THE result of the 2020 presidential election was the slowest to be called since 2000. Covid-19 restrictions, a mass switch to early voting, high turnout and tight margins in swing states led to four anxious days of vote-counting, nail-biting and Twitter-refreshing before Joe Biden was declared president-elect.

This year, given heavy early voting, many expect the counting will be slow again. Officials insist that ballot tallying will be faster. And although the contest is close—with six days to go, The Economist’s forecast model had it as a dead heat—there is a good chance of a decisive victory for either Donald Trump or Kamala Harris, due to a normal polling error. The results could be known just a few hours after polls close—as they were for seven of the past ten elections (see chart).

The first states to conclude voting will be on the east coast. Six states, including the key battleground of Georgia, will finish voting statewide at 7pm eastern time (midnight in London). By 8pm, 19 more states will have joined them and a flurry of data will be published. Readers should exercise caution: little of substance will be revealed at this stage of the night, unless the election is a landslide.

Exit polls will be published in states that have completed counting. Unlike such polls in many countries, the data will not include estimates of candidates’ share of the vote. Instead, these polls include information on the composition of the electorate, their policy views and top issues—none of which will reveal who has won.

In some states, where one candidate is heavily favoured, the election result will be called almost immediately. Unless there is a major upset or a striking trend, these calls may not say much about the election overall. One of the first states to be called in 2020 was Vermont, which Ms Harris is overwhelmingly likely to win. The absence of a call may be more informative: if Virginia is not called soon after polls close, it may indicate that Mr Trump is having a good night. The reverse is true for Ohio.

The first sets of counted votes are unlikely to reveal much, either. In many states, where large urban counties that lean Democratic are slow to count, the vote will appear more Republican than the final tally. In 2020 this effect was compounded in some states by mail-in ballots (which skewed Democratic) being slowest to count. Hence the “blue shift” phenomenon: Republican vote leads wiped out by late-counted Democratic ballots, fuelling false claims of electoral fraud.

So what will be the first solid pointers on election night? One metric to watch is the change between county-level results in 2020 and 2024 (this will appear on each state’s results page on economist.com). By comparing counties which have completed their tallies, we can measure the change in support for each party’s candidate.

For example, in a key state such as Pennsylvania—with 67 counties—the early results might come from a selection of counties that Mr Biden won by ten percentage points in 2020. Suppose those counties show Ms Harris winning by five points. If that shift were replicated across the state, Mr Trump would be on track to win Pennsylvania as a whole by four points (Mr Biden won it by one point in 2020).

When the first states conclude counting, we will get more clues as to how the election has panned out. Florida finished counting before midnight eastern time in 2020. Although the state is not likely to be competitive (our forecast gives Ms Harris a five-in-100 chance of an upset), it could still indicate who has the upper hand. Using simulations from our forecast, we can see how the result in Florida relates to Ms Harris’s chances of winning overall. If she loses Florida by seven percentage points, she has a one-in-two chance of winning the presidency. If she loses the state by more than 11 points, her chances of winning the election sink below one in five.

Both of these measures are imperfect. The first counties and states to tally their votes may be unrepresentative. In 2020 Florida moved two points towards Mr Trump whereas the country as a whole moved two points towards Mr Biden.

The final result will probably come down to seven key states. In our forecast, Ms Harris has a 93% chance of becoming president if she wins Pennsylvania, for example, and Mr Trump has a 95% chance if he wins Michigan. Of the seven states, Georgia and Michigan may be the fastest to count. Georgia has mandated that results from early voting (around 70% of Georgia’s total vote) must be announced by 8pm eastern time. Michigan has changed the law to allow the processing of early votes before election day, speeding up the tally compared with 2020. North Carolina is also traditionally quick to count but may experience disruption due to Hurricane Helene.

Others could well be slower. Pennsylvania, the most likely pivotal state according to our forecast, will not start processing millions of postal ballots until the morning of election day. Arizona and Nevada, in the west, finish voting later that day and take longer to count their mail-in ballots, which are popular in both states. Nevada accepts and counts ballots which arrive after election day, too (although these are unlikely to flip the state).

The timing of the final call will depend on how close the election is. In 2000, when the presidency was decided by just over 500 votes in Florida, it took weeks to determine the result. In 1984, when Ronald Reagan won by a landslide, the result was called at 8pm eastern time, while voters on the west coast were still casting ballots. A decisive victory for either candidate would reduce the opportunities for spurious litigation and election denialism—a pastime of Mr Trump’s which may slow the announcement of the final result.

Eyes on the prize

The median scenario from our forecast has Ms Harris winning her pivotal 270th electoral-college vote by less than half a percentage point. But there is also a substantial chance of a polling miss of a scale that would give one or other of the candidates a comfortable win. In one in six scenarios from our forecast, the winning margin in the pivotal state is greater than five points—matching Barack Obama’s re-election in 2012. If that were to happen, we would probably have a clear indication early in the night (the 2012 election was called before midnight eastern time). In three out of four forecast scenarios, the margin of victory in the pivotal state is larger than Mr Biden’s in 2020.

The tail risk of election-night becoming election-week or election-month is still significant. If the presidency comes down to a few thousand votes in Wisconsin or Pennsylvania—the central estimate of our forecast—it could take weeks to resolve. Election interference could extend the wait even further. But there is also a fair chance that the result is known sooner than many expect. Election-watchers, adjust your sleep schedule accordingly. ■

Stay on top of American politics with The US in brief, our daily newsletter with fast analysis of the most important electoral stories, and Checks and Balance, a weekly note from our Lexington columnist that examines the state of American democracy and the issues that matter to voters.