Former President Trump has a 2-3 chance of winning the White House in November, according to The Economist’s election forecast model launched Wednesday.
The model predicts Trump has a 66-100 chance of winning, while President Biden has a 33-100 chance of winning.
The results are similar to a forecast model from Decision Desk HQ and The Hill released late last month. In that model’s most recent update, Trump holds a 56-100 chance of winning the presidency, while Biden has a 44-100 chance.
The Economist’s forecast model, which is set to be updated daily, puts the likelihood of a Trump victory at the highest level since the beginning of March — the earliest daily prediction The Economist included retroactively in its forecast model.
For about three months, Trump’s chances of winning were in the mid-to-high 50s, occasionally reaching 60 or 61. Biden’s chances, conversely, were mostly in the low-to-mid 40s, occasionally dropping down to 39 or 40.
As opposed to national polling averages — which suggest a toss-up race between Trump and Biden — the forecast model looks at the electoral college vote map and calculates the likelihood of a candidate reaching the necessary 270 votes to win.
The Economist model identifies six battleground states, worth 77 electoral votes, that will be decisive in November, noting Trump won five of the six in 2016, and Biden won all six in 2020.
Four states are categorized as “uncertain,” meaning the likelihood of a Trump victory in the state is less than 65 in 100: Nevada (64 in 100), Pennsylvania (63 in 100), Wisconsin (59 in 100) and Michigan (56 in 100).
The model predicts a Trump victory is “likely” — meaning 65-85 in 100 — in Georgia (76 in 100) and in Arizona (69 in 100).
The Economist notes that each of those six states has a varying degree of importance to each candidate. The model predicts the likelihood of Pennsylvania deciding the election outcome is 24 percent, and the likelihood for Michigan is 14 percent. The other states are below 10 percent.
For Biden, victories in Michigan, Pennsylvania and Wisconsin are especially important to win the electoral college. The model predicts Biden has a 5-100 chance of reelection if he loses Michigan, a 7-100 shot if he loses Pennsylvania and a 9-100 shot if he loses Wisconsin.
A loss in Arizona or Nevada would bring Biden’s chances down to 20 in 100, while a loss in Georgia would bring his chances to 24 in 100.
Pennsylvania will be key for a Trump victory as well. A loss in Pennsylvania brings the likelihood of his victory down from 66 in 100 to 21 in 100. A loss in either Michigan or Wisconsin would bring the likelihood down to 30 in 100. An Arizona loss would mean he still has a 35-100 shot, a Georgia loss gives him a 37-100 shot, and a Nevada loss would give him a 42-100 shot.
The model combines national and state-level polls with various other data points — including historical voting patterns, economic indicators and demographic information — which have proven to be predictive in previous presidential election cycles.
The model then runs 10,001 scenarios, which each contain “different vote shares in each state and different values for the impact of polling biases and other characteristics.” The final determination reflects the percentage of scenarios each candidates won.
This story was updated at 11:49 a.m.