V/AFL CHARTS

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.

AVERAGE POINTS PER GAME IN PORTIONS OF THE H&A SEASON

Data: Every Home & Away game from 2000 to R4 of 2016

Topic: Variability in average aggregate scores for games played at different points in the Home & Away season

Points of Interest

  • Most common pattern is for aggregates to be higher in the early and late parts of the Home & Away season compared to the remainder
  • Most obvious exception was 2007

EFFECT OF DAYS REST ON PERFORMANCE RELATIVE TO EXPECTATION

Data: Games from 2000 to 2016 R2

Topic: Variability in teams' final game margins relative to expectation after having had varying levels of rest since the previous game

Points of Interest

  • For most teams, having less than 7 days' rest appears to have little to no effect on their performance relative to expectations
  • The Kangaroos and Fremantle are the obvious exceptions
  • The Western Bulldogs appear to do relatively better on less rest

Never Before Seen Final Game Scores

Winning Score & Losing Score Pairing as Points

Data: Every game from 1897 to R5 of 2016

Topic: Proportion of final scores in a season (eg 72-67) that had never before been recorded in V/AFL history

Points of Interest

  • General decline in proportion of never-before seen scores
  • Spike in late 1970s / early 1980s as higher-than-ever scores being recorded
  • Now about 1 game in 6 produces a record

 

Winning Score & Losing Score Pairing in G.B format

Data: Every game from 1897 to R5 of 2016

Topic: Proportion of final scores in a season (eg 10.12-9.13) that had never before been recorded in V/AFL history

Points of Interest

  • Similar decline in proportion of never-before seen scores
  • Similar Spike in late 1970s / early 1980s as higher-than-ever scores being recorded
  • Now about 65% of games produces a record

WINNING AND LOSING RATES IN CLOSE AND NON-CLOSE GAMES

Data: All games from 2000 to R5 2016

Topic: Evidence that some teams are better or worse at winning / losing close games relative to their overall winning rate across the period

Points of Interest

  • Adelaide is the only team for which the 2 SD confidence intervals do not overlap
  • Majority of teams have a closer to 50% record in close games than they have in other games
  • The size of the error bars on the Close Game bars (roughly) reflects the number of close games in which a team has been involved

TEAM FINAL SCORES THAT WERE PRIME NUMBERS

Proportion

Data: All games from 2000 to R5 2016

Topic: Extent to which teams' final scores were prime numbers

Points of Interest

  • Brisbane has registered a substantially smaller proportion of final scores that have been prime numbers, perhaps because of its spate of high scoring in the early 2000s (see this post for why that matters)
  • Low-scoring teams such as GWS, Gold Coast, Sydney and Fremantle have registered a slightly large proportion of final scores that have been prime (again, the blog post linked to above explains why these things are related)

Streak

Data: All games from 2000 to R5 2016

Topic: Number of consecutive games in which a team has avoided registering a final score that was a prime number

Points of Interest

  • Brisbane and Geelong have both gone very long periods without finishing on a score that's prime

All Teams' Winning and Losing Scores By Era

Data: All games from 1897 to R14 2016

Topic: Number of games during an era in which the final winning or losing score was prime or non-prime

Points of Interest

  • Winning scores are more likely than Losing scores to be prime, especially in more recent eras, partly because of the relative paucity of prime numbers amongst scores over 100
  • In the modern era, the most-common Losing Score has been prime (73 points), as was, surprisingly, the most common Winning Score in the 1980-1999 era (113 points)

CHANGE IN TEAM CONVERSIOn RATES AFTER WINS AND LOSSES IN the home-and-away season

Data: All home-and-away games from 1997 to R12 2016 (excluding those for teams whose previous game ended as a draw).

Topic: How do teams' conversion rates tend to respond after a win and after a loss. Since teams that win tend to convert at slightly better-than-average rates and teams that lose tend to convert at slightly worse-than-average rates, one hypothesis would be that teams' conversion rates will regress towards the mean, falling after a win and increasing after a loss.

Note that conversion rate is defined as Goals / (Goals + Behinds).

Points of Interest

  • The hypothesis is very clearly borne out: in every season, winning teams tend to kick less accurately in their next game, while losing teams tend to kick more accurately
  • As well as being true for all of the seasons shown here, it in fact holds for all but a handul of seasons across the entirety of V/AFL history. In only 7 seasons have losing teams more often seen their conversion rate fall than increase, and in only 8 seasons have winning teams more often seen their conversion rate increase rather than fall.

TEAM WINNING RATE IN PREVIOUS 100 GAMES

Data: All games from 1897 to 2015

Topic: Teams' winning rate in 100 most-recent games, Home & Away and Finals, with draws counted as half-wins

Points of Interest

  • It's rare for a team to dip below 25% or rise above 75%
  • The generally cyclical nature of team performance levels across time is evident

RELATIONSHIP BETWEEN SCORING AND WINNING RATE

Scoring Shots

Data: All games from 1897 to R6 2016

Topic: Relationship between Scoring Shot production and Winning Rate by Era

Points of Interest

  • Relationship is S-shaped for all eras, but steeper in some eras than in others
  • Fewer Scoring Shots are associated with a 50% win rate (about 25) in the 2000-2016 era than in the 1980-1999 era (about 28)
  • Increasing from 25 Scoring Shots to 29 Scoring Shots (ie just one Scoring Shot per quarter) in the 2000-2016 era is associated with a 23% point increase in Winning Rate (from 51.6% to 75.1%)

