Since its introduction in the early 1970s, the Mackerras pendulum has been used to predict overall election outcomes by looking at what would happen if the overall swing towards one or the other major party were uniform across every seat.
The electoral pendulum is a tool for predicting net seat changes, but it does not make notably better predictions than other tools make. It also tempts commentators and politicians to misuse it to predict results in individual seats, where its predictions are more likely to be wrong than right.
This paper argues that Australia's electoral pendulum performs no better than an alternative method (the cube law) in predicting the overall result of an election. In its common, alternative use as tool to predict individual seat changes, it is successful less than half of the time.