(CNN) The battle between the parties for control of the closely divided US Senate has become a contest between the exception and the rule.
The rule is that over the past quarter century, it has grown much more difficult for even the most talented Senate candidates to win in states that usually vote for the other party in presidential races. That trend, which is affecting both parties, has created a very narrow balance of power in the Senate -- and triggered frequent changes of partisan control -- because each side has established a natural advantage at the presidential level in about half the states.
The paradox is that because the balance of power between the two parties is so precarious, the Senate majority now often tips based on the few exceptions to this pattern -- the candidates who find ways to win, in effect, behind enemy lines in states that usually back the other party's presidential nominees.
This year, that test will fall primarily on the five Democratic senators seeking reelection in states that have invariably voted Republican for president in recent years. Five other Democrats are trying to hold seats in states that have been more evenly balanced between the parties, but which tipped to Donald Trump in 2016. By contrast, Dean Heller of Nevada is the only Republican this year defending a Senate seat in a state that Democrats have won in any of the past three presidential elections.
Democrats have an outside chance of regaining the majority in November if they can win in Arizona and Tennessee, states that have leaned reliably Republican in recent presidential elections, but which may be offering new opportunities for the party amid the backlash to Trump among suburban voters and minorities alike. But given how many seats in red-leaning terrain Democrats must defend, the more achievable goal for them this year may be to avoid losses that put the Senate out of reach in 2020 when they will have more pick-up opportunities in states that lean their way in presidential races (Colorado and Maine), are already presidential battlegrounds (North Carolina and Iowa), or may be evolving in that direction (Georgia).
The increasing alignment of states' presidential and Senate preferences reflects the broader decline in split-ticket voting over the past several decades. At its height in the 1970s, about one out of four voters typically voted for one party in the presidential election and the other party in Senate races, according to an analysis by Emory University political scientist Alan Abramowitz.
But the increasingly partisan, even tribal, nature of modern politics has made split-ticket voting much more rare. More voters have started to treat both Senate and House races as quasi-parliamentary elections, viewing them less as a referendum on the merits of the individual candidates than as a choice on which party they prefer to control the chamber. The result, as I have written, is that for congressional candidates, the name on the back of a candidate's jersey now usually matters less than the color on the front of it.
That trend becomes vividly apparent when looking at Senate results through the lens of how states have recently voted for president.
One place to start is 1992, when Bill Clinton's first election shattered the GOP's dominance of the White House over the previous quarter century, and coalesced a new partisan alignment in which Democrats have carried the popular vote in six of the past seven presidential contests (though they have won the Electoral College, and thus the White House, just four times.)
Over those seven elections starting with 1992, 24 states have voted Democratic at least four times. Democrats now hold 40 of their 48 Senate seats, or 83.3%. Over those seven presidential elections, 26 states have voted Republican at least four times. Republicans hold 43 of their 52 Senate seats, a virtually identical 82.6%.
It may be somewhat outdated to characterize states based on their preferences since 1992, however, since Clinton managed to hold a number of blue-collar Southern and border states, including Arkansas, Kentucky, Louisiana and Tennessee. Each of those states has grown more reliably Republican since, as the GOP has deepened its hold on older, non-college educated, evangelical and rural white voters. Meanwhile, Colorado, Virginia and Nevada have migrated more reliably into the Democratic camp in presidential contests, amid growing racial diversity and declining Republican strength among well-educated whites.
Comparing the party's Senate results to the presidential outcomes over the past three presidential elections may offer a more revealing map of the opportunities and challenges ahead for the parties.
Measured against that yardstick, the relationship between presidential and Senate preferences remains very powerful. In the three presidential elections starting with 2008, 20 states have voted Democratic all three times: Democrats now hold 37 of their 40 Senate seats. The only Republicans left in these states are Heller, who faces a tough reelection battle this fall, as well as Cory Gardner of Colorado and Susan Collins of Maine, who will both face reelection in 2020.
Six more states voted twice for Barack Obama but then flipped to Trump in 2016. The parties now evenly divide those states' Senate seats, holding six each. But Democrats in November will be defending five of their six seats in this category. In descending order of vulnerability, they include Bill Nelson in Florida and Tammy Baldwin in Wisconsin, followed by Sherrod Brown in Ohio, Bob Casey in Pennsylvania and Debbie Stabenow in Michigan. Republicans are not defending any of the seats they hold in these states.
