Wednesday, April 18, 2012

Why the AFL Draft Seems To Work, Or Why There is Hope For Gold Coast and GWS

Recently on the excellent Wages of Wins Journal there have been a few articles explaining how “tanking” for higher draft picks – i.e. losing more games to move up in the draft order - doesn’t work in the NBA. This was proven by showing that NBA teams that are good one season are generally also the good teams the next season, and the season after that, and the season after that. Given that the “tanking” debate seems to rear its head every season in the AFL, I was curious to find out if this finding also applied to the AFL.

To work out the answer I calculated the correlation between a team’s winning percentage in one AFL season with their winning percentage the year before that. The correlation is some figure between -1 and +1; a high positive correlation means that the teams that were good in that season were also the good teams the season before it. I also then calculated the correlations for 2 seasons apart, 3 seasons apart, 4 seasons apart and 5 seasons apart. The results since 1940 are shown in the figure below. (To remove some of the volatility of the series I averaged each series using the corresponding correlations over the past five seasons.)

Unsurprisingly a team’s winning percentage in one season is positively correlated with its winning percentage the previous season – that is, the good teams in a season are likely to be the good teams from last season. Also unsurprisingly the strength of the correlation is reduced as the seasons become further apart, so the good teams in a season have often been the good teams from five seasons ago but the relationship is not as strong as it is for the “one season apart” series.

What is interesting though is that the strength of the correlations has been historically low from about the 1997 season onwards. Indeed, once we get to the mid-2000s there are negative correlations for the 3, 4 and 5 season aparts series, indicating that the good teams in a season were the bad teams from a few seasons ago. If one postulates that teams did not begin to use the draft well until the mid-1990s (it was instituted in 1986) then this could be taken as an indication that the draft has helped bad teams to become good. And if this is the case then that gives hope to Gold Coast and Greater Western Sydney that their high draft picks will pay off for them in a few years.

No comments: