Our mission to help you navigate the new normal is fueled by subscribers. To enjoy unlimited access to our journalism, subscribe today.
As a businessman, Donald Trump fashioned himself as a star “closer,” a great negotiator who could charm, cajole, or bully players across the table into buying his West Side Yards in Manhattan, or surrendering a sweet deal on a trophy Atlantic City casino. And whatever you think of Trump as President, it’s hard to dispute that though he botched the first debate and antagonized seniors by dismissing the pandemic’s perils, he’s a hell of a closer. Even his critics acknowledge that Trump showed a manic thirst for combat in barnstorming the swing states, hitting as many as four battlegrounds a day, in the final stretch of the campaign.
But how successful were Trump’s raucous rallies in capturing the extra votes he’ll need to win reelection? This writer analyzed where and when Trump chose to hold his Make America Great Again extravaganzas to see how the polls shifted after Trump swooped down on those prizes, often multiple times. The conclusion: Trump made generally shrewd choices by campaigning extra hard in electoral vote–rich, must-win states such as Pennsylvania and North Carolina. He wisely bombarded states where he trailed but was just close enough to stand a decent shot at winning. On the other hand, he appears to have squandered a lot of exhausting, frenzied campaigning trolling for votes in places where his odds always looked barren and didn’t improve, notably in the upper Midwest.
At mid-afternoon on Election Day, it looks like Trump’s closing offensive dramatically lifted his odds in no fewer than six major swing states. In each, he either went from a deficit to a lead or narrowed a wide gap. He owes his gains—which right now, don’t look big enough to win him reelection—to the raw energy that enabled him to push so hard in so many places. The President took his bandwagon to so many places, so often, that he hit the ones that really counted plenty hard, even though he also wasted a lot of effort on hopeless territory.
What’s a good yardstick for determining where Trump was smart to campaign, given his standing at the time? Thomas Miller, a data scientist at Northwestern University, has developed a measure for determining which states are most important. The relative electoral importance is determined by what he describes as “a flexible machine learning model.” (Miller posts his election analysis on his site www.data-science-quarterly.com.)
Miller uses a blend of two variables: first, the number of electoral votes, and second, the volatility in how the odds are shifting between the candidates. The more the red and blue lines jump around, the more you want to target that state (more on why in a moment). Instead of relying on polls, Miller deploys the odds on the political gaming site PredictIt as a gauge. He argues that data from the voter surveys are out of date by the time they’re published, since they reflect opinions collected several days before. The political exchanges, he notes, have over time been better at forecasting the winner than the final, pre–Election Day average of polls. Miller also stresses that PredictIt hosts over 100,000 bettors who are wagering their own cash, and hence are dead serious about handicapping not whom they hope will win, but the candidate they predict will prevail.
In order of importance, Miller’s list comprises Arizona (11 electoral votes), Pennsylvania (20), Florida (29), Georgia (16), North Carolina (15), and Iowa (6). “If I were running a campaign, I wouldn’t want to guess where to go next; I’d want to know,” he says. “This measure shows where the candidate should go and not go to maximize their chances.” He notes that at the time he did the analysis about a week ago, Trump had always trailed in two of the states, Pennsylvania and Arizona, and weeks earlier was behind in Florida and North Carolina, even in Georgia, and virtually tied in Iowa. “You want to target states where first, you have a good chance or are slightly ahead. That means ruling out the ones where your chances are 35% or below, or where you’re 65% or above.”
The second requirement, he says, is that the states where you’re in the running also show lots of volatility. “In that case, you know that since the odds move around a lot, it’s likely that your campaigning can have a big impact on your chances,” says Miller. Hence, pounding the right issues can help candidates catch up where they’re behind, and protect or lengthen a lead where they’re ahead.
Trump held plenty of rallies in the five states on Miller’s list. It’s impossible to know for sure if those mainly open-air spectaculars swelled his chances. But it’s worth exploring how the places where Trump swooped down, and kept alighting, align with the states where he notched the gains that, for now, appear to have made the contest far closer.
From Sept. 30 to Nov. 2, Trump crisscrossed America to hold a total of 45 big campaign events in 12 states. Forty-one were what he called Make America Great Again rallies, optimistically relabeled “Victory rallies” starting on Oct. 23. Four were speeches, such as his Forgotten Fighting Men and Women address in North Carolina the next day.
