MAFL 2013 : Team Dashboard for Round 23

/

Before providing the final Team Dashboard of the season I want to present a summary view of each team's performance this season, which I've done by creating a bubble chart where the area is proportional to the victory margin in the relevant game and the colour reflects the result from the viewpoint of the Home team.

One aspect of the season that I think this chart makes particularly clear is the general narrowness of the Roos' defeats. To see this, peruse the collection of small, red circles on the row labelled "Kangaroos" and the assemblage of small, green circles - which also denote Roos losses - in the column labelled similarly.

Also highlighted by this chart is the magnitude of many GWS and Melbourne losses and, more subtly, the inequity of the home-and-away clashes that were missed as a consequence of the non all-plays-all home-and-away nature of the AFL draw. Which team wouldn't, for example, want one of those large green circles that proliferate in the columns labelled "GWS" or "Melbourne"?

With that by way of entree, here's the season's final Team Dashboard:

(I've left the Dons in 7th though I realise that, officially, they'll finish 9th.)

The Roos, therefore, have missed out on a place in the Finals despite finishing the season having scored, on average, about 20% more points than their opponents, ranking them 5th amongst all teams, while Port Adelaide have made it after racking up only a bit over 2% more points than they conceded.

Somewhat unusually, half of the teams that have lived to fight another week found themselves on the end of a loss this week, while five of the teams that won't participate in the Finals were victorious. The end of any AFL season is often a curious time ...

At the end of the home-and-away season only two teams have recorded more than two wins fewer than would have been expected based on their scoring shot data and MAFL's Win Production Function: Adelaide, who recorded 2.6 wins fewer, and the Roos, who recorded over 5 wins fewer. No team recorded more than 1.6 wins more than would have been expected, and half of the teams finished within 1 win - plus or minus - of the number we'd have expected.

Hawthorn finished the home-and-away season as the team generating the greatest number of scoring shots per game, while the Dees finished as the team generating the fewest. Fremantle conceded the fewest and GWS the greatest number per game.

Fremantle also recorded the highest scoring shot conversion rate, and Carlton the lowest, while GWS allowed the greatest proportion of scoring shots to be converted to goals and Carlton allowed the smallest proportion - how else could the Blues have hoped to have made the Finals given their own inability to convert?

The Roos were best at 1st Quarters, Sydney at 2nd Quarters, Geelong at 3rd Quarters, and Port Adelaide at 4th Quarters, while worst, respectively, were the Dogs, Dees, Giants and the Saints.

Rank correlations between final ladder position and ranking on the various metrics were as follows:

  • Competition Ladder vs Own Scoring Shot production: +0.78
  • Competition Ladder vs Opponent Scoring Shot production: +0.90
  • Competition Ladder vs Own Conversion rate: +0.52
  • Competition Ladder vs Opponent Conversion rate: +0.29
  • Competition Ladder vs Q1 Performances: +0.76
  • Competition Ladder vs Q2 Performances: +0.95
  • Competition Ladder vs Q3 Performances: +0.74
  • Competition Ladder vs Q4 Performances: +0.56

This year, it seems, success was very much about ensuring that opponent's scoring opportunities were hard to convert and about performing especially well in 2nd terms.

MAFL 2013 : Team Dashboard for Round 22

/

Should Port Adelaide defeat Carlton next week and, as some are suggesting might the case, the Dons are excluded from the Finals or choose not to participate, then it's possible we'll have a team, the Brisbane Lions, playing in the Finals with a percentage below 90 and another team, the Roos, not playing despite recording a percentage during the home-and-away season above 120. 

In truth, even if they lose, Carlton would still be favoured to grab any vacated Finals berth however, as the scenario just outlined would require that the Lions draw with or defeat the Cats, which seems unlikely, especially in light of the Lions' near loss to the lowly Dogs this week after leading by over 50 points..