This is partially a matter of timing. Senators serve for six years in a term and Democrats are defending, for the second time, a number of the seats they won in 2006, when they took control of the Senate from Republicans. They lost control in 2014, when Republicans beat several Democrats who had been swept to victory during Barack Obama's resounding win in 2008. Republicans will defend their 2014 gains in 2020.
Moving to the Republican side of the line, Indiana and North Carolina voted for Obama in 2008 but have backed the GOP candidate in the two elections since.
Republicans already hold three of their four Senate seats and Democrats this fall will be defending their sole beachhead in this group: Joe Donnelly in Indiana.
The main event in November's Senate election will unfold in the 22 states that have voted Republican in each of the past three presidential races. Republicans already hold 39 of those states' 44 Senate seats (almost 89%, or just slightly below the Democratic share in the 20 states they have carried three straight times).
But Democrats this year must defend four of their five seats in this group. In descending order of vulnerability they include Claire McCaskill in Missouri, Joe Manchin in West Virginia, Heidi Heitkamp in North Dakota and Jon Tester in Montana. The twist is that even as Republicans press those opportunities, they must defend Senate seats in at least three other states they have won each time since 2008: Democrats are mounting strong challenges for open Senate seats in Arizona and Tennessee and have found a dynamic, though still long-shot challenger, in Beto O'Rourke in Texas.
Even Mississippi might become a long-shot target for Democrats if Tea Party insurgent Chris McDaniel wins a primary later this year against Cindy Hyde-Smith, the more mainstream Republican that Gov. Phil Bryant recently appointed to succeed retiring Republican Sen. Thad Cochran.
The long-term trajectory in these numbers is toward an unmistakable alignment between the way states vote in presidential and Senate elections. Yet because the parties hold a presidential edge in about the same number of states -- with perhaps just slightly more leaning toward Republicans at this point -- it's the small deviations from the overall trend that may decide Senate control in the years ahead.
Three factors seem most important in shaping those exceptions and determining which side controls the upper chamber.
One factor is which side wins what are often very close races in the remaining battleground states where neither side has established a consistent advantage at the presidential level. (I define those as the states that have voted two of the past three for either party, with the exception of Indiana, whose 2008 vote for Obama looks more like an anomaly over time.). Democrats missed a golden opportunity in 2016 to strengthen their hands in these battlegrounds when they failed to beat any of five Republican incumbents in Wisconsin, Ohio, Iowa, Pennsylvania and Florida -- partly because of Hillary Clinton's weakness, especially among blue-collar white voters, in those states. This year, Republicans face a comparable test against the five Democratic incumbents in the states that switched from Obama to Trump.
The second key factor in shaping exceptions is candidates who can either slow or advance the clock in their state's political evolution. The candidates who have stopped the clock are Democrats like Manchin, Heitkamp, and McCaskill, and Republicans such as Collins, who are holding on in states that have long since tilted to the other side in presidential races. None are guaranteed success in their next election, but it's hard to imagine anyone else from their party holding their seats in the near future after they have left the Senate.
Those who advance the clock are candidates who can establish an advance beachhead in states whose loyalties may be shifting at the presidential level. Kyrsten Sinema will try to do that for Democrats in Arizona this year, and the party is likely to make a serious run in 2020 at Republican David Perdue in Georgia. Conversely, it's easy to imagine Republicans making a big push in 2020 for a Senate seat in Minnesota, whose heavily blue-collar demographics will encourage Trump to vigorously contest it, though no Republican presidential nominee has won there since Ronald Reagan in 1980.
Finally, timing matters. Incumbent senators from the party out of the White House have rarely lost in midterm elections since the 1990s. Abramowitz notes that in the 2006, 2010 and 2014 midterms, the only incumbent senators who lost were from the president's party.
"If we were heading toward this midterm with Hillary Clinton in the White House most likely Democrats would be looking at losing several of those (red state) seats," he says. "The thing that is helping them out a lot is it's a midterm with an unpopular Republican president in the White House."
In states that Democrats are defending such as West Virginia, North Dakota, Missouri and Indiana, Trump is more popular than he is nationally. But he may not be quite as strong even there as he was in 2016. And that small shift may be just enough to allow some of the red state Democrats to defy the mounting odds for senators trying to survive behind enemy lines.