Tuesday, September 23, 2014

The AFL’s Brownlow Medal: The Problems of 3-2-1


Over the past day I have seen a lot of comments basically saying that the voting system for the Brownlow Medal is not a good way of picking the ‘best’ player in the AFL (yes, I know it is technically ‘best and fairest’ in the AFL … ) The voting system has bothered me a little bit over the years, but it was only tonight that I tried to think through exactly what its problems were.

In the Brownlow Medal voting system, umpires award three votes to the best player for the match, two votes for the second best player, and one vote for the third best player. By contrast, in many leagues around the world, the voting for the ‘best’ or ‘most valuable’ player often takes place through an end-of-season vote in which players receive votes for their performances over the season as a whole. Compared to this system, there are a few main problems that the Brownlow Medal system has in determining in the best player for the season.  These problems exist independently of who is doing the voting, whether it be umpires, coaches, players, or the media.

The first problem is that only three players are awarded with votes for each match. Hence, no distinctions are made between any players from the fourth best player through to the forty fourth best player. An end-of-season vote, on the other hand, would in theory make these distinctions.

The second problem is that three players have to be awarded votes for each match. The saying that it is easier to get votes on bad teams has sounded to me like a cliché, particularly since voting has shifted towards winning teams in recent years. But putting these voting biases aside it is actually true, even if its effect is often exaggerated. For a player of given ability playing against a given opposition, the player’s chances of getting votes will be better the worse his teammates are, given that three players have to get votes. Again, this should not theoretically be true in the end-of-season vote model.

The third problem is that players can only receive three votes, two votes, one vote, or no votes. Therefore, regardless of how much better the best player was than the second best player he will only receive one more vote (and the best player in each match, regardless of how good they were, will receive the same number of votes).

There are two potential advantages of the Brownlow Medal voting system, but they have to do with how the voting is done, rather than the 3-2-1 voting system. The first is that, by doing the voting immediately following the match, it increases the chance that the voting is a fairer reflection of how the player actually performed at the match, rather than being affected by imperfect recollections. The second is that the voting is done by people who were actually at the match. In an end-of-season vote it is often the case that voters will not have seen all of a player’s performances.

Putting all these points together a better system would be this, although it would be obviously time-consuming and unlikely to happen. Each week ‘knowledgeable observers’ (whoever they might be) could watch each match, and rate every player. This would keep the advantages of immediacy, and having voters who have actually watched the matches, while avoiding the flaws of the 3-2-1 system. On the other hand, having the AFL boss read out forty four votes for each match probably would not make for riveting television.

No comments: