The model was first brought to my attention by Jacques Black in his book Spread Betting to Win. He suggested a slight alteration to the model—cutting out the points awarded to a party for having an ethnic presidential or vice-presidential candidate. The basis for the alteration was that having a non-white Anglo-Saxon, middle-class candidate was more a hindrance than a help on the few occasions it has been seen.


That tweak would have meant the system would work for every US presidential election in 70 years and not many systems can claim a record like that. To start with the model weights the candidates according to their party—the Democrats base figure is higher than the Republicans to account for their higher number of registered party members and their built-in advantage in the Deep South of America. Other key variables are then factored in, including the state of the economy, the length of time a party has been in power, whether one of the candidates is incumbent and whether the USA is at war or not.


Whichever party scores highest is the one that wins the election. ‘The BBC website before the election had a program which would calculate the likely number of seats based on the percentage votes for the parties. I played around with it going on the numbers in the latest polls. Even the most optimistic prediction for the Conservatives was way outside what the spread firms predicted. I sold their seats for £150 a seat at various prices. I won £15,000.