After sorting through the Alabama results and comparing them to other 2017 special elections, I figured it was time to look ahead, so I did just that at New York.

[T]he [Alabama] results were entirely consistent with the pro-Democratic trend that has persisted throughout 2017’s special and off-year elections. That would have been the case even if Roy Moore had eked out a narrow win. Republicans can, as Donald Trump has done, rationalize this or that 2017 defeat as being an anomaly. But it is impossible to take an honest look at the overall pattern of 2017 contests without hearing the not-so-distant rumbling of a likely 2018 wave for Democrats.

Harry Enten conducted a comprehensive analysis of 2017 special elections — all 70 of them — taking into account the established partisan “lean” of the jurisdiction being contested.

“The Democratic margin has been 12 percentage points better, on average, than the partisan lean in each race. Sometimes this has resulted in a seat flipping from Republican to Democratic (e.g. in the Alabama Senate face-off on Tuesday or Oklahoma’s 37th state Senate District contest last month). Sometimes it has meant the Democrat barely lost a race you wouldn’t think a Democrat would be competitive in (e.g. in South Carolina’s 5th Congressional District in June). Sometimes it’s merely been the case that the Democrat won a district by an even wider margin than you’d expect (e.g. in Pennsylvania’s 133 House District last week).

“The point is that Democrats are doing better in all types of districts with all types of candidates. You don’t see this type of consistent outperformance unless there’s an overriding pro-Democratic national factor.”

The best elections to examine in order to figure out whether Democrats can win back the U.S. House in 2018 are the seven congressional special elections of 2017. Republicans won five and Democrats two (a winning percentage that’s not surprising since all but one of these elections were triggered by members of Congress joining the Trump administration). But as Enten notes, the average vote-percentage swing to Democrats from prior established partisan levels was 16 points. In a polarized electorate, that’s a large swing indeed.

In thinking about this pattern, keep in mind that the demographic groups most likely to vote Democratic typically don’t proportionately turn out for non-presidential elections, and particularly for special elections. There is a powerful trend under way.

While any single special congressional election is not necessarily predictive of future election results, in larger batches they are highly correlated to the next election coming down the pike. Enten looks at special elections prior to the last six midterms and finds that on average the partisan swing in the former is within three percentage points of the partisan swing in the latter. That would suggest a double-digit Democratic swing (or something close to it) in 2018.

If that seems extravagant, look at the congressional generic ballot (a simple polling question about which party the respondents would like to control the U.S. House), itself highly correlated with the national House popular vote. According to the RealClearPolitics polling average, Democrats currently have an 11-point advantage, the highest they’ve enjoyed since last year’s elections.

The question of exactly how big a margin in the national House popular vote Democrats would need to gain the 24 net seats required for control of the House is a difficult one. Political scientist Alan Abramowitz has just published an analysis of House elections dating back to 1946, which also takes into account the impact of GOP-controlled redistricting after 2010, and concludes that a Democratic win as small as four points could do the trick. David Wasserman of the Cook Political Report thinks a seven- or eight-point win would be necessary.

Despite the clear trends, there remain a lot of unknown variables as we head toward the midterms, most notably presidential approval ratings and retirements. But the current occupant of the White House has a highly polarizing approach to politics that almost certainly caps his approval ratings (which have never been above 46 percent in any event). And Republican retirements are definitely outpacing those of Democrats; 26 House Republicans are either calling it a day or running for other offices. There’s no telling where the much-rumored investigations of sexual misconduct by large numbers of congressmen will lead. But as Jonathan Chait points out, there are 219 Republican men in Congress as opposed to just 132 Democratic men, so the odds of net damage to the GOP (and to a GOP-controlled institution) are high.

There is more at stake next year, obviously, than control of the U.S. House. Thirty-six states will hold gubernatorial elections, and all but a few will hold state legislative elections. Partisan performance at the state level could have a crucial effect not just on the public policies of the jurisdictions involved, but on positioning for the next redistricting cycle, which will begin between 2020 and 2022. And even in Washington, Democrats now see an opportunity to win back the U.S. Senate, which would have seemed laughably impossible a year ago.

All in all, we will probably look back a year from now and see 2017 as a harbinger of a strong Democratic performance in the midterms. Its precise strength will determine whether Donald Trump enters the second half of his presidential term merely embattled or fully caged and cornered.