Which Teams Fare Better as Favourites?

In this blog, the next is a series in which I've been exploring the all-time MARS Ratings I created for every game from the start of 1897 to the end of the 2012 season, I'll be looking at how well each team has performed depending on the relative strength of its opponent, as measured by that opponent's MARS Rating. So, for example, we'll consider how well Collingwood tends to do when playing a team it is assessed as being much stronger than, a little stronger than, about as capable as, and so on.

For the purposes of this analysis I've defined a team as being: 

• An Extreme Favourite if it enjoys a MARS Rating more than 40 points higher than its opponent (who will be labelled an Extreme Underdog)
• A Heavy Favourite if it enjoys a MARS Rating more than 25 and up to 40 points higher than its opponent (who will be labelled a Heavy Underdog)
• A Moderate Favourite if it enjoys a MARS Rating more than 10 and up to 25 points higher than its opponent (who will be labelled a Moderate Underdog)
• A Narrow Favourite if it enjoys a MARS Rating less than 10 points higher than its opponent (who will be labelled a Narrow Underdog).

In the very rare cases in which the two teams have identical MARS Ratings, the designated Home team is labelled a Narrow Favourite and the Away team a Narrow Underdog. This is the only circumstance in the current analysis under which I've taken Home team status into account.

STATISTICS FOR 1897-2012

Firstly, let's take a look at how often each team has carried each label across the entire history of VFL/AFL.

(This is going to be a table-heavy blog since I think the numbers themselves are more important than the overall impression that a chart would provide in this case.)

We see then that 7% of all contests have involved an Extreme Favourite/Underdog pairing and that Carlton, Collingwood and Geelong are the teams whose game mix is most overrepresented in this category. GWS, Gold Coast, University, Fremantle and the Western Bulldogs (Footscray) are the teams with, relatively speaking, the fewest games where they've been the Extreme Favourite.

Conversely, GWS, Gold Coast and University are the teams with the highest proportion of games in which they've been the Extreme Underdogs, whereas Adelaide, Carlton, Collingwood and Essendon are the teams with relatively few such games in their respective histories.

Gold Coast, GWS and University aside (since their short histories and very low Ratings have resulted in quite skewed statistics for them) other notable team differences relative to the overall average are: 

• Carlton as Heavy Favourites (16% vs 10%) and Fremantle and St Kilda as Heavy Underdogs (16% vs 10%)
• Collingwood as Moderate Favourites (24% vs 18%) and Fremantle as Moderate Underdogs (31% vs 18%)

To my original question then, which teams fare best carrying which designation?

The first thing to note is that, as you'd expect, virtually every team's winning rate declines as you move from left to right across the columns, which is equivalent to saying that teams tend to win less often as their MARS Rating declines relative to their opponents'. Fremantle, Port Adelaide and St Kilda are the only exceptions to this and even they are exceptions only for a couple of Favourite or Underdog designations.

Sydney (South Melbourne) are the team you want to be on when they're Extreme Favourites. They've won 96% of the time when they've enjoyed this status - which, I should also note, has been relatively rare - in comparison with the all-team average of 88%. Fremantle, who've also carried the Extreme Favourite status only rarely, have been especially prone to being upset in this situation, winning only 57% of such games, which is actually substantially lower than their winning rate of 85% as Heavy Favourites, as alluded to earlier.

A few other general observations I think are worthy of note: 

• Adelaide, St Kilda, Port Adelaide and University have markedly different success rates as Narrow Favourites than they do as Narrow Underdogs
• Collingwood, Geelong and, especially, Adelaide, fare significantly better than other teams when they've Extreme Underdogs. Adelaide wins at twice the all-team average rate when it is in this position
• Hawthorn wins at less than half the all-team average rate when it is an Extreme Underdog 

STATISTICS FOR 1980-2012

Stretching the analysis all the way back to 1897 is an interesting and instructive exercise, but none of us can attest on the basis of lived experience to the veracity of the conclusions. 

