Tuesday featured the last federal special election before November’s midterms. And while it’s too close to call and Republicans appear likely to eke it out, it remains an ominous sign for the GOP.

Republican Troy Balderson leads by less than one point over Democrat Danny O’Connor in Ohio’s 12th District, where both President Trump and Mitt Romney won by double digits in the past two presidential elections. It was the second major special election in a row in which Republicans struggled in clearly conservative territory, after Democratic Rep. Conor Lamb’s win in neighboring Pennsylvania. The GOP also lost the Alabama Senate special election in late 2017, albeit for unusual reasons.

We all have a tendency to oversell the importance of a special election, given the dearth of actual races between Novembers. These are races in one-435th of the country, after all, and crazy things tend to happen in these races — bigger shifts than you'd see in a general election, for sure. There is also little practical difference between one party narrowly losing and narrowly winning a race in such a district. Winning is nice and serves as vindication of your ability to focus on one district; November affords no such specialization.

Sometimes special elections even suggest the opposite of what lies ahead. Democrats won a conservative-leaning Pennsylvania district in a special election just six months before the 2010 midterm. That November, they got destroyed.

But Tuesday’s result was merely confirmation of a long-building trend, and history suggests Democrats are headed for big gains, now that the next federal election on the calendar is the Big One, fewer than three months away.

Democrats have been far outperforming their 2016 presidential vote margin in almost every special congressional election between the 2016 and 2018 elections. The only races in which they didn't were both in districts where Trump's performance undersold the GOP's historic strength. Georgia's 6th District had featured the sixth-biggest drop from Mitt Romney's vote to

Trump's, while Utah's 3rd featured the biggest drop. In both cases, Democrats far outperformed their vote margins from presidential elections before 2016.

The trend is even more pronounced at the state legislative level, where Democrats have flipped lots of GOP seats and often overperformed their past voted totals by even more than in the races above.

Polls currently have Democrats leading by less than earlier this cycle on the generic ballot, which tests an unnamed Democratic House candidate against an unnamed Republican one. But the margin is still around seven points, which is about where it needs to be to possibly tip the House in the party’s favor, according to the Cook Political Report's David Wasserman. And that margin could grow as pollsters shift from measuring registered voters to likely voters, given Democrats have a clear enthusiasm advantage by most measures.

Perhaps the most significant development from a party perspective is that the GOP is dealing with a tougher map than it should be. That's because of retirements, which have plagued more than one-fifth of all Republican districts — a historically fast onslaught of GOPers calling it a career. And retirements generally make it much more difficult to hold a seat, given that you no longer have an incumbent on the ballot. Of the 10 Republican districts that Cook rates as likely to flip to Democrats, eight are open seats. Another 11 open seats are rated as “toss-up” or “lean Republican,” meaning about half of GOP-held open seats are ripe for the Democratic picking in ways they otherwise might not be.

History is also running against the GOP. Midterms are traditionally brutal for a president’s party, which loses an average of 23 seats over the past century, according to Gallup — a number that is exactly what Democrats need to reclaim the majority today. Combine that with the fact that Trump is more unpopular at this juncture than most recent presidents, and the math begins to make sense.

Perhaps the most compelling recent case for a Democratic wave, though, came from the Cook Report’s Amy Walter. She ran the numbers and found that we have tended to undersell the looming wave at this juncture in a midterm:

In July of 2006, The Cook Political Report rated just 14 GOP-held seats as highly vulnerable. By November, the number of GOP-held seats in danger had tripled to 43. We saw a similar pattern in 2010. In August of that year, we listed 36 Democratic-held seats as highly vulnerable. By November, the number of vulnerable Democratic-held seats had more than doubled to 78. On Election Day of 2006, Republicans lost 30 seats; Democrats lost 63 seats in 2010.


For example, between July 2006 and November 2006, we moved 24-GOP held seats from Lean/Likely Republican to Toss-Up (or worse). Between August and November of 2010, we moved 35 Democratic-held seats from Lean/Likely Democratic to Toss-Up (or worse).

In fact, of the 30 seats that Democrats won in 2006, 21 of them (or 70 percent), weren’t classified as the most vulnerable GOP-held seats in July. Almost half of the Democratic seats Republicans won in 2010 were classified as Lean or Likely Democrat in August.

That tells us one of two things: Either waves tend to build late, or things changed late in those election cycles. But it’s difficult to see what changed fundamentally late in those election cycles, given approval numbers for George W. Bush and Barack Obama didn't shift appreciably.

And the fundamentals for the 2018 election seem well established right now.