MAFL 2013 : Team Dashboard for Round 20

/

It's been a little while since we looked at the strengths and weaknesses of each team in terms of the metrics on the MAFL Team Dashboard, so let's do that this week: 

  • Adelaide, 12th on the ladder, are 17th on Own Conversion rate, but 3rd in final terms
  • Brisbane, 13th on the ladder, are 18th on Own Conversion rate, but 8th on Opponent Conversion rate and in final terms 
  • Carlton, 9th on the ladder, are 13th on Own Conversion rate and in final terms, but 1st on Opponent Conversion rate
  • Collingwood, 5th on the ladder, are 13th on Opponent Conversion rate, but 4th on Opponent Scoring Shots per game
  • Essendon, 7th on the ladder, are 11th on Opponent Conversion rate, but 4th in final terms
  • Fremantle, 4th on the ladder, are 14th on Own Scoring Shots per game, but 1st on Opponent Scoring Shots per game
  • Geelong, 2nd on the ladder, are 16th on Opponent Conversion rate, but 1st on 3rd Quarter performances
  • Hawthorn, 1st on the ladder, are 9th on Opponent Conversion rate, but 1st on Own Scoring Shots per game
  • Kangaroos, 11th on the ladder, are 11th on 3rd Quarter performances (yes, 11th is the worst ranking they have on any metric), but 1st on 1st Quarter performances and on Own Conversion rate
  • Melbourne, 17th on the ladder, are 18th on Own Scoring Shots per game and on 2nd Quarter performances, but 10th on Opponent Conversion rate
  • Port Adelaide, 8th on the ladder, are 15th on Opponent Conversion rate, but 1st in final terms
  • Richmond, 6th on the ladder, are 14th on Own Conversion rate, but 4th on Opponent Conversion rate
  • St Kilda, 16th on the ladder, are 17th on performances in final terms, but 7th on Opponent Conversion rate
  • Sydney, 3rd on the ladder, are 11th on performances in final terms, but 1st on 2nd Quarter performances
  • West Coast, 10th on the ladder, are 18th on Opponent Conversion rate, but 4th on Own Scoring Shots per game and in 3rd terms
  • Western Bulldogs, 15th on the ladder, are 18th on performances in 1st terms, but 6th on Own Conversion rate

Having read that list, it's probably no surprise that the weakest rank correlation between competition ladder position and ranking on any single metric is for Opponent Conversion rate, for which it is only +0.18. The next-weakest correlation is with Own Conversion rate where it's +0.51.

Strongest correlations are for Opponent Scoring Shots per game (+0.93), and for performances in 2nd Quarters (+0.92). Surprisingly - to me anyway - the correlation between performances in final terms and competition ladder position is a relatively weak +0.60.

MAFL 2013 : Team Dashboard for Round 18

/

Time, perhaps, again to look at the significant differences between each team's competition ladder position and their ranking on the various Team Dashboard metrics.

Here are some of the major differences: 

  • Carlton, 9th on the ladder, are 4th on Scoring Shots per game, and 1st on Opponent Conversion Rate
  • Collingwood, 7th on the ladder, are 13th on Own Scoring Shot Conversion and 16th on Opponent Scoring Shot Conversion
  • Essendon, 4th on the ladder, are 8th on Opponent Scoring Shot Conversion
  • Fremantle, 5th on the ladder, are 14th on Scoring Shots per game
  • Geelong, 2nd on the ladder, are 13th on Opponent Scoring Shot Conversion
  • Gold Coast, 14th on the ladder, are 5th on Opponent Scoring Shot Conversion
  • GWS, 18th on the ladder, are 9th on Own Scoring Shot Conversion
  • Hawthorn, 1st on the ladder, are 12th on Opponent Scoring Shot Conversion
  • The Kangaroos, 10th on the ladder, are 2nd on Own Scoring Shot Conversion
  • Melbourne, 17th on the ladder, are 8th on Own Scoring Shot Conversion and 9th on Opponent Scoring Shot Conversion
  • Port Adelaide, 8th on the ladder, are 12th on Own Scoring Shot Conversion and 15th on Opponent Scoring Shot Conversion
  • Richmond, 6th on the ladder, are 15th on Own Scoring Shot Conversion
  • St Kilda, 16th on the ladder, are 11th on Own Scoring Shot Conversion and 11th on Opponent Scoring Shot Conversion
  • Sydney, 3rd on the ladder, are 6th on Own Scoring Shot Conversion
  • West Coast, 11th on the ladder, are 7th on Own Scoring Shots per game, and 18th on Opponent Scoring Shot Conversion
  • The Western Bulldogs, 15th on the ladder, are 3rd on Own Scoring Shot Conversion and 8th on Opponent Scoring Shot Conversion

As is apparent from the disparity between ladder position and Scoring Shot Conversion rates laid bare by this list, in 2013 Scoring Shot Conversion rates have mattered little - instead it's been about Scoring Shot production and prevention.

