Four years ago, Trump won the presidency on the margin of three states—Michigan, Wisconsin and Pennsylvania—by a combined 77,744 votes. It was, to use poker terminology, an “inside straight,” the sort of hand that seemingly comes together more by luck than by skill.

The story isn’t that Trump’s luck appears to have run out. The story is that Biden is poised to beat him with the same exact hand.

Wisconsin has counted all of its ballots, and the final result puts Biden ahead by roughly 20,500 votes—almost the mirror image of Hillary Clinton’s loss in Wisconsin by 22,748 votes.

In Michigan, the final ballots from heavily Democratic precincts in Detroit are still being tallied. Those will add some cushion to Biden’s current lead of 120,000 votes. Whatever the final number, Biden seems destined to win the state by about a couple of percentage points, four years after Trump carried Michigan by just 10,704 votes.

Pennsylvania will take the longest of the three to complete its count. As of Wednesday night, Trump still clung to a shrinking lead of some 190,000 ballots. But with nearly a million ballots still outstanding—the lion’s share of them coming from dark blue precincts in Philadelphia and elsewhere—Trump’s lead is expected to shrivel over the coming 24 hours and then disappear completely. How much Biden ultimately might win by is unclear. But it may not be by much more than Trump’s victorious margin four years ago of 44,292 votes.

In studying the election results from these states, I’m fascinated by some of the similarities to the 2016 election. But I’m all the more struck by key divergences between the two campaigns, and how movement at the margins with certain voting groups will be the difference between a second term for Trump and a new Biden administration.

Here are three reasons Democrats are poised to sweep the three critical Midwestern battleground states and win back the White House.

1) Biden kept Trump from running up the score with working-class whites

No single location has received more political attention this year than Scranton, Pennsylvania. It’s Biden’s birthplace, a city that’s integral to his brand as a scrappy, middle-class, train-riding everyman. It’s also an ancestral Democratic stronghold: Scranton, and surrounding Lackawanna County, is the embodiment of the old school, labor-anchored Democratic coalition. This is why both Trump and Biden made regular stops in the area, lavishing attention on voters there and saturating the local media market with ads and earned media.

In every campaign spanning Reagan’s reelection in 1984 and Barack Obama’s reelection in 2012, Democrats carried Lackawanna County by comfortable (and sometimes huge) spreads. And then Trump came along. He didn’t win the county, but he closed the gap, from a 27-point Democratic victory in 2012 to just a 3-point victory in 2016. That neck-snapping swing of 24 points caught the attention of every pol in Pennsylvania.

Trump entered his reelection bid with every confidence that he would perform even better in Lackawanna County this year. His team envisioned adding another 5 points or so to his 2016 showing, officially flipping the county and eating further into the Democratic Party’s dwindling base of white working-class supporters.

But the opposite happened. Biden won Lackawanna County by 8 points. Instead of gaining 5 points on his 2016 performance, the president lost 8 points.

With Pennsylvania once again destined to be decided a percentage point or less, every one of those lost votes in Lackawanna County will haunt the Trump campaign.

It’s easy to chalk up these specific results to Biden benefiting from a home-field advantage. But that would fail to explain what happened in Macomb County.

Another symbolically important location, the blue-collar suburbs of Detroit figured to test the president’s theory of how he could grow his vote in 2020. Fully expecting losses among college-educated voters in upscale areas, Trump’s team was determined to drive up bigger margins in the middle-class communities like those in Macomb County.

Home to the fabled “Reagan Democrats” of the 1980s, Macomb has drifted steadily rightward over the past three presidential elections. Democrats carried it by nearly 9 points in 2008 and by 4 points in 2012, only to watch Trump dominate the county with a 12-point win in 2016. There was real reason for optimism, among Republicans in southeast Michigan, that Trump could add another 3 or 4 points to that spread, padding his margins in friendly territory.

Instead—once again—the opposite happened.

Trump won Macomb County by 8 points, losing 4 points off his 2016 total. This was arguably the most surprising result in Michigan, and it was highly symbolic to boot: The president’s failure to match or exceed his 2016 performance, in a county tailor-made to his politics, was part of a broader letdown in his efforts to juice white working-class votes across the board.

2) Biden peeled away Trump’s support in conservative suburbs

The suburbs outside of Milwaukee constitute the most conservative metropolitan area in the country. Each of the three counties that surround the city—Waukesha, Ozaukee and Washington, the “WOW” counties—have voted Republican in every presidential election since 1968, and by double-digit margins. These counties are each wealthy, exceptionally well-educated and north of 90 percent white.