Goals

Data: All games from 1897 to R6 2016

Topic: Relationship between Goals production and Winning Rate by Era

Points of Interest

  • Relationship is also S-shaped for all eras and steeper in some eras than in others
  • Fewer Goals are associated with a 50% win rate (about 14) in the 2000-2016 era than in the 1980-1999 era (about 15)
  • Increasing from 14 Goals to 16 Goals (ie just one Goal per half) in the 2000-2016 era is associated with a 23% point increase in Winning Rate (from 54.2% to 77.2%)

Scoring Shots by Team (in Modern Era)

Data: All games from 2000 to R6 2016

Topic: Relationship between Scoring Shot production and Winning Rate by Team

Points of Interest

  • Only a handful of teams are able to win at a greater than 50% rate registering only 23 or 24 Scoring Shots (Geelong 54%, Kangaroos 53%, Sydney 55%).
  • Others register winning rates of over 60% with just 25 or 26 Scoring Shots (GWS 75% [but from only 4 games], Geelong 65%, Sydney 64%, Fremantle 63%, Kangaroos 61%)
  • Carlton's winning rate goes above 50% only when it registers 29 or 30 Scoring Shots

RELATIONSHIP BETWEEN TOTAL SCORE AND VICTORY MARGIN BY ERA

Data: All games from 1897 to R8 2016

Topic: Relationship between the Total Score in a game and the Margin of Victory in that game

Points of Interest

  • In the modern era, higher Margins have been associated with slightly lower Total Scores than in the previous era. For example, the average Total Score for a Margin of 20 points has been about 183 points in the modern era, whereas it was about 191 points in the previous era.
  • If we consider games from the last three eras with a Total Score of 200 points, the expected Margins would be: 85 points for 1960-1979; 39 points for 1980-1999; 81 points for 2000-2016.

RELATIONSHIP BETWEEN EXPECTED DIFFICULTY OF PREVIOUS GAME AND PERFORMANCE IN CURRENT GAME RELATIVE TO EXPECTATION

Data: MoSSBODS-based expectations and actual game margins for all home and away games from 2000 to Round 10 of 2016

Topic: It's commonly asserted that teams benefit more from a 'tough' game than an 'easy one. We proxy game difficulty here by expected margin and assess performance relative to MoSSBODS expectation.

Points of Interest

  • No team shows substantial evidence of any relationship between its performance in one game relative to expectation and the difficulty of its previous game
  • The overall correlation is about +0.02

CLOSE GAMES AND BLOWOUTS IN 2016 AFTER 8 ROUNDS

Games Won by less than a Goal

Data: All game margins from games played in the first 8 rounds of 2016

Topic: What proportion of games were "close", that is, decided by less than a goal

Points of Interest

  • In 2015 only 14 of 206 games finished with a margin under a goal
  • In the first 8 rounds of 2016, 10 games had already had such a final margin
  • Close games were especially common in some of the early years of the competition and in the 1950s and 1960s
  • They were relatively rare in many of the seasons during the 1980s

Games Won by more than 5 Goals

Data: All game margins from games played in the first 8 rounds of 2016

Topic: What proportion of games were "blowouts", that is, decided by more than 5 goals

Points of Interest

  • In 2015 106 of 206 games finished with a margin over 5 goals
  • In the first 8 rounds of 2016, 43 games had already had such a final margin
  • Blowouts have been especially common in since about the early 1980s
  • They were relatively rare in the 1920s, 1950s and 1960s

DAYS SINCE WINNING AND LOSING SCORE COMBINATION LAST SEEN

Data: All winning and losing score combinations from 1897 to Round 11 2016

Topic: Depict how long it has been since every witnessed combination of winning and losing scores has occurred

Points of Interest

  • Most of the combinations of very low winning and losing scores haven't been seen for thousands of days, some since the first years of the V/AFL competition, which is over 40,000 days ago (or a bit over 4 in log base 10)
  • Darker colours are seen for scores that are more common in the modern era although there are a surprising number of gaps even in that area representing scores that have never been witnessed

TEAM MoSSBODS RATINGS AFTER 8 ROUNDS

With Combined Rating (CR) lines for specific CRs

Data: All teams' MoSSBODS Ratings after the Official Round 8 from 1897 to 2016

Topic: Comparison of 2016 team Ratings and the historical Ratings of Grand Finalists (more correctly, Premiers and Runners Up) across history. This version of the chart includes lines reflecting specific Combined Ratings (CRs)

Points of Interest

  • No eventual Premier has had an Offensive Rating less than -1.52, a Defensive Rating less than -2.27, or a Combined Rating less than -0.81 after Round 8
  • No eventual Runner Up has had an Offensive Rating less than -3.26, a Defensive Rating less than -1.81, or a Combined Rating less than -2.73 after Round 8

With Combined Rating (CR) lines for specific quantiles of CRs as at the end of Round 8 for all Grand Finalists and with outliers labeled

Data: All teams' MoSSBODS Ratings after the Official Round 8 from 1897 to 2016

Topic: Comparison of 2016 team Ratings and the historical Ratings of Grand Finalists (more correctly, Premiers and Runners Up) across history. This version of the chart includes lines reflecting quantiles for the Combined Ratings of Grand Finalists after Round 8 (eg the lowest Combined Rating of any subsequent Grand Finalist [labelled "Min CR for GF"], the Combined Rating of the Grand Finalists in the lowest 10% [labelled "CR obtained by 90% of GF"], and so on) 

Points of Interest

  • The lowest Combined Rating after Round 8 of any Grand Finalist is -2.73 (Sydney 1996)
  • The 10th percentile of Combined Ratings after Round 8 of all Grand Finalists is +1.05
  • The median Combined Rating after Round 8 of all Grand Finalists is +4.47
  • The 90th percentile of Combined Ratings after Round 8 of all Grand Finalists is +7.49