(The Roos are currently the only team who have won or lost (in their case, lost) more than 2 games more than their scoring statistics would suggest they should have.) 

MAFL 2013 : Team Dashboard for Round 17

/

This week I thought I'd look again at the rank correlations between teams' competition ladder positions and their ranking on some of the metrics from the Team Dashboard.

Currently, the metric most highly correlated with ladder position is the number of Scoring Shots conceded per game for which the rank correlation is +0.90. The next most highly correlated metrics are Scoring Shots created (+0.83) and the teams' performances in 2nd Quarters (also +0.83).

Thereafter come teams' performances in 3rd Quarters (+0.79), teams' performances in 4th Quarters (+0.72), and then teams' performances in 1st Quarters (+0.66). Teams that have led at the first change this season have won only about two-thirds of the time. That's similar to the rate achieved in 2011 but significantly lower than the rate of about 75% that we witnessed in 2012.

The only metrics less correlated with teams' ladder positions are the teams' scoring shot conversion metrics. Teams' own scoring shot conversion rate correlates at only +0.32 with their ladder positions, while the average conversion rate of their opponents correlates at only +0.21. As I think I've noted before, this season has been about the volume of scoring shots created and conceded by teams, not about the rate at which those scoring shots have been converted.

(I'll just note in passing that the Roos, on the back of another loss with a score than would win most games (16.12), have now won almost 4.5 games fewer than their aggregate scoring shot performances for the season and MAFL's Win Production function suggest they should have.)

MAFL 2013 : Team Dashboard for Round 16

/

Seven of the teams in the competition's top 8 recorded victories this week, all as favourites in their respective games, while the two remaining victories came from the teams now in 9th (also as favourites) and in 13th position.

What now separates the teams in positions 1 and 2 is their respective scoring abilities, as both are equal on competition points and, coincidentally, also on points conceded. Further down the ladder, it's also offensive ability that predominantly separates 4th from 5th, with Sydney's 1,550 points eclipsing Freo's 1,322.

Even further down the ladder, the Roos continue to look oddly out of place, after yet another agonising and narrow loss has left them in 11th place with a 116 percentage, the 2nd-best conversion rate in the league, and the best Q1 statistics of any team.


MAFL 2013 : Team Dashboard for Round 14

/

Having just recorded another week's profitable wagering, I'm in a charitable mood tonight, so I'm going to highlight for each team the metric (or metrics) on which it is currently performing best.

  • Adelaide are 4th best in 4th Quarters
  • Brisbane Lions are 7th best on Opponent Conversion rate
  • Carlton are best on Opponent Conversion rate
  • Collingwood are 8th best on Opponent Scoring Shots per Game and on 1st and 2nd Quarters (and 8th on the ladder)
  • Essendon are 2nd best on 3rd Quarters
  • Fremantle are best on Opponent Scoring Shots per Game and on 2nd Quarters
  • Geelong are best in 3rd Quarters
  • Gold Coast are 9th on 1st and 3rd Quarters
  • GWS are 8th on Own Conversion rate
  • Hawthorn are best on Own Scoring Shots per Game, Own Conversion Rate (and 1st on the ladder)
  • Kangaroos are best on 1st Quarters
  • Melbourne are 10th on Own Conversion rate
  • Port Adelaide are best on 4th Quarters
  • Richmond are 3rd on 1st Quarters
  • St Kilda are 6th on Opponent Conversion rate
  • Sydney are 2nd on Opponent Scoring Shots per Game, Opponent Conversion rate, and on 2nd Quarters (they're also 2nd on the ladder)
  • West Coast are 4th on Own Scoring Shots per Game and on Opponent Scoring Shots per Game
  • Western Bulldogs are 5th on Own Conversion rate

MAFL 2013 : Team Dashboard for Round 13

/

Here's the latest Team Dashboard, which this week returns to ordering teams in the same way as the official competition ladder since all teams have now played the same number of games.

(Just one comment this week: it must be a long time since a team with a 107 percentage and the best record in the league in 1st terms has sat 5 places and 3 wins outside the finals. Tough year to be a Roos supporter.)

MAFL 2013 : Team Dashboard for Round 12

/

Here's the latest Team Dashboard (along with a reminder that I order the teams in the Dashboard based on the competition points earned as a percentage of the maximum possible for a team, given the number of games that it has played. That's why, for example, I have Essendon, who've played 12 games, behind Sydney and Fremantle who have fewer competition points but from fewer games.)

