Underachieving and Overachieving Teams

A couple of blogs back I described some win production functions, which relate a team's winning percentage in the home-and-away season to characteristics of its scoring during that season, in particular to its rate of scoring shot production and its conversion of those scoring shots relative to its opponents'.

One win production function that I found, which explained about 90% of the variability in winning percentages for every team in every ladder from 1897 to 2010, was the following:

Predicted Winning Percentage = logistic(0.164 x (Own Scoring Shots per game - Opponent Scoring Shots per game) + 6.18 x (Own Conversion Rate - Opponent Conversion Rate))

One way to think about this equation is that it describes the winning percentage that a team can expect to record given that it generates X scoring shots per game while allowing its opponents' Y, and that it converts these opportunities at a rate of A% to its opponents' B%.

If we take that perspective and apply it the final home-and-away season results for every team that's ever played VFL/AFL, we can identify two dozen teams that have won at a rate 15% or more lower than they might have expected. I bring you the All-Time Underachievers.

(Please click for a larger version of this image.)

Atop the list is the North Melbourne side of 1926, which managed only a single draw during its 18-game season, despite converting its (admittedly scant) scoring opportunities at a rate more than 5% points higher than its opponents. On the strength of their scoring statistics they might have expected to win about 26% - say 4 or 5 - of their matches. Instead, their winning percentage was just 3%.

Next worst - or next most unlucky if you prefer - is the Fitzroy team from the very first season, 1897. It won 4 and drew 1 of its 19 games that year, for a winning percentage of 32%, when its statistics suggest it should have won about 7 games. It generated almost 1.5 more scoring shots per game than its opponents which, despite its poorer conversion rate, allowed it to register a percentage of 104.7, a number far more often associated with better seasons than theirs.

A little further down the list we find some more recent examples of teams with competition points mismatched to their scoring statistics. There's the 1996 Tigers with a 50% winning percentage but with superior scoring shot production and superior conversion performance than their opponents , the 1994 North Melbourne side with astonishingly good scoring statistics for a team with only a 59% winning percentage, the 1997 Pies with a 111.4 percentage but a 45% winning rate, and the 2008 Fremantle team with undeniably mediocre scoring statistics, but not sufficiently mediocre to deserve a 6 and 16 season.

As history dishes out undue punishment, it also serves up undeserved reward. Here then are the All-Time Overachievers.

(Please click for a larger version of this image.)

Having waited 56 years to erase the inequities visited upon their 1897 brethren, Fitzroy's 1953 team squeezed lemons and made lemonade by turning a deficit of 3.5 scoring shots per game and a narrow conversion rate deficit into a 10 and 8 season, enough for 6th spot of the final home-and-away ladder. In so doing they became the greatest overachievers in VFL/AFL history.

The St Kilda side of 1919 were similarly fortuitous, registering over 4.5 fewer scoring shots per game than their opponents and converting at a rate almost 6% points lower yet still winning 44% of their matches.

More recently, the 2000 Fremantle team produced 8 fewer scoring shots per game than their opponents and converted at a rate 2% points lower but won 8 of their 22 home-and-away games, finishing 12th. Also, in 2006 Richmond won half their games after generating almost 4 fewer scoring shots per game than their opponents, and in 2004 Carlton finished 10 and 12 having generated 5 fewer scoring shots per game than their opponents.

A few other recent anomalous performances that didn't quite make the cut in the All-Time lists above are set out in the following table.

(Please click for a larger version of this image.)

These then are all examples of single-season sorrow and surprise. To finish, here's a table showing the aggregated underachievement and overachievement by team in 10 year blocks, where negative values connote aggregate underachievement and positive values aggregate overachievement across those 10 years.

(Where necessary I've collapsed the results for merged teams into the one row - so, for example, the data labelled Western Bulldogs relates to Footscray's performances in the appropriate decades.)

Looking just the last two decades, Adelaide, Collingwood and Sydney stand out as the greatest underachievers, while Port Adelaide and Richmond shine as the teams that have won most with least.

Taking a longer view, the short-lived University team take the prize as the competition's chronic underachievers. Amongst the teams with longer histories, Carlton, the Roos/North Melbourne and Richmond have the best records for overachievement, while Adelaide, Hawthorn, the Lions/Fitzroy/Bears, and Geelong have the worst records for underachievement.