The forthcoming special issue of American Political Quarterly will describe six different election forecasting models. Pity it has turned down the model proffered by one Harvard academic who bases his predictions on the number of people who join bowling leagues.