Midterm Years & the Market: 15.4% return after every time?

You'll see one number repeated before the US midterm elections on November 3: 15.4%. It's the average gain in the S&P 500 over the year after a midterm election since 1950. Capital Group, the fund manager, states it plainly:

I reran it on daily S&P 500 closes (data as of October 19, 2026) and get 15.1%. The index was higher a year after all 19 midterm elections from 1950 to 2022. So the number holds. What it doesn't settle is whether 19 observations are enough to lean on, and whether the midterm is what's doing the work. After that, I'll show how the four-year political calendar sits next to the 18.6-year land cycle I follow.

I have no view on who wins on November 3, and nothing here depends on it. The test never looks at party. Last week I said I'd check whether the pattern people quote holds up. This is that check.

A midterm election is the one held two years into a president's four-year term. I took each of the 19 from 1950 to 2022 and measured the S&P 500's price change over the 3, 6 and 12 months after Election Day, with no dividends. A November start has its own seasonal drift, so I set each window against three comparisons. One is the 19 presidential Election Days. Another is the same November Tuesday in odd-numbered years, when no federal election is held (37 or 38 of them, depending on the window). The last is every possible start day since 1950.

The 6-month window is the clearest. After midterm Election Days the index gained 14.1% on average and was up all 19 times. After the comparison starts it gained 4% to 5%. I shuffled the labels on the dates 100,000 times to see how often chance alone produces a gap that big, and it did 8 times.

The 12-month window is closer. The gap against the other Election Days is 7.9 percentage points, and shuffling produces one that big about 6 times in 100. A cruder test looks better: since 1950 a 12-month stretch has been positive 75% of the time, so 19 in a row would happen by luck about 4 times in 1,000. But I tested this pattern because it's the one being quoted, so the real odds of luck are higher than either number says.

The pattern is often told as “stocks bottom in the midterm year, then rally.” That version starts from the low. From the lowest close of each midterm year, the next 12 months averaged 31.8%. From Election Day they averaged 15.1%. In 18 of the 19 midterm years the low had already passed by Election Day, and the October low of the folklore fits only 5 of them: 1966, 1974, 1990, 2002 and 2022.

The gap between the first two bars is the part only hindsight can see. Election Day is known in advance, while the low is only found afterwards, so the larger number describes a start date nobody could have picked in real time.

The average calendar-year return in midterm years (so from mid term until end of that calendar year) is the lowest of the four at 4.7%, but that average is noisy. The median is 1.1%, and 11 of the 19 years were up. The sturdier signal is the drop along the way. The average worst fall from a peak inside a midterm year was 17.5%, against about 12.5% in the other years. In 6 of the 19 midterm years the index fell 20% or more at some point, against 7 of the 57 other years. Removing any single year doesn't change that, and shuffling gives a gap this big about 3 times in 100.

I should be plain about how I found it. I compared all four kinds of year, and the worst of four always looks a little unusual. Shuffling the year labels, one of the four looks this bad on drops about 9 times in 100, and on returns about 28 times in 100. So I'd call the drop suggestive and the return gap weak.

Two ideas get offered. One is that investors demand extra return while government policy is uncertain, and that premium shrinks once a vote resolves it (Pástor and Veronesi, 2013). If that were the whole story, I'd expect a lift after presidential elections too, and the record shows none. After those, the S&P 500 gained 4.6% over 6 months, the same as from any day.

The other idea is the calendar. A November midterm sits where the weakest year of the four ends and the strongest begins. The third year of a presidential term has averaged 17.2% and been up 17 times in 19. So the post-midterm rally and the third-year effect may be one observation under two names. Neither idea tells me why it should keep working.

My read is that the midterm record describes how the years since 1950 went. It isn't a rule I'd size anything on, and it's no reason to change what you hold because of a date on the calendar. I could be wrong in either direction. If the pattern is real, I'm giving it too little weight. If 1950 to 2022 was a lucky stretch, a 6-month gap of 14.1% against 4% to 5% will still look more convincing than it should.

Quick note: the readings and figures here are as of October 20, 2026, when I wrote this — markets move, so for the live picture I keep Smart X Terminal updated in real time.

Below the line, I take the question this record raises for my own work. The 12 months after a midterm election have been up 19 times out of 19, which looks like it clashes with the cycle view I've laid out in recent issues. I test that against the midterms nearest a crisis date in my land-cycle record. I also show the two earlier midterms that match 2026 on price, and how I'm positioned through the vote.

On the free edition, you can still open the same cycle-position reading in Smart X Terminal.