Trump’s No. 1 target was Pennsylvania; he visited the Keystone State no fewer than nine times. During the final swing, his first visit came on Oct. 13, when his odds barely touched Miller’s 35% minimum. But Trump kept coming back, and the yawning gulf kept narrowing. The same day he emceed his next-to-last rally in Montoursville on Oct. 21, his PredictIt odds jumped from 40% to 45%. It seems that Trump’s relentless focus on Pennsylvania greatly improved his odds. But as of mid-afternoon Nov. 3, he’s still lagging 40% to Biden’s 60%. The bettors are wagering that Trump’s late surge in this must-win state will fall short.
Florida ranked second in the occasions Air Force One touched down for a Trumpian gala, usually staged at an airport. On Oct. 7, Trump was once again in a hole, trailing by around 45% to 55% for Biden on PredictIt. But Trump apparently saw Florida as a quicksilver territory where events and issues can quickly change votes. So over the next three weeks, he hit the Sunshine State seven times, from Orlando to Ocala to Miami to senior haven The Villages. By Oct. 19 he took the lead and never looked back. His odds have been stable for the past several days at roughly 60%.
Tied for fifth (we’ll get to three and four shortly) were Arizona and North Carolina at four visits apiece. Trump needs not one but both of them, along with Pennsylvania, to win. It’s because hitting that trifecta is such a long shot that Trump’s odds have never risen much over 40% on PredictIt and are in the high 30s in the mid-afternoon of Nov. 3. In the Tar Heel State, Trump was way behind as late as mid-October. But after his first October rally in Greenville on the 15th, his numbers improved. Trump drew even on Oct. 19, but the lead kept toggling practically by the hour from blue to red. So Trump picked up the pace with three more MAGA rallies in the last two weeks. They may have been the elixir that boosted his current standing to a much better, though still precarious, 55%.
Arizona followed a similar trajectory. In mid-October, Trump’s odds floated in the high 30s. But even then, betting on the Grand Canyon State was jumping all around, showing that good campaigning could cause big, swift swings. Trump responded by hosting four rallies, two each on Oct. 19 and 28. Right after the second rally, the odds tightened, and by the afternoon of Nov. 2, Trump took his first lead, advancing to 55% the morning of Election Day. In a bad sign for the President, he has slipped late on Nov. 3 to 47%. As Arizona goes, so goes the election. The same can be said of Pennsylvania and North Carolina. The rub for Trump is that he’s trailing by a modest margin in Pennsylvania, enjoys a slight lead in the North Carolina, and is watching his advantage of a few hours ago evaporate in the Arizona. Yet he must win all three.
Finally, let’s deal with numbers three and four, Michigan and Wisconsin. The Great Lakes State merited six visits from the President, while the Badger State venues greeted Air Force One on five occasions. Both are a case study in places not to lavish lots of energy. They’re traditionally Democratic strongholds. Trump won both by tiny margins in 2016, prevailing in Michigan by 0.23% and in Wisconsin by 0.77%. By the way, Trump also held two events in Michigan’s next-door neighbor Minnesota, another state he lost—by 1.5 points in 2016.
In early October, Trump stood at 27% in Michigan, touching 36% for a hot minute at the peak. He’s now at 33%. In Wisconsin, he started last month at around 30% and has pretty much flatlined since, and he’s fared even worse in Minnesota, where he’s been consistently stuck in the high 20s. Considering his knife-edge wins in Michigan and Wisconsin in 2016, and his ultra-weak status in early to mid-October, Trump was wasting his time in those three states. Trump and his team were obeying their hopes and emotions, not following the numbers, as Miller advocates.
Fortunately for Donald Trump, he had energy to burn, and he channeled that horsepower into enough of the right places so often that he may have turned a potential runaway into a horse race, though it’s a contest he’s unlikely to win. In the annals of campaigns, it’s a sight we’ll probably never see again.
After all, being a great closer doesn’t always make you a winner.
More politics coverage from Fortune:
USPS update: The latest on how absentee ballots are being counted in swing states
Florida votes to raise minimum wage to $15
The state ballot measures the business community should watch in the 2020 election
Trump fancies himself an incredible closer—but is it enough to make him a winner?
How Trump can and can’t use the courts to shape the election
This story was originally featured on Fortune.com