# Home and Away Season Competition Points and Scoring Metrics

On Twitter today someone made the observation that teams' accuracy or conversion rate differentials seemed to have been strongly correlated with their final ladder position. In other words, if we calculated the difference between a team's conversion rate - the proportion of scoring shots that it converts to goals - and that of its opponents', then this difference would tend to be related to their competition performance.

That struck me as an interesting thing to explore, not just for recent seasons, but also for the entire history of the competition, and the results of that exploration are what I've summarised in this blog.

The data I've used comes from the end of Home and Away season ladders for every season from 1897 to 2014, from which I've collected five data elements:

• Competition Points (ignoring deductions)
• Goals For
• Behinds For
• Goals Against
• Behinds Against

These last four data elements have been used to calculate, for each team:

• Own Conversion = Goals For / (Goals For + Behinds For)
• Opponents' Conversion = Goals Against / (Goals Against + Behinds Against)
• Own Scoring Shots per Game = (Goals For + Behinds For)/Games Played
• Opponents' Scoring Shots per Game = (Goals Against + Behinds Against)/Games Played

The correlation of each of these measures with the teams' competition points was then calculated for every season.

At the end of this blog is the long table with the actual data, which are summarised in the two charts below.

In the first chart we look at the history of the correlation between the three conversion rate metrics and competition points and find that, as my Twitter correspondent had noticed, teams' conversion rate differentials are generally quite highly correlated with team competition performance, especially in recent seasons where it's been +0.71, +0.57, and +0.70.

Further, in most seasons, the correlation with the differential has been greater in absolute terms than the correlation with either Own Conversion or Opponent Conversion rates. In fact, in only 37 of 110 seasons has the absolute value of the correlation between Opponent Conversion Rate and Competition Points exceeded that between the Conversion Rate Difference and Competition Points, and in only 31 of 110 seasons has the value of the correlation between Own Conversion Rate and Competition Points exceeded that between the Conversion Rate Difference and Competition Points.

Lastly, in 57 of the 110 seasons, the Conversion Rate Difference has been more highly correlated with Competition Points than either Own or Opponent Conversion rates.

In the second and final chart we look at the history of the correlation between the three Scoring Shot metrics and Competition Points and discover that the correlations are much higher in absolute terms but that, again, that between Competition Points and the Difference in Scoring Shot production tends to be higher in absolute terms than that between Competition Points and either of Own or Opponent Scoring Shot production.

Now we find that, in only 13 seasons is the absolute value of the correlation between Opponent Scoring Shot production per game and Competition Points greater than that between the Difference in Scoring Shot production per game and Competition points, and that in only 8 seasons is the value of the correlation between Own Scoring Shot production per game and Competition Points greater than that between the Difference in Scoring Shot production per game and Competition points. We also find that in 97 seasons the Difference in Scoring Shot production per game is more highly correlated with Competition Points than either Own or Opponent Scoring Shot productions and that, in the same number of seasons the Difference in Scoring Shot production per game is more highly correlated with Competition Points than the Difference in Conversion rates.

Overall then it seems that:

• Scoring Shots are a better predictor of Competition Points than are Conversion Rates
• Differences in Scoring metrics are better predictors of Competition Points than are the Own and Opponent metrics that comprise those differences

Comment