 48% of the time, and has been wrong for the last three Grand Finals. The mean absolute error (MAE) of its forecast final margins has been about 33 points per game.

Some of the records for this era (all of which are also all-time records) are:

  • Highest rated team: Essendon 2000 (+44.5)

  • Highest aggregate Ratings: Geelong v Collingwood 2011 (+78.6)

  • Highest forecast margin: Geelong v Port Adelaide 2007 (34.2 pts)

  • Highest absolute error of forecast: Geelong v Port Adelaide 2007 (84.8 pts)

  • Biggest Upset (as measured by expected victory margin for loser): Sydney v Hawthorn 2012 (20.8 pts)

Other era records are:

  • Lowest rated team: Collingwood 2002 (+5.2)

  • Lowest aggregate Ratings: Richmond v GWS 2019 (+27.6)

One notable feature of this era is the fact that the average defensive ratings of winners and losers have been about equal, whereas winning teams have been, on average, about 4.5 points stronger offensively than losers. It has so far produced only 172 points per game, and an average victory margin of 32.5 points per game.

1980 to 1999

Next, we look at the era from 1980 to 1999, where MoSHBODS’ accuracy improves considerably.

In this era, MoSHBODS tips 15 of the 20 winners, despite an average forecasted winning margin of just 9.7 points per game, which is more than 4 points lower than for the 2000 to 2018 era. MoSHBODS MAE for the era is just over 7 goals per game.

Some of the records for this era are:

  • Highest rated team: Hawthorn 1991 (+36.4)

  • Highest aggregate Ratings: Carlton v Hawthorn 1987 (+66.4)

  • Highest forecast margin: Hawthorn v Melbourne 1988 (27.3 pts)

  • Highest absolute error of forecast: Hawthorn v Essendon 1983 (79.2 pts)

  • Biggest Upset (as measured by expected victory margin for loser): Adelaide v North Melbourne 1998 (15.2 pts)

  • Lowest rated team: Melbourne 1988 (+3.0)

  • Lowest aggregate Ratings: Essendon v Carlton 1993 (+24.1)

In this era, winning and losing teams had roughly equal offensive ratings, but winning teams had defensive ratings, on average, almost 5 points higher. This era also produced over 200 points per game in Grand Finals, and an average victory margin of 47 points.

1960 to 1979

Now we move to the 1960 to 1979 era, where MoSHBODS’ record improves in terms of MAE, but deteriorates in terms of accuracy.

Again giving half-credit for the 1977 draw, MoSHBODS’ accuracy comes in at 64% for the era, but its MAE is a more impressive 23.5 points per game, albeit in a lower scoring era than 1980 to 1999.

Some of the records for this era are:

  • Highest rated team: Richmond 1967 (+33.1)

  • Highest aggregate Ratings: Hawthorn v St Kilda 1971 (+56.3)

  • Highest forecast margin: Melbourne v Collingwood 1964 (24.5 pts)

  • Highest absolute error of forecast: North Melbourne v Hawthorn 1975 (51.4 pts)

  • Biggest Upset (as measured by expected victory margin for loser): Carlton v Collingwood 1970 (14.7 pts)

  • Lowest rated team: Collingwood 1964 (+3.9)

  • Lowest aggregate Ratings: Essendon v Carlton 1962 (+17.7)

In this era, winning teams were roughly 2 points stronger offensively, and 1 point stronger defensively than the teams they defeated. This era produced 178 points per game in Grand Finals, and an average victory margin of just under 24 points per game.

1940 to 1959

MoSHBODS had greater accuracy in the previous era, 1940 to 1959, but a higher MAE.

MoSHBODS’ accuracy for the era is just under 70%, bolstered by tipping 4 of the last 5 winners of the era, and its MAE is just under 31 points per game.