For this next section, then, I'll instead be focussing on what I'll call the "modern age" of football, which I've defined based on a statistical analysis of scoring patterns and margins. If you look, for example, at the first chart in the blog post I linked to earlier you can see an obvious step-change around 1980 in the average MAPE of the predictive model I built using only MARS Ratings and home team status as inputs.

With this time frame in mind, let's again firstly look at the proportion of games for which each team has carried each Favourite and Underdog label.

The first thing I'd note is the remarkably similar profile of the all-team proportions at the foot of the table to those we had in the all-time view. It's not the case then, contrary to what I'd believed before I performed this analysis, that more modern games have tended to pit teams of wider relative abilities against one another.

For example, only 8% of the games in the modern age have pitted Extreme Favourite against Extreme Underdog, compared with 7% across the entire history of VFL/AFL.

On a team-by-team basis - and I'll again ignore commenting on the data for GWS and Gold Coast due to the small sample sizes that underpin it - I'd make the following observations: 

• Carlton, Essendon, Geelong and Hawthorn are all teams that have been Extreme or Heavy Favourites far more often than the all-team average during this period.
• Adelaide, Collingwood and West Coast are all notable for the proportion of games in which they've been Moderate or Narrow Favourites
• Fremantle have been Moderate or Heavy Underdogs at almost twice the all-team average rate
• Brisbane, Melbourne and St Kilda have been Extreme Underdogs at or near to twice the all-team average rate 

The teams' winning rates when carrying each label during this period are provided by this next table.

We again see the general pattern of declining winning rates as we move left to right, which, if nothing else, tells us that MARS Ratings are at least crudely useful. Again too we see that Sydney (South Melbourne) have performed extraordinarily well as Extreme Favourites, though Collingwood and the Western Bulldogs (Footscray) have done almost as well in the modern age.

One other feature of this table that caught my eye was the slightly lower winning rates of Favourites of all kinds in the 1980-2012 period compared to 1897-2012. For example, Heavy Favourites have won only 71% of the time in the modern era versus 77% of the time across the entire span of history.

Also notable are the relative performances of Carlton, Melbourne and St Kilda as Narrow Favourites when compared with their performances as Narrow Underdogs. In St Kilda's case, the difference in winning rate is a startling 27% points.

Essendon also emerge as strong performers when Underdogs of any sort other than Narrow. They win at 45%, 38% and 19% respectively as Moderate, Heavy and Extreme Underdogs compared with all-team averages of 37%, 29% and 15%.

(By the way, in this and in the previous table of the same kind for the longer time period, the rightmost column provides each team's overall winning rate for the period in question.)

STATISTICS FOR 2000-2012

Lastly, and briefly, I want to constrain the analysis to the last 13 seasons only.

Comparing this table with the same one for the 1980-2012 period I'd note that: 

• Carlton, Hawthorn and West Coast have been Extreme and Heavy Favourites much less often
• Brisbane, St Kilda and Sydney have been Extreme Underdogs much less often, and Carlton much more often

Finally, the team-by-team winning rates.

We still have the general pattern of declining winning rates as we move left to right, though we now have more exceptions to that tendency.

A few of the highlights from this table for me are:

• The 92% or higher winning rates of Collingwood, Geelong, Sydney (South Melbourne) and Western Bulldogs (Footscray) as Extreme Favourites, contrasted with the 69% or lower rates for Fremantle, Melbourne, Richmond and West Coast when they're similarly designated 
• The large and positive difference between the winning rate as Extreme Favourite compared to that as Heavy Favourite for Brisbane, Geelong, Collingwood, Port Adelaide, Sydney (South Melbourne) and Western Bulldogs (Footscray) 
• Fremantle winning far more often as Heavy Favourites than they do as Extreme Favourites, and their winning relatively infrequently as Moderate or Narrow Favourites compared to the all-team averages
• Adelaide winning at almost the same rate whether they're Extreme, Heavy or Moderate Favourites, and whether they're Moderate or Heavy Underdogs
• Brisbane winning at over four times the rate when they're Extreme Underdogs compared to when they are Heavy Underdogs

There are lots more observations I could make about this and the earlier tables, but it'll be more memorable I think if you uncover them for yourself.