Footy Forensics: Stars vs. spread

Every team would love a 'Dangerwood' combination. But is an even spread of contributors more valuable than a superstar or two?

Last Saturday, a familiar scenario played out. Geelong were in serious trouble of suffering an upset loss before the dynamic duo of Patrick Dangerfield and Joel Selwood stepped up in scintillating fashion to drag the Cats over the line, this time against Melbourne. Their game-winning heroics are fast becoming the stuff of legend. But such has been their dominance since Dangerfield partnered up with Selwood last season it begged the question - is it better to have a champion team or a team of champions? Of course, every club in the land would love a one-two punch of 'Dangerwood' quality, but the evidence suggests having a more equal team - one with multiple paths to goal and several ball-magnets - is a more reliable way to win than relying on one or two superstars.

Inequality has been falling in the AFL for years, at least on the field. Teams are sharing the ball around - and sharing the shots at goal - much more than their predecessors did in the 1990s. To measure this inequality in disposals and scoring shots, we can use the Gini coefficient -- a common measure that is equal to 100 if a single player had all the disposals or scoring shots, or equal to 0 if the ball was shared completely equally among a team's players. A higher value means more inequality.

Disposals and scoring shots are shared more equally within teams than in the past:


The trend has clearly been towards more equal use of the ball within teams - both disposals and scoring shots are less concentrated than they were in the past. This trend has developed for seemingly good reasons: sides that share the ball around more tend to win games more often. On average, the more an outfit shares the ball around in a given game, the better the margin is likely to be for them.

But maybe you're not convinced that disposals are a good measure of inequality within a side. In that case, how about Brownlow votes? There, too, the stats suggest that more equal teams tend to fare better. A club like the 2011 or 2012 Suns, in which their best player received more than half of the team's total Brownlow votes, tends not to be very good. On average, if a team's best player gets a bigger share of its Brownlow votes, it tends to win fewer games.

Teams with more concentrated Brownlow votes tend to win fewer games:


Again, the striking thing here is that this relationship - in which more equal teams tend to do worse - holds up even when we statistically control for the team's overall quality. If we take into account both the total number of Brownlow votes that a team received, as well as the share of those votes that went to the team's biggest vote-getter, we find that clubs that win more votes overall tend to win more games, as you'd expect; meanwhile if a higher share of those votes go to the top player, this tends to reduce the club's winning percentage. If a team wins an extra Brownlow vote, it's probably a good sign for the club if it goes to a mediocre team member rather than a superstar.

This chimes with the record from 2016, in which the cellar-dwellars tended to have a high share of their Brownlow votes going to their top vote-getter. The only good team in which a large share of the votes went to the best player was the Cats, with Dangerfield snagging nearly 42 percent of their votes.

Teams that are more even seem to do better, on average. We see an interesting example of this if we compare the 2016 premiers to the wooden spooners. Last season, the Lions' best two players performed better than the Bulldogs' best, at least as measured by AFL Fantasy points. But the further we look down the list of the teams' best 22 players for the year, the bigger the gap grows in favour of the Bulldogs. The difference between the top and bottom teams doesn't seem to be about the quality of their superstars, but about the quality and contribution of the supporting cast.

The best 22 players of the 2016 Lions and Bulldogs:


It makes sense that a more balanced team would have more success. In AFL football, with 22 players on each side roaming freely around almost absurdly large grounds, no single player can hope to have the impact of a superstar basketballer or soccer player. More balanced outfits are more resilient to injuries, less threatened by health woes of their top ball-getters. Teams that don't rely on a small number of superstars have more paths to goal and are less easily shut down, with their midfields less susceptible to tagging tactics.

Research from other sports also backs up the idea that more equal teams tend to do better, at least in interdependent sports like footy. Researchers have found teams with higher player quality, but more evenly distributed quality, tend to win more games. Meanwhile psychologists have found evidence that there can be such a thing as 'too much talent' for a team, a point at which adding top-end talent can detract from team dynamics.

So which teams are more unequal on the field? So far this year, Richmond stands out for having moved in the direction of more equality. They're sharing the ball around, and sharing the scoring shots, much more than they did in 2016.

Inequality in disposals and scoring shots by team:


So, what does this all tell us? Of course, every club wants to produce and/or recruit match-winners but teams looking to bolster their prospects in the coming years might want to pay attention to the tail end of their lists. Superstars can't do it by themselves.

Matt CowgillMatt Cowgill is an economist by day and footy stats obsessive by night who thinks there's no such thing as too many graphs. He uses data analysis to tell stories for ESPN's Footy Forensics, as well as his own blog The Arc.

More Stories