Some of the records for this era are:

  • Highest rated team: Melbourne 1955 (+27.5)

  • Highest aggregate Ratings: Essendon v North Melbourne 1950 (+47.7)

  • Highest forecast margin: Melbourne v Collingwood 1955 (23.4 pts)

  • Highest absolute error of forecast: Essendon v Carlton 1949 (63.5 pts)

  • Biggest Upset (as measured by expected victory margin for loser): Collingwood v Melbourne 1958 (17.8 pts)

  • Lowest rated team: Collingwood 1958 (+2.7)

  • Lowest aggregate Ratings: Collingwood v Melbourne 1958 (+19.8)

In this era, winning teams were roughly 1.4 points stronger offensively, and 1.5 point stronger defensively than the teams they defeated. This era also produced only 163 points per game in Grand Finals, and an average victory margin of just over 34 points per game.

1920 to 1939

MoSHBODS had another poor record on accuracy in the 1920 to 1939 era, but a relatively low MAE, helped by the lower rates of scoring during the era.

MoSHBODS’ accuracy for the era is barely over 50%, including just 2 of the last 6 in the era, but its MAE is only just over 4 goals per game.

Some of the records for this era are:

  • Highest rated team: South Melbourne 1934 (+30.4)

  • Highest aggregate Ratings: Richmond v South Melbourne 1934 (+50.0)

  • Highest forecast margin: Collingwood v Richmond 1928 (12.2 pts)

  • Highest absolute error of forecast: Melbourne v Collingwood 1926 (60.9 pts)

  • Biggest Upset (as measured by expected victory margin for loser): Collingwood v Richmond 1928 (12.2 pts)

  • Lowest rated team: Fitzroy 1922 (+7.7)

  • Lowest aggregate Ratings: Fitzroy v Collingwood 1922 (+15.6)

This era also had the game with the all-time lowest forecast margin - the 1933 Grand Final in which South Melbourne defeated Richmond by 6 goals. MoSHBODS had South Melbourne as 0.02 point favourites for this game.

In this era, winning teams were roughly 2.6 points weaker offensively, but 2.1 point stronger defensively than the teams they defeated. On average, then, the winners’ Combined Ratings were lower than the losers’. The 1928, 1934 and 1936 Grand Finals, in which the winning teams were all Rated 11 to 15-points higher than the teams that beat them, contributed heavily to that result.

This era also produced only 147 points per game in Grand Finals, and an average victory margin of just over 4 goals per game.

1897 to 1919

In the final, extended era of 1897 to 1919, MoSHBODS was moderately accurate but quite precise in its margin forecasts.

MoSHBODS’ accuracy for the era is 64% (but includes correct tips in 6 of the last 7 in the era), and its MAE a very impressive 15 points per game, aided considerably by the fact that Grand Finals in this era produced, on average, only 90 points per game.

Some of the records for this era are:

  • Highest rated team: Essendon 1901 (+25.0)

  • Highest aggregate Ratings: Carlton v Collingwood 1915 (+42.4)

  • Highest forecast margin: Collingwood v Fitzroy 1917 (15.2 pts)

  • Highest absolute error of forecast: Carlton v Fitzroy 1906 (55.0 pts)

  • Biggest Upset (as measured by expected victory margin for loser): Fitzroy v Carlton 1916 (13.8 pts)

  • Lowest rated team: Fitzroy 1917 (-0.9). This is the all-time record, and the only time a team with a negative Combined Rating has played in a Grand Final.

  • Lowest aggregate Ratings: Fitzroy v Carlton 1904 (+6.5). This is the all-time record. Fitzroy’s Combined Rating for that game was +2.0, which is the all-time lowest for any winning team.

In this era, winning teams were roughly 1.2 points stronger offensively, and 0.6 points stronger defensively than the teams they defeated. The era also produced, as noted, only 90 points per game in Grand Finals, and an average victory margin of just 16.5 points per game.

Across all of history, MoSHBODS has picked the winner 62% of the time, and recorded an overall MAE in Grand Finals of 28.2 points per game. Make of that what you will as you assess its forecasts for this year’s final game.