Do stocks rally after the midterms?

What do people believe about midterm years?

Americans vote for Congress every two years, and the vote halfway through a president's four-year term is called the midterm. The market folklore around it is confident.

Midterm years are bumpy, because nobody knows who will control Congress. Then the vote happens, the fog lifts, and stocks rally.

It's usually traced to a trading almanac from the late 1960s that follows the four-year presidential cycle.

What exactly did we test?

The belief has two parts, so we tested both.

Part one: is the stretch from New Year to election day weaker in midterm years than the same stretch in other years? Part two: are the 12 months after election day stronger in midterm years than the same 12 months in other years?

We used the whole US stock market from July 1926 to August 2026: 25 midterm elections from 1926 to 2022, against 74 other years. Part one uses 24 of them, from 1930, because our data start in July 1926.

We measure every return above what cash in Treasury bills earned at the same time, so different interest-rate eras don't muddy the picture. When we compare two averages, the gap is in percentage points, or points for short.

What did we find?

Before the vote, midterm years really have been sluggish. From New Year to election day they averaged +0.2% above cash, against +8.0% for the same stretch in other years. Only 50% of midterm years beat cash by election day, against 76% of other years.

Some midterm years were rough. In 1974, stocks were 28.2 points behind cash by the vote, and in 2022 they were 20.8 points behind.

After the vote, it starts even better. Over the full century, the 12 months after a midterm averaged +14.4% above cash, against +7.0% in other years. Those post-midterm years beat cash 88% of the time, against 68% of other years.

But look at the last pair of bars. In the last 20 years, the year after a midterm beat cash all 5 times, yet other years did even better, by 4.8 points.

How did each midterm do?

Averages hide a lot, so here is every midterm alone, against the average other year of its span.

For the year after the vote, the bar rose: other years averaged +5.8% up to 2004 and +11.6% since. Up to 2002, the year after a midterm beat the average other year 16 times in 20. In the last 20 years, it did 0 times in 5, though stocks still beat cash all 5 times.

From 1930 to 2002, midterm years trailed the average other year by election day 12 times in 19. In the last 20 years, they did 4 times in 5. We call that a rough start, but trailing isn't falling: some of those years still beat cash.

So what's the grain of truth?

The weakness from New Year to the vote has shown up consistently. Midterm years trailed other years by 7.3 points before 1977 and by 8.4 points since, and by 8.0 points in the last 20 years.

It passed our test, but only just. The likely range of the gap runs from -15.3 to -0.1 points, so it stays clear of zero by a whisker.

The rally after the vote is the closer call. Over the full century, its likely range of +1.0 to +13.9 points stays above zero.

But our verdict rule also asks a pattern to hold in every period we check. The gap was +11.0 points from 1926 to 1976 and +3.6 points since, and in the last 20 years it turned negative.

So we call this part a myth, in a precise sense. Stocks usually did fine after a midterm, but you can't count on them doing better than in an ordinary year. It isn't faded either: the early decades alone had too few midterms to be sure.

Does it hold everywhere?

We repeated the test on the 43 of the 49 industries with enough history. Before the vote, midterm years were weaker in all 43 of them. After the vote, the year after a midterm did better in 39 of 43.

Industries all move with the market, so these aren't independent votes.

Does it matter who wins?

The groups are small, so treat this as a sketch, not a law.

In 10 midterms the president's party came out controlling both the House and the Senate. The next 12 months averaged +17.1% above cash. In the other 15 midterms, they averaged +12.7%.

In 10 midterms a chamber changed hands, and the year after averaged +10.2%. In the other 15, where both chambers stayed with the same party, it averaged +17.2%.

Neither gap is reliable. The groups are small and single years swing a lot, so their likely ranges, -7.2 to +14.6 points and -16.1 to +2.8 points, both include zero.

The groups also bunch by era. Most midterms where the president's party controlled both chambers came before 1977, and most flips came after. So these gaps may largely echo the early-versus-late difference above; with so few midterms, we couldn't check within each era.

The last 20 years can't settle it either. All 5 midterms since 2006 flipped a chamber, so there's no recent case to compare. The year after averaged +6.8%, against +11.6% in other years.

Who won didn't reliably show up in the next day's move either: the outcome groups were at most 0.8 points apart.

A fun footnote: the president's party gained House seats in only three midterms since 1926. The year after was strong all three times: +32.0% after 1934, +20.4% after 1998 and +20.1% after 2002. Three is a story, not a statistic.

What about the month before and after the vote?

Neither month shows a clear midterm effect, over the century or in the last 20 years. If anything, the month before the vote was stronger than usual.

Each window is 21 trading days. The month before the vote ends at the last market close on or before election day, and the month after starts there.

In the month before the vote, midterm years averaged +2.95% above cash, against +0.12% in other years. They beat cash 68% of the time, against 64%.

In the last 20 years, with just 5 midterms, it was +2.15% against +1.63%. That's a handful of elections, so treat it as a glance, not a verdict.

Over the century, that gap looks big, but it isn't clear. A single month swings a lot, and with only 25 midterms even a gap this size could be luck. We widened the range because we checked several things at once: it runs from -0.65 to +6.35 points, which includes zero.

So the final weeks before the vote weren't weak; if anything, they were stronger. Midterm years still lagged from New Year to the vote, so that lag must have built up before the final month.

The month after the vote averaged +1.19% in midterm years, against +0.60% in other years, 0.59 points apart. It beat cash 68% of the time, against 62%. In the last 20 years it was +1.22% against +0.91%. Widened the same way, its range of -2.00 to +3.02 points includes zero too.

Why might it seem to work?

One common explanation is uncertainty: markets dislike not knowing who will control Congress, and the vote ends the suspense, whatever the result.

Another is politics: presidents may push for good news before the next presidential election. We didn't test these explanations; they are reasons people give.

And there's plain luck. With only 25 midterms, a handful of big years can shape the averages.

What's the story in numbers?

25 midterms from 1926 to 2022. Rough starts (trailing an average year by election day): 12 in 19 up to 2002, 4 in 5 in the last 20 years. Rallies (a year after the vote that beat an average year): 16 in 20 up to 2002, 0 in 5 in the last 20 years.

Three things to take away. A slow run-up to a midterm vote is nothing new. The rally after the vote is no sure thing. And timing the midterms fell less, but grew more slowly.

What does this mean for you?

This is history, not a forecast. Midterm years have tended to lag from New Year to the vote, but a tendency over 25 elections tells you little about any single year.

A myth doesn't mean you should do the opposite. Selling after midterms because the rally hasn't stood out lately would be its own untested bet.

Timing the cycle would have meant slower growth. Holding stocks only in the 12 months after each midterm, with cash otherwise, grew +6.2% a year in total return. Staying invested grew +9.9% a year in total return. In the last 20 years, it grew +3.2% a year, against +11.5% staying invested.

Its worst fall was smaller, -32.3% against -83.9%, mainly because it held stocks only 24% of the time. In the last 20 years, its worst fall was -20.3%, against -54.7%.

What else did we test?

See all the calendar effects we tested, or how we test a myth.