This week I'll just highlight a few things that have caught my eye from the Competition Ladder itself. Mostly they're of no particular import and are probably best described as "interesting" rather than "insightful".

  • Fremantle have scored fewer goals than any other team in the top 13 positions on the ladder. But, they've also conceded the fewest goals of any team in the competition.
  • Two-thirds of the teams in the competition have a percentage above 100, which highlights the feasting that's been going on when the top teams have played the bottom teams.
  • Carlton, in 10th spot on the ladder, have scored more behinds than any other team in the competition. They've also conceded third-most, better only than Melbourne and GWS.
  • Between them, despite making up only one-ninth of the teams in the competition, Melbourne and GWS have conceded over one-sixth of the goals and almost 15% of the behinds.
  • Amongst them, over the past 5 weeks, the bottom 4 teams have recorded just a single win from the 17 games they've played.

MAFL 2013 : Team Dashboard for Round 11

/

The latest Team Dashboard follows, for which I'd note that my ordering of the teams will differ from the "official" view. This is because I adjust for byes by ranking teams based on the competition points they've accrued relative to the opportunities they've had to accrue them, not by the total points they've actually accrued. Practically, right now that affects the rankings only of Fremantle, Essendon, Richmond and Port Adelaide.

This week I'll restrict my comments to those relating to team Quarter-by-Quarter performances, about which I'll note that:

  • Adelaide are poor starters but strong finishers in each half of games, being ranked 11th and 12th in Q1 and Q3, but 4th and 2nd in Q2 and Q4
  • Carlton are particularly weak in Q2 (ranked 13th) but strong in Q3 (ranked 4th)
  • Collingwood are notably weak in Q4 where they're ranked 12th and have a percentage of just 100
  • Fremantle are strong in Q2 where they're ranked 2nd
  • Geelong are the best team in the competition in Q3 but only 12th in Q1
  • Gold Coast have much better 2nd halfs than 1st halfs - they're ranked 6th and 8th in Q3 and Q4, and 10th and 14th in Q1 and Q2
  • Hawthorn are surprisingly poor in Q2 where they're ranked only 11th having won only 55% of these quarters
  • The Roos are the best team in the competition in Q1 where they've a 91% win rate and a 217 percentage
  • Port Adelaide, similar to Adelaide, are poorest in Q1s and Q3s where they're ranked 16th and 10th, and strongest in Q2s and, especially, Q4s where they've ranked 5th and 1st.
  • Richmond are most notably weak in Q3 where they're ranked 13th
  • Sydney, unlike the Swans of previous years, are relatively poor finishers being ranked 11th in Q4s

MAFL 2013 : Team Dashboard for Round 9

/

This week I'll be highlighting significant differences between the teams' competition ladder positions and their Scoring statistics: 

  • Carlton are 8th on the ladder, but 1st on Opponent Conversion and 16th on Own Conversion
  • Essendon are 3rd on the ladder, but 10th on Own Conversion
  • Fremantle are 4th on the ladder, but 12th on Own Scoring Shots and 1st on Opponent Scoring Shots
  • Geelong are 2nd on the ladder, but 13th on Opponent Conversion
  • GWS are 18th on the ladder, but 7th on Own Conversion
  • Hawthorn are 1st on the ladder, but 11th on Opponent Conversion
  • The Kangaroos are 13th on the ladder, but 6th on Own Scoring Shots, Own and Opponent Conversion
  • Melbourne are 17th on the ladder, but 5th on Own Conversion
  • Port Adelaide are 9th on the ladder, but 15th on Opponent Conversion
  • St Kilda are 15th on the ladder, but 3rd on Opponent Conversion
  • Sydney are 5th on the ladder, but 13th on Own Scoring Shots and 12th on Opponent Conversion
  • West Coast are 6th on the ladder, but 1st on Own Scoring Shots and 17th on Opponent Conversion
  • The Western Bulldogs are 16th on the ladder, but 4th on Own Conversion

One feature from this list is the apparent lack of importance of Opponent Conversion rates - teams high up on the ladder seem to have poor Opponent Conversion statistics and teams lower on the ladder seem to have better Opponent Conversion statistics. Indeed, the rank correlation between ladder position and ranking based on Opponent Conversion is just +0.18. 