In 2016, the president carried the WOW counties by yawning margins. Washington was decided by 40 points, Waukesha by 27 and Ozaukee by 19 points. (The Ozaukee result was particularly interesting: It was the tightest race in generations, and yet, no Democrat had broken 40 percent of the vote there in a half-century.)

Four years later, Biden closed the gap in all three. Trump won Washington by 38 points, Waukesha by 21 points and Ozaukee by 12 points. Biden’s vote share in Ozaukee? You guessed it: 43 percent.

In a vacuum, those totals might not seem noteworthy. But taken together—as a picture of suburban Milwaukee and as a wider snapshot of wealthy white suburbs across the Midwest—they are the difference between a President Trump and a President Biden.

According to the final (though yet uncertified) election results in Wisconsin, Biden carried the state by just over 20,000 votes. With a margin of defeat that narrow, any number of things are key to the outcome. But it’s tough to argue that Trump’s slippage in the Milwaukee suburbs wasn’t the essential ingredient of his defeat in Wisconsin.

The same reality was apparent in Michigan.

Livingston County, which contains the exurbs of Detroit and Ann Arbor, has long been the most reliable source of Republican votes in the state. Four years ago, Trump carried the ultraconservative county by 30 points, and his team hoped to match that performance in 2020.

There was, however, reason to doubt this. Certain bedroom communities in Livingston County have grown more prosperous, attracting more advanced-degree households and high-income earners, over the past decade. Republicans wondered whether Trump’s struggles in more generic suburbs would carry over to deeply, fundamentally conservative suburbs like those in Livingston County.

Sure enough, Trump carried Livingston County by 22.5 points against Biden. It’s a healthy margin, no doubt, but it’s a dramatic falloff from his showing against Clinton four years ago.

Margins matter in tight races. The story of 2020, in the Midwest and elsewhere, was Biden whittling down the president’s margins in the conservative suburbs where Trump’s team thought he might be safe.

3) Biden got Black voters to turn out in big numbers

Some things in politics are pretty straightforward. This is one of those things: Clinton lost to Trump because she did not mobilize Black voters.

This was true across the battleground map. But it was especially conspicuous in the three determinative Midwestern states, not only because of their photo-finish results but because of the sizable Black populations in the biggest cities of Wisconsin, Michigan and Pennsylvania.

Let’s start with Milwaukee and surrounding Milwaukee County, home to the biggest share of Black voters in Wisconsin. In 2012, Barack Obama won roughly 328,000 votes in Milwaukee County. Four years later, Clinton won fewer than 289,000 votes in Milwaukee County. The challenge for Biden wasn’t necessarily to get all the way back to that Obama 2012 number; rather, at the bare minimum, it was to split the difference between these figures. He did that and then some: With all the votes counted, more than 317,000 people in Milwaukee County voted for the Democratic ticket, and Biden needed every single one of them.

It was a similar story in Detroit, a city that’s more than 80 percent Black, and surrounding Wayne County. In 2012, Obama won nearly 596,000 votes in Wayne County. Four years later, Clinton won fewer than 520,000. Once again, the question in Michigan—as in Wisconsin—was whether Biden could push that figure somewhere close to that Obama 2012 number, even if it was unrealistic to get all the way there. In fact, Biden might just get all the way there. As of this writing, roughly 15 percent of Wayne County’s ballots are still outstanding. But Biden has already won 568,000 votes there, far surpassing Clinton’s performance from 2016.

Finally, we have Philadelphia, a city with a plurality of Black voters, and surrounding Philadelphia County. The case against Clinton was less cut-and-dried there. In 2012, Obama won some 557,000 votes in Philadelphia County, and Clinton actually passed that mark in 2016, winning 584,000 votes there. That said, a closer examination of precinct-level data revealed that Clinton’s strong turnout came in whiter and wealthier precincts of the county, rather than its working-class and less affluent neighborhoods. Biden’s team knew that he would need both in order to best Trump in 2020. While there’s still a ways to go in the counting, it appears Biden will blow past both the Obama 2012 and Clinton 2016 numbers in Philadelphia County: He has already banked 458,000 votes, and with hundreds of thousands of votes from the area still outstanding, he figures to get well into the 600,000-vote range.

Using only the biggest cities and counties gives an incomplete window into Black voter turnout, but the numbers track with smaller Black-majority areas as well. In Flint, Michigan, for example, and surrounding Genesee County, Democrats went from winning nearly 129,000 votes in 2012 to some 103,000 votes in 2016. This year? Biden topped 120,000 votes in the county.

To win the presidency, Biden never needed Obama-era levels of turnout and support from Black voters. He just needed significant improvement on the performance of Clinton in 2016.

He has gotten exactly that—and with it, more than likely, a four-year term as president.