Most people in this country do not support Donald Trump and never have. As The New York Times recently reported, he “is the only president in the history of Gallup polling never to earn the support of a majority of Americans even for a single day of his term.” This simple fact should be the starting point for all analysis of the 2020 presidential election, but rather than stating it at the outset, most pundits and columnists ignore a crucial reality. In doing so, they distort their coverage and default to Trump.

I almost hesitate to point this fact out because we need to sustain a strong sense of urgency in order to win back the White House, but it is important to allocate time, energy, and resources in the most informed and effective fashion. What we think has to happen in order to win in 2020 will drive our efforts in 2019, so we need to have this discussion now. Mathematically and empirically, as we enter this election cycle, Donald Trump is not favored to win reelection. The president is not the front-runner.

In normal election cycles, incumbent presidents are seen as likely to win reelection by virtue of their name recognition and media attention. Trump has both in droves, but this is no normal election cycle. The country is currently engaged in a titanic struggle that is evocative of the 1860s Civil War era and 1960s civil-rights movement. CNN political analyst Ron Brownstein describes it as a battle between the “coalition of restoration” and the “coalition of transformation.” The essence of the struggle, as Brownstein explains, is whether one embraces or opposes the “relentless economic, demographic and cultural changes reshaping American life.”

Brownstein’s framework is more on point than the pervasive view that the 2020 contest is a boxing match in which victory will belong to the party that selects the most effective pugilist. What will decide the 2020 election is not how many blows one can land on Trump, but how many bodies one can get to the polls to vote him out.

If we fail to learn the right lessons from 2016 (and 2018), we will waste many millions of dollars, squander precious time, and fail to properly reassemble the coalition that twice voted to put a Black man in the White House. An essential fact to keep in mind is that Hillary Clinton won the popular vote by nearly 3 million votes. But her opponent won more electoral votes, so he must be in the stronger position today, right? Wrong.

Three things conspired to make the 2016 election a perfect and catastrophic storm, whereby the election was determined by a mere 78,000 votes in three states: Wisconsin, Michigan, and Pennsylvania (none of which, notably, gave Trump 50 percent of the popular vote). First, African-American turnout plummeted to its lowest level in nearly 20 years. Second, many Obama voters defected to the third- and fourth-party candidacies of Gary Johnson and Jill Stein. (The vote increase for Stein in Michigan from 2012 to 2016 was greater than the margin of difference between Clinton and Trump). Third, some Obama voters likely did vote for Trump, but not nearly as many as people think. More Obama voters switched to Johnson and Stein than to Trump, and Trump received fewer votes in Wisconsin than Romney did in 2012, troubling the popular perception that a wave of voters switched sides to swell the Republican ranks. Had any one of those three developments gone differently, Clinton would likely be president.

Obviously, since Trump is in the Oval Office, it is possible for him to win again. But is he favored to do so? No. Just as three key factors handed the election to Trump, three significant changes to the country and the electorate since 2016 now give the electoral advantage to the Democrats. First, the country continues to get browner by the hour (hence this administration’s maniacal focus on restricting and reversing immigration trends). Based on the analysis I did of census and government data for my book Brown Is the New White, there are 7,000 more people of color added to the US population every single day—as compared to just 1,000 whites—through a combination of births, deaths, and legal immigration. The Center for American Progress and other think tanks released a report last year showing that, if we had a do-over of the 2016 election, with every demographic group voting in 2020 exactly as they did in 2016, including third-party votes, Trump would lose, simply because there will be that many more people of color in the electorate in 2020. Roughly one-third of all eligible voters will be people of color next year, in what will be the “least white and most diverse electorate ever.”

The second changed reality on the electoral landscape is the tangible effect of the 2018 elections, which brought a new slate of Democratic leaders and a push to expand the electorate. Whereas the previous Republican governors of Wisconsin and Michigan worked overtime to restrict the franchise and aid and abet voter suppression, those states now have Democrats in their governor’s mansions (and African Americans in the lieutenant governors’ offices). The governmental apparatus can now be brought to bear to remove obstacles to voting instead of erecting them. In states that were decided by 11,000 (Michigan) and 22,000 (Wisconsin) votes, that’s a major development.

The work done to expand the electorate for the 2018 elections also created new battleground states out of formerly comfortable Republican territory. Democrats fielded historically strong gubernatorial candidates in Georgia, Florida, and Arizona last year, with the nominees all winning more votes than any prior Democratic gubernatorial nominee in their state. Although their candidacies fell agonizingly short of prevailing, the states saw major infrastructural changes: Florida restored the right to vote to people with felony records; Georgia was forced to overturn a harsh automated voter-purge system; and Arizona elected a secretary of state who supports automatic voter registration and has pledged to keep a close eye on the rolls. These changes will give Democrats a competitive edge in the three states, all of which are on the front lines of the demographic revolution. Moreover, the gubernatorial candidate standard-bearers in Florida and Georgia (Andrew Gillum and Stacey Abrams, respectively), among the party’s top rising stars, are focused on turning their 2018 supporters into permanent political forces.

The third development is that, finally, some whites who supported Trump have had enough. The 2018 election saw Republican support among white college-educated women drop from a 10-point GOP advantage to a statistical tie between the parties. That shift helped the Democrats take back control of the House. In determining political priorities and investment strategy, it is critical to realize that the data shows that the most movable white Republicans are white college-educated women, who many Democrats seem to have forgotten in their obsession over white working-class male Trump supporters.

To be sure, Democrats and progressives could still mess this up (I’m looking at you, Jill Stein and Howard Schultz). But the mathematical foundation is plain and simple: Trump starts out behind. The coalition of transformation is bigger than the coalition of restoration. The majority of people don’t support this president, and never have. If we can keep them from splintering or becoming dispirited, the Democratic nominee will become the 46th president in January of 2021.