The Midterm Market Pattern

What History Says About the Next 15 Months

Global Perspectives | October 2026

• Since 1950, the S&P 500 has returned an average of 29.9% over the 15 months from October of a midterm year through the following December, roughly double the 14.7% average for the same stretch in other years, and all 19 of those periods were positive.
• The pattern owes more to the calendar than to who wins. When the party out of the White House took both the House and the Senate in a midterm election, which has happened four times since 1928, subsequent returns showed no consistent partisan direction.
• Stocks enter this window already up 12.7% for 2026 through September, far above the typical midterm year, so history argues for patience and portfolio discipline.

A Familiar Temptation

As the midterm elections approach, many investors feel the pull to get defensive and wait for the political fog to clear. History offers a reason to resist that instinct. The months after a midterm election have been among the best stretches of the four-year presidential cycle for US stocks.

The political stakes this year are real. Prediction markets now give Democrats better-than-even odds of winning both the House and the Senate: 61% on Kalshi as of October 1 (Figure 1). But the market record around midterms has had far more to do with timing than with which party prevailed.

Assessing the Midterm Market Pattern

We measured S&P 500 total returns, including reinvested dividends, from the end of September through the end of the following December.¹ That window covers 15 months, starting in each year from 1950 through 2024. When the window began in a midterm year, it averaged 29.9%, and all 19 cases ended higher. When it began in any of the other three years of the presidential term, it averaged 14.7%, and 42 of 56 cases, or three in four, ended higher (Figure 2).

No other starting point came close. Windows that began the year after a presidential election averaged 12.1%, those that began the year before averaged 16.8%, and those that began in an election year averaged 15.1%. The average over every 15-month period since 1950, whatever its starting month, was 16.2%. By that yardstick, the post-midterm stretch has delivered nearly twice the typical gain.

The gap is too large to dismiss as luck. If the 75 October-start windows are shuffled at random, a group of 19 matches or beats the midterm group’s 15.2 percentage point advantage only about once in 750 tries. Nineteen gains in a row, in periods where gains otherwise occur three times in four, would be expected well under 1% of the time if each window were independent.² As far as finding calendar-based market anomalies goes, the midterm market pattern seems statistically significant.

Where the Gains Have Come From

The 15-month result is built from two better-known pieces of the cycle. The midterm year itself has been the weakest of the four, averaging 8.3%, with gains in 12 of 19 years. Its second quarter averaged a 2.0% loss, the only negative quarter of the cycle on average (Figure 3).

The recovery has typically begun late in the year. Midterm fourth quarters averaged 7.6%, compared with 4.2% for fourth quarters in other years. The advantage came from the size of the gains rather than their frequency: stocks rose in 84% of fourth quarters in both groups. The first quarter of the following year averaged 8.2%, the best of all 16 quarters. The year after the midterm, the third year of the presidential term, averaged 20.9% and rose in all 19 cases since 1950, with gains ranging from 1.4% to 37.6%.³

These results come with an important caveat: past performance is not indicative of future results. The cycle pattern emerged from this same history, and 16 quarters offer many chances for one to stand out by chance. Explanations such as policy stimulus ahead of a presidential election, or relief as election uncertainty lifts, are cetainly plausible but hard to prove. Nineteen observations also make for a small sample.

The Averages Hide the Ride

A winning streak says nothing about what investors had to sit through. Although all 19 midterm windows ended higher, 12 of them included a decline of at least 5% from a prior month-end high, and four included a decline of 10% or more (Figure 4). The worst came in the window that began in 1986, when the October 1987 crash cut the market’s value by nearly 30% from its month-end peak, before the period ended with a gain. More recently, the fourth quarter of 2018 fell 13.5%, despite its favorable spot on the calendar.

Election Outcomes Matter Less Than the Calendar

A sweep of both chambers by the party opposing the president would be unusual. Only four midterms since 1928 flipped both the House and the Senate: Republican gains in 1946 and 1994, and Democratic gains in 1954 and 2006.⁴ The results that followed ranged widely. After the Democratic sweep of 1954, stocks returned 31.4% in 1955. After the 2006 sweep, they returned 5.5% in 2007. The comparable figures after Republican sweeps were 5.2% in 1947 and 37.6% in 1995 (Figure 5). Across the four cases, the 15-month window averaged 26.8%, close to the 29.9% for all 19 midterms since 1950. The switches followed the general pattern rather than improving on it. Limiting the comparison to the three switches since 1950 gives the same answer: their fourth quarters averaged 6.4%, versus 7.8% for the other 16 midterms, while their following years averaged 24.8%, versus 20.2%. Which comparison looks better depends on the period chosen, as one would expect from a handful of cases.

A Democratic takeover of either chamber would produce divided government. Over the full 1950–2025 period, stocks did somewhat better under unified government, at an annualized 13.2% versus 11.4%. Excluding recession months reverses the ranking, to 14.2% under divided government versus 12.1% under unified (Figure 6). The reversal reflects timing more than politics: recessions made up 16% of divided-government months but only 7% of unified-government months.

It therefore appears the economic outlook matters more than the election result. And for what it’s worth, the bond market is not flashing a recession warning. According to the New York Fed’s model, based on the gap between 10-year Treasury yields and 3-month bill rates, the probability of a recession 12 months ahead was 13.9% as of August (Figure 7). That barely differs from the 12% of months since 1960 when the economy was in recession.

What Could Make 2026 Different

The biggest difference this year is where stocks are starting from. The S&P 500’s total return was 12.7% for 2026 through September, compared with an average of 0.9% for the same months of midterm years since 1950. Historically, the midterm window often began after a rough patch, and part of its strength was recovery. After midterm years in which stocks were already up through September, the following 15 months averaged 25.4%, compared with 37.7% after losing starts; this year’s 12.7% gain is exactly the up-year average (Figure 8).

We would still argue that fundamentals matter more than the election calendar. Corporate earnings, inflation, and borrowing costs are likely to drive equity markets, as is almost always the case. Also, given the AI theme’s driving role, investors are likely to focus on whether heavy AI spending generates sustainable profit growth. Investors will also watch closely how the newly elected Congress influences taxes, government spending, and regulation. And we doubt that a divided government will bring smaller deficits or lower interest rates.

Perspective for Portfolio Decisions

The main lesson is not to use election dates as a reason to enter or exit the market. Selling on political uncertainty forces a second decision about when to buy back, and history suggests the months after a midterm have been an expensive time to wait on the sidelines. The same history shows that sizable interim declines are common, so a favorable calendar does not justify taking more risk than an investor’s plan allows.

History may not repeat itself this time around. But knowing the favorable market history around midterm elections may help investors avoid headline-driven decisions.

 

William P. Sterling, Ph.D.,

Global Strategist

 

1 All returns are for the S&P 500 on a total return basis, including reinvested dividends, based on month-end index levels.

2 To test the strength of the midterm market pattern, we randomly relabeled the 75 October-start windows from 1950 to 2024, 1,000,000 times, and counted how often 19 randomly chosen windows beat the other 56 by at least the observed 15.2 percentage points; this occurred 0.13% of the time. The windows overlap the following calendar year, so the 15-month, third-year, and first-quarter results are not independent confirmations of one another.

3 Over the longer 1928 – 2025 sample, the third year averaged 17.9% and the midterm year 7.4%. The third year declined twice in that longer history: a 43.9% loss in 1931 and a 0.9% loss in 1939.

4 A complete switch means one party controlled both chambers before the election and the other party controlled both afterward, based on party divisions published by the House and Senate historians. Gaining one chamber while keeping the other, as in 1986 and 2014, does not count as a complete midterm switch, nor do presidential-election years.