Democrats Lead the Generic Ballot Average by About Seven Points, and History Says the President's Party Usually Loses House Seats

What the field shows now

With three months left before the November 3 midterms, the party out of the White House leads the generic congressional ballot in the polling averages, and the size of that lead is about seven points. Nate Silver's Silver Bulletin average read Democrats plus 6.3 in early July and has stayed near there since (Silver Bulletin, July 2026). The Race to the WH average put it at Democrats plus 7.0, 48.1 percent to 41.1 percent (Race to the WH, August 2026). One recent survey, from Emerson College, went further, 53 percent to 42 percent, an 11 point margin (The Hill, 2026).

This site treats numbers like these the same way every time. One poll is a rumor. The field is the fact. The 11 point Emerson reading is the outlier at the edge of the range. The average near seven is the signal, and the honest way to state it is as an average, not as any single headline.

The president's approval, read all the way down

The other number driving the commentary is presidential approval. The Silver Bulletin average had President Trump near 38 percent approval and 59 percent disapproval in early August, a net around minus 21 and a low point for his second term (Silver Bulletin, August 2026). About 21 percent of Americans said they strongly approve. Among independents, approval ran near 34 percent (Silver Bulletin, August 2026).

Those figures point in one direction, and it would be dishonest to pretend otherwise. It would be equally dishonest to call them a forecast. Approval is an input to a midterm result, not the result. What the record says about that input is worth laying out plainly, because it cuts against the easy read in both directions.

The pattern older than this president

The reason a seven point lead and a 38 percent approval rating do not settle the fall is that the midterm has its own gravity, and it pulls on every president. In the 22 midterm elections from 1934 through 2018, the party holding the presidency lost House seats in 19 of them, and lost an average of 28 seats (American Presidency Project, 2022). That is not a claim about one party. Bill Clinton's Democrats lost 54 House seats in 1994. Barack Obama's Democrats lost 63 in 2010. Donald Trump's Republicans lost 40 in 2018 (American Presidency Project, 2022). Same ruler, different years.

Approval sharpens the pattern. Gallup found that presidents sitting below 50 percent approval at the midterm saw their party lose an average of 37 House seats, while presidents above 50 percent lost an average of 14 (Gallup, 2018). The measured correlation between a president's approval and his party's House seat loss is 0.66 (American Presidency Project, 2022). A 38 percent approval rating sits on the wrong side of that line. The history says the party in power should expect to lose ground. It does not say how much.

The honest other side of the same record

Read the full table and it holds the exits as well as the warnings. The president's party has gained House seats at the midterm three times since 1934, in 1934, 1998, and 2002, each time on the strength of an unusual national moment (American Presidency Project, 2022). And a low approval number does not fix the scale of a loss. In 2022, with the president's approval near 42 percent, his party lost only nine House seats, far fewer than the model would have guessed (American Presidency Project, 2022; official House returns).

Summer leads also travel. Generic ballot margins measured in July and August have frequently narrowed by Election Day, because a midterm is decided seat by seat, and candidate quality, local turnout, and the shape of each district do work that a national average cannot see. It is August. A full campaign is still unwritten. The number is a real input, honestly read, and it is not a result.

What the rankings measure underneath the polling

The generic ballot is a national mood. Control of Congress is 435 House races and 35 Senate races, and US Political Rank scores the ones that decide it. The Senate math is specific: Republicans hold 53 seats and defend 23 of the 35 on the ballot, and Democrats need a net gain of four for a majority, a picture tracked in the ranking of the 2026 Senate races by flip likelihood (Cook Political Report, 2026). The historical spine of the House number lives in what history says about the 2026 House.

The last thing worth saying is about the polls themselves. Not every survey is worth the same weight, which is why US Political Rank keeps a running ranking of pollster accuracy rather than reacting to whichever number is loudest on a given morning. The count that decides the House and the Senate is taken on November 3. Today's average is a reading on the way there, and the fair way to hold it is exactly as heavy as the evidence makes it, and no heavier.