This talk will review the experience of attempts to forecast the 2015 and 2017 British general elections and the 2016 Brexit referendum. I will discuss both my own forecasting methods and those of others. The talk will consider issues with the opinion polls that under-pinned many of the forecasts and the extent to which forecasts still rely on opinion polls or whether other data sources, such as betting markets, are more effective. Most of the talk will be accessible to a general audience but there will be some points of detail on the statistical models.