In 2016, I decided to create a election forecasting model for the State of Florida, to determine which political party was more likely to win that year’s election. The main theory behind the model is that voter registration by party can be used a a predictor of vote intention. Of course, there are more votes cast than there are people registered by political party (since there are a large number of voters who do not register with a political party). Therefore, the remaining votes from a previous election would be take into account who has turned out, and then split the residual vote among both parties.

Since I still use this predictive model for current political clients, I don’t disclose the formulas used in this model. However, during any job interview process, I will be more than happy to discuss the formulas and methodology behind the models