The 2020 election will play out in 10 states: five in the rustbelt and five in the sunbelt. Three are “blue wall” states that Donald Trump narrowly won and will need to win again. The other seven states are various shades of red and trending in different directions. As Democrats begin the long process of narrowing twenty-odd candidates to one, they would be well served to consider which potential nominees can best compete where the election will be won.

Every projection is based on assumptions, and I’m assuming that Democrats are likely again to win the 20 states Hillary Clinton carried (plus the District of Columbia). Some, like Minnesota and New Hampshire, were excruciatingly close but retain a stubborn Democratic lean. Others, like Colorado, Virginia and Nevada, once reliably red, voted Democratic in the past three elections and continue to trend blue on the strength of demographic change. Together, they total 232 electoral votes. You may or may not want to add the second congressional district of Maine (a state that assigns electoral votes by congressional districts), which Democrats narrowly carried in last year’s House contest, to get to 233.

Republicans can count on 20 states of their own in the Deep South, Great Plains, Rocky Mountain West and Alaska. Absent something dramatic, it’s difficult to see any of these states defecting, with the possible long shot exception of Nebraska’s second congressional district around Omaha (like Maine, Nebraska divides its electoral votes by CDs). These states total 126 electoral votes, over 100 less than the Democrats’ base.

That’s where things get interesting. The five northern-tier states depicted in green on the map to the left are Rust Belt states where Democrats have performed well in the past but lost to Donald Trump in 2016. They fared poorly last year in Ohio, losing the governor’s race and failing to make inroads to a highly gerrymandered House map. Democrats did better in Iowa, picking up two House seats, but they lost the gubernatorial contest, and things were very tight in Wisconsin. Their best states in the region were Pennsylvania and Michigan, which appeared to revert to pre-2016 patterns. Should this foreshadow next year, the addition of these two states would stretch the Democrats’ total to 269 — and a tie in the Electoral College. They would still need a win somewhere else.

That’s where the five green southern-tier states come in. They are historically Republican strongholds where demographic trends are edging Democrats closer to competitiveness — the next generation Nevada, Colorado and Virginia — but it is unclear how many are ready to flip next year. Florida is a perpetual swing state and is always hotly contested. North Carolina and Georgia are closely divided and suffer from suffocating voter suppression efforts. Texas has been an unsatisfying shiny object for a long time. Arizona holds the most promise following a convincing senate win last year.

This makes the Democrats’ initial electoral vote advantage more tenuous than it may appear. Although they begin on the cusp of victory, although the election will play out almost entirely on Republican turf, and although Donald Trump will need to win almost every state in play to be re-elected, it is not impossible to imagine entire regions moving as a bloc as they did in 2016. If we were looking at the projected electorate of 2024 or 2028, Democrats would hold a commanding position. But 2020 is a transitional electorate, much like it has been for the better part of the decade, with support in some traditional Democratic strongholds eroding before growth in future base states is fully realized. As their support recedes in the rustbelt it has not quite emerged in the sunbelt, at least not to the point where they can count on trading Georgia for Ohio or Arizona for Wisconsin. This means Democrats will need to compete in both regions and should look to nominate a candidate with appeal to older working class white voters in the north, younger multicultural voters in the south, and suburban women everywhere. As the campaign takes shape over the coming months, we will get a better sense of who that candidate might be.