US election forecaster: Our model predicts a narrow victory for one candidate

Democratic Party
Republican Party
270 votes to win
Donald Trump will win the US election by the narrowest of margins, The Telegraph’s election forecaster now predicts.
Using years of election results, demographic trends and both current and historical polling, our model forecasts that Trump is on track to win 270 Electoral College votes to Kamala Harris’s 268 votes. 
The model suggests that the Republicans are the most likely to win tightly fought races in Arizona, Georgia, North Carolina and Wisconsin.
However, the Democrats are on track to secure the Great Lake state of Michigan, as well as the Sun Belt’s Nevada. Ms Harris is also set to hold Pennsylvania with its 19 Electoral College votes.
The Telegraph’s daily forecast has placed Ms Harris ahead of Trump for the past two months. The crossover is yet another sign of the Trump campaign’s growing momentum. 
The predictor is just one of many other models run by analysts, pollsters and media outlets. In the past few days, The Economist, polling-aggregator FiveThirtyEight’s and statistician Nate Silver’s models have all shifted to forecasting Trump as the most likely winner. 
Election models are similar to polls, which have moved against Harris recently, but attempt to add more variables to the mix.

Democrat
Republican

Turnout, tactical voting and other factors can influence results dramatically. 
How states with large ethnic minority or well-educated populations have voted in the past might help inform how those with changing demographics vote today. 
To factor this in, The Telegraph’s model runs 2,500 different simulations, adjusting the influence of these and other factors in each case across every seat, to give a range of different results. 
But the variety of outcomes in this and other models highlights just how close the race remains.
Currently, in 70 per cent of those simulations, the Republicans win, with the Democrats winning the rest.
Wisconsin, Michigan, Nevada, Georgia and Pennsylvania were won by a candidate with a probability of 51 per cent or lower. 
Forecasts are, inevitably, heavily influenced by the quality of polling at any given point. However, using historical results will — in theory — help factor in the issues faced by pollsters in 2016 and 2020, when they underestimated the support of Trump.
Modelling and projections by Peter Childs and Thomas Lawrie. Further analysis by Ben Butcher, Meike Eijsberg and Ollie Corfe. Visual production and development by Connor Ibbetson.