For nearly seven years, President Obama has pushed America leftward while relying less on persuasion than on power politics. He and his allies rammed Obamacare through Congress against the clear will of the American people, and he has since largely circumvented Congress—and the separation of powers—and has imposed his will through executive actions.

Nor does Obama’s resulting unpopularity seem to concern him much. Indeed, his entire presidency appears designed to test James Madison’s claim that “in all free governments,” “the cool and deliberate sense of the community” will “ultimately prevail over the views of its rulers.”

What Madison knew, however, is that a free citizenry has the benefit of elections—and Obama’s unpopularity has put his party’s intended successor at a significant disadvantage in the election that will take place in just ten months. For the historical evidence suggests that, the more popular a two-term president, the more likely it is that his party’s intended successor will win. And it suggests that parties holding the presidency for two consecutive terms only win a third term if the outgoing president is a lot more popular than Obama.

If the Democrats’ intended successor loses—and if a new Republican president shows anywhere near Obama’s level of determination—Obamacare stands to be repealed in 2017 and replaced with a limited-government alternative. And most of Obama’s constitutionally dubious executive actions stand to be revoked.

So, what does the historical evidence suggest will likely happen in 2016?

Focusing only on non-incumbents running for their party’s third consecutive term, the evidence suggests that a president needs to have about a +15 net approval rating at the beginning of an election year to give his party’s intended successor an even-money chance of winning.

Moreover, every five-point increase or decrease from that mark moves the expected popular-vote margin about one point for or against his party’s intended successor. The basic formula, therefore, derived from historical experience, is to take a president’s net approval rating in Gallup’s first poll released during a given election year, subtract 15 points from that figure, and then divide by 5.

That’s the predicted advantage or disadvantage in the popular vote for the intended successor from a president’s party. In other words, if a president’s net approval rating is +20 at the start of election year, then his party’s intended successor would be expected to win the popular vote by about 1 point: (20-15)÷5=1.

Let’s examine how this has played out over time. According to the University of California at Santa Barbara’s American Presidency Project, President Harry Truman’s net approval rating in the first Gallup poll released in 1952 was -43 points (23 percent approving, 66 percent disapproving). Based on the formula, Adlai Stevenson, the Democrats’ intended successor, should have lost by 11.6 points in the popular vote. He instead lost by 10.9, thus over-performing by 0.7 points.

President Dwight Eisenhower’s net approval rating at the start of 1960 was +49 points, so based on the formula, Republican Richard Nixon should have won the popular vote by 6.8 points. He instead lost by 0.2 points to Democrat John F. Kennedy, underperforming by 7 points. After Kennedy’s assassination, Lyndon Johnson’s net approval rating of +10 (in the first polling of 1968, the last full year of the 8-year presidency that he and Kennedy split) indicated that Democrat Hubert Humphrey should have lost the popular vote by 1 point. He lost by 0.7 points to Nixon, over-performing by 0.3 points.

The next non-incumbent running for his party’s third term (after Gerald Ford took over for Nixon and ran as an incumbent in 1976) was Republican George H. W. Bush in 1988. Based on Ronald Reagan’s net approval rating of +16 points in Gallup’s first polling of that year, Bush should have won the popular vote by 0.2 points. He won by 7.7, exceeding expectations by 7.5 points. Twelve years later, Bill Clinton’s net approval rating of +29 points indicated that Democrat Al Gore should have won by 2.8 points. He won by 0.5, underperforming by 2.3 points—while losing the electoral vote to George W. Bush. And Bush’s net approval rating of -32 points implied that Republican John McCain should have lost by 9.4 points. He lost by 7.3, over-performing by 2.1 points.

Tallying the results, in four of the six races, the popular vote was within 2.3 points of expectations. And in five of the six races, the expected winner did in fact win the popular vote. (Moreover, Nixon may really have won the popular vote in 1960, which would bring that tally to six-for-six.)

So, what about Obama? His net approval rating in the first Gallup poll released in 2016 was -6 points (45 percent approving, 51 percent disapproving). The expectation, therefore, would be that the Democrats’ intended successor will lose the popular vote by about 4.2 points: (-6-15)÷5=-4.2.

Of course, that doesn’t guarantee a Democratic loss. Nixon fared 7 points worse than expected in 1960, and George H. W. Bush fared 7.5 points better than expected in 1988. It is worth examining these two outlier results in a bit more detail.

In Nixon’s case, Eisenhower’s popularity was at something of a high-water mark at the start of 1960. By the last poll taken before the election, Ike’s +49 net approval rating had dropped to +26—which, using the same formula, would have indicated that Nixon should have won by 2.2 points, just 2.4 point off from what actually happened. (In George H. W. Bush’s case, Reagan’s popularity remained at +16 points on the eve of the election.)

Perhaps more importantly, as I have argued elsewhere, the evidence suggests that the cake is pretty much baked long before election year. By July 4 of year-6 of an incumbent’s presidency, voters mostly seem to have made up their minds about what direction the nation needs to head next. Indeed, in the last Gallup poll taken before July 4 of year-6, Eisenhower’s net approval rating was +22 points, and Reagan’s was +45 points—suggesting that Nixon should have won by 1.4 points (just 1.6 points off of what actually happened) and Bush should have won by 6 (just 1.7 points off of what actually happened). As for Obama, his net approval rating as of July 4 of year-6 was -9 points, suggesting his party’s intended successor will lose by 4.8 points.

But doesn’t all of this hinge on who the Republican nominee is?

To a degree, yes. However, the historical trends suggest that the specific choice of nominee doesn’t really matter all that much—at least in terms of winning the presidency. (Expertly fulfilling the responsibilities of the office is quite another matter.) And recent polling (which at this early stage should be taken with several grains of salt) suggests that, among four of the five leading GOP contenders—Ted Cruz, Marco Rubio, Ben Carson, and Chris Christie—there’s not all that much difference in how they would fare against Hillary Clinton, the leading Democratic contender. (The polling suggests ‎that Donald Trump would fare a bit worse than these four, but not dramatically so.) In other words, the historical trends will likely hold regardless of who the Republican nominee is—provided it isn’t‎ John Kasich.

So the eventual Democratic nominee appears to be starting 2016 in a 4-point hole, which is a big hole to climb out of. But the real test of whether Obama’s legacy will be reversed—of whether Obamacare will be repealed and replaced with a limited-government alternative—will come not in 2016 but in 2017. The former will merely set the stage for the latter, and the latter will decide who’s right: Madison or Obama.

Anderson, author of “An Alternative to Obamacare,” is a Hudson Institute senior fellow.