What's mattered instead this season has been how many scoring shots opponents a team has allowed its opponents. The correlation between the ranking on that metric and ladder position now stands at +0.86. 

MAFL 2013 : Team Dashboard for Round 8

/

This week I'll draw your attention to the right-hand portion of the Scoring Shot Data, which provides information about each team's actual performance in terms of competition points earned, relative to what might be expected based on the team's scoring shot metrics and the win production function I derived back in this blog from 2011. (Note that I'm using the first equation from that blog).

According to this analysis, the teams that have achieved significantly more wins than their scoring shot data would predict are: 

  • Geelong, who've won 88% of their games when their scoring statistics would predict a 67% record,
  • Collingwood, who've won 63% of their games when their scoring statistics would predict a 46% record, and
  • The Brisbane Lions, who've won 38% of their games when their scoring statistics would predict a 24% record

Confirmatory evidence for the relative good fortune of these three teams comes from the fact that both the Cats and the Pies have recorded 3 victories by less than 3 goals, and that the Lions have recorded 2 victories by less than 2 goals.

Conversely, the teams that have achieved significantly fewer wins than their scoring shot data would predict are: 

  • Adelaide, who've won 50% of their games when their scoring statistics would predict a 72% record,
  • The Roos, who've won 38% of their games when their scoring statistics would predict a 59% record, and
  • St Kilda, who've won 25% of their games when their scoring statistics would predict a 40% record

Again we can find support for these assessments in relevant game margins: Adelaide have lost 2 games by less than 2 goals, while the Roos have lost 4 games and the Saints 3 games by less than 3 goals.  

MAFL 2013 : Team Dashboard for Round 6

/

This week I'm going to discuss some of the rank correlations between various of the metrics on the Team Dashboard and teams' competition ladder positions.

Firstly, looking at the metrics in the Scoring Shot Data section, we find the following correlations with ladder position (the equivalent correlations for the Team Dashboard at the end of the 2012 home-and-away season are in brackets): 

  • Correlation with Own Scoring Shots per Game : +0.79 (+0.88)
  • Correlation with Opponent Scoring Shots per Game : +0.78 (+0.77)
  • Correlation with Difference in Scoring Shots per Game : +0.95 (+0.89)
  • Correlation with Own Conversion Rate : +0.27 (+0.70)
  • Correlation with Opponent Conversion Rate : +0.23 (+0.29)
  • Correlation with Difference in Conversion Rate : +0.39 (+0.69)

The only substantial differences in these correlations from the equivalent figures for last season are those for Own Conversion Rate and the Difference in Conversion Rate. This season then, so far, it's been more important for teams to create lots of scoring shots and prevent their opponents from doing the same than it has been to convert those opportunities or prevent their opponents from doing so. Put another way, it's been a season for quantity of opportunities rather than quality. 

Finally, consider the correlations between quarter-by-quarter performance rankings and competition ladder position: 

  • Q1 performance : +0.39 (+0.79)
  • Q2 performance : +0.72 (+0.76)
  • Q3 performance : +0.84 (+0.62)
  • Q4 performance : +0.65 (+0.70)

The obvious conclusion from this is that 1st terms have been far less important and 3rd terms somewhat more important so far this season than they were in the home-and-away season last year. I'll have more to say about this in a blog I'm planning to write this week. 

MAFL 2013 : Team Dashboard for Round 4

/

A handful of things that caught my eye as I reviewed this week's Team Dashboard:

  • The first 11 teams on the ladder have percentages over 100. When was the last time that was true, I wonder
  • Adelaide, in 8th, have scored only about 10 points per game more than GWS, in 18th. Of course they've conceded over 18 points per game fewer as well, which is what explains their more-elevated status
  • Geelong's combined victory margins across its four wins is just 48 points
  • Port Adelaide have registered almost 12 scoring shots per game more than their opponents. Melbourne have conceded almost 19 more per game than their opponents
  • Geelong have the best record of any team in the 3rd quarter, but are 14th and 15th-best in 1st and 2nd terms
  • Hawthorn, the Roos and Richmond all have the opposite pattern, faring relatively poorly in 3rd terms and much better in the rest of the game
  • Both the Gold Coast in 2nd terms and Melbourne in 3rd terms score only 18 points for every 100 points scored by their opponents in these respective quarters