There is a gap in the existing literature for models linking prices in prediction markets with those for financial markets. We bridge this gap using a model based on the assumption that a binary political event has a constant effect on the difference of the conditional expectations of financial prices, given that event. This leads to a model where returns can be partitioned into a political factor, driven by changes in the likelihood of an election outcome, and a non-political component. Contrary to the existing literature, this model is based on economic principles and applies in a general setting. This model is naturally extended to equities using the Fama-French 5 factors to model the non-political part of returns variance. We test the model for six elections and referendums from the US and UK. Strong support is found for the theory for four events, and weak evidence for one. The remaining election, the 2017 UK general election, does not appear informative for asset prices. An exploration of the political factor loadings reveals pleasing relationships between firm characteristics and political sensitivity. Internationalisation of revenue, location and nationalisation risk are found to be significant. This is consistent with the existing literature, as well as the idea that firms can diversify away from local political risk using offshore sales.