This page contains charts with minimal commentary about miscellaneous topics that have grabbed my attention along the way.

All of the charts can be clicked on to access a larger version.

Data: All games from 1897 to 2017

Topic: Profile the game margins (ie own score less opponent score) for all games played by a team against all opponents during a particular era

Points of Interest

Certain teams have distinctly different-shaped profiles in some eras - for example, compare North Mebourne and the Western Bulldogs in the 2000 to 2017 era
One way of assessing the relative performance of any team in a particular era is to see how high its mean margin is compared to all other teams in the era. That mean is given by the notch inside the team's boxplot for that era.

Data: All games from 1897 to 2017

Topic: Relationship between the actual Total Score and that predicted by the MoSSBODS Rating System

Points of Interest

Less of the variance in Total Scores (11-13%) has been explained in the two most-recent Eras than in any before tem
The mean absolute error (MAE) in this Era has been lower than in either of the two preceding Eras
The spread of Expected Total Scores has vary quite markedly across Eras, narrowing especially in the most-recent Era

Data: All games from 1897 to 2017

Topic: Scoring in each quarter of a game as a proportion of the final score in that game

Points of Interest

The median for each quarter and for each era is around 25%, as we might expect
For the most-recent three eras, however, the tendency has been for there to be a slightly higher proportion of points registered in 3rd and 4th Quarters
The spread of percentages is quite large, with up to 60% and in some cases 0% of points in a single game being registered in one of the quarters

By Era

Data: All games from 1897 to 2017

Topic: Relationship between Scoring Shot production and Conversion Rate

Points of Interest

Mild evidence that teams generating more Scoring Shots tend to be more accurate
Small effect size
Not true in all eras
More prominent at higher Scoring Shot levels
Note that the spread of Conversion Rates reduces with the number of Scoring Shots (ie if we think about conversion as sampling from a Bernoulli distribution, we're seeing the effects of increasing the sample size, n, and so reducing the variance).

By Team in the Modern Era

Data: All games from 2000 to 2017

Topic: Relationship between Scoring Shot production and Conversion Rate for each team

Points of Interest

Mild evidence that, for a number of teams in the modern era, their generating more Scoring Shots tends to be associated with their being more accurate
Small effect sizes
Not true for many teams
More prominent at higher Scoring Shot levels for those teams for which it is true
Again, note that the spread of Conversion Rates reduces with the number of Scoring Shots (ie if we think about conversion as sampling from a Bernoulli distribution, we're seeing the effects of increasing the sample size, n).

By Venue in the Modern Era

Data: All games from 2000 to 2017

Topic: Relationship between Scoring Shot production and Conversion Rate for each venue where at least 50 games have been played during the Era

Points of Interest

Some evidence that, for a number of venues in the modern era, teams that generate more Scoring Shots tend to be more accurate
Small effect sizes (and sample sizes for some Scoring Shot counts) at many venues
More prominent, if anything, at higher Scoring Shot levels for those venues where it is true
Again, note that the spread of Conversion Rates reduces with the number of Scoring Shots (ie if we think about conversion as sampling from a Bernoulli distribution, we're seeing the effects of increasing the sample size, n).