We are conducting the autopsy of our General Election 2017 model. Last time we looked at our individual predictions for each party. The issue with those is that we have little say over the location of these predictions. For example, if the bookies believe a party will make X seats, then our model will to, even if polling says they will make significantly fewer or more.
This time, we are looking at an element of our model that is definitely in our court - the chances of parties winning individual seats. Specifically, we will be looking at how we convert bookies' odds into probabilities.