How Did the Stock Market Do After Every Midterm Election Since 1950?

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-11-02 · Updated 2026-11-03

The question here is narrow on purpose: on the last trading day before each US midterm election, where did the S&P 500 close, and where was it three, six and twelve months later? Every one of the 19 answers is in the table below, with the dates, so nothing has to be taken on trust.

Download the data: all 19 midterms as CSV. You may reuse the table, the chart and the file with a credit and a link back, under the site license.

Watch the 60-second version of this study on YouTube.

Findings you can quote

• The S&P 500 was higher 12 months after all 19 midterm elections from 1950 to 2022. The average gain was 15.4% and the median 15.0%, on price alone.
• Six months after, it was also 19 for 19 (average 14.5%). Three months after, it was 17 for 19 (average 7.9%).
• Across all 19,079 twelve-month windows since 1950, 74.7% ended higher, with an average of 9.4% and a median of 10.6%.
• Weakest 12 months: 1986, +2.0%. Strongest: 1954, +33.0%.
• In 18 of 19 midterm years the index dipped below its starting close at some point in the following 12 months. The deepest dip was 14.1%, on December 24, 2018.
• The 12 months leading into the vote were ordinary: up in 12 of 19 years, average 3.3%.

How it works

Start. Federal Election Day is set by law as the Tuesday after the first Monday in November of every even-numbered year (2 U.S.C. § 7), so it always falls between November 2 and November 8. The start of each window is the last S&P 500 close before that Tuesday. The daily data has no close on Election Day itself in any year from 1950 to 1968, or in 1972, 1976 and 1980, so “the close before” is the one date that exists in every year. It also means the result of the vote was not yet known at the start.

End. The first close on or after the same calendar date 3, 6 or 12 months after the start close. Starting from Monday, November 7, 2022, the 12-month end is the close of Tuesday, November 7, 2023.

Return. End close divided by start close, minus one, from the unrounded daily closes. Dividends are not included. The S&P 500 index is a price index, so every figure here understates what a holder of an index fund earned, by roughly the dividend yield of that year.

Base rate. The same rule applied to every trading day since January 3, 1950 that has a full window in the data: 19,079 twelve-month windows, 19,203 six-month windows and 19,265 three-month windows. The data runs to October 30, 2026.

Same-season control. Midterm windows all start in early November, so the fairest comparison is the same start, built the same way, in years without a midterm: the close before the Tuesday after the first Monday in November of every year from 1950 to 2024. That gives 19 presidential-election years and 37 odd years. 2025 is left out because its 12 months are not finished.

A worked example: the 2022 midterm

The 2022 midterm was held on Tuesday, November 8. The last close before it was Monday, November 7: 3,806.80. Twelve calendar months later is November 7, 2023, a trading day, and the S&P 500 closed at 4,378.38.

4,378.38 ÷ 3,806.80 = 1.1501. Subtract one and the 12-month return is +15.0%, almost exactly the median of all 19.

The path inside that year is the part a single number hides. The lowest close was the day after the vote, November 9, 2022, 1.5% below the start. Three months in, on February 7, 2023, the index stood at 4,164.00, up 9.4%. Six months in it was up 8.7%: it had given a little back. Anyone redoing this needs only two rows of the daily closes and one division.

The original data

Every midterm since 1950, with the start close and the return to each horizon. The last column is the lowest close inside the 12 months, against the start, and its date.

Summary by horizon, midterm years against every window of the same length since 1950:

The two 3-month losses were 2002 (−6.6%), when the index slid back toward its October 2002 low, and 2018, down 0.02%, after the sharp December fall that took the index 14.1% under its start.

Is the midterm pattern more than chance?

Against the same season in other years, the gap is large. Of the 56 non-midterm years from 1950 to 2024, 38 rose over the same early-November-to-November window (67.9%), with an average of 7.3%. The 19 presidential-election years rose 12 times (average 9.0%); the 37 odd years rose 26 times (average 6.5%).

Three ways to put a number on luck:

1. If each year were a coin that lands up 74.7% of the time, the base rate, nineteen ups in a row would happen 0.39% of the time, about once in 258 tries.
2. Shuffle the labels instead. Drawing 19 years at random from the 75 early-November windows (1950 to 2024) 200,000 times, all 19 were positive in 0.22% of draws, and the average reached 15.4% or more in 2.2%.
3. Against the spread of all 19,079 twelve-month windows (standard deviation 15.8 points), a 19-year average of 15.4% is only about 1.6 standard errors above 9.4%. A gap that size turns up by chance about one time in ten.

What that adds up to. The unbroken streak is hard to explain as luck on these tests; the size of the average gain is less convincing. Both tests also overstate confidence for one reason they cannot see: the midterm pattern became famous because people noticed it in this same history, and a pattern found first and tested second always looks stronger than it is. U.S. Bank’s own statistical test, over 31 midterms from 1900 through 2025, did not establish a dependable midterm effect. Nineteen events can be broken by the twentieth.

Does the start day matter? Not much. Starting from the first close after Election Day, when the result was known, all 19 were still higher at 12 months, with an average of 15.0% and a median of 14.4%. Published versions that start from the end of October or use monthly averages land a little higher or lower because their start and end points differ.

The year before the vote, and the dips inside the year after

The 12 months leading up to each midterm were unremarkable: up in 12 of 19 years, average 3.3%, median 5.7%. The worst was 1974, down 30.7% into the vote, during the 1973 to 1974 bear market. Strong run-ups happened too: 1954 rose 28.9% in the year before and 33.0% in the year after.

The year after was never a straight line. Only 1954 stayed above its starting close the whole way. The 1986 window held the October 19, 1987 crash, when the index closed 8.5% under its November 1986 start, and still ended 2.0% up two weeks later. 2002 sat 11.8% under its start in March 2003. Anyone reading “19 out of 19” as “safe for a year” would have sat through each of those, which is what a drawdown looks like from the inside.

The 2026 midterm

The 2026 midterm is on Tuesday, November 3, 2026. The starting close for this study is November 2, 2026, which was not yet in the data when this page was measured. Its 3-month figure arrives in February 2027 and its 12-month figure in November 2027. Until then the 2026 row reads “not yet known”, and this page makes no estimate of it.

When it fails

A small sample is the first limit. Nineteen events make a wide range of outcomes look tidy. Two of them, 1986 and 2010, ended less than 3% up, and each had closed below its start within the last five weeks of its window (October 19, 1987 and October 3, 2011).

Many windows started just after a bear market’s low. Of the 11 falls of 20% or more since 1950, 7 hit bottom in a midterm year before the vote: 1962, 1966, 1970, 1974, 1982, 2002 and 2022. A window that opens weeks or months after a low has a head start, so part of the record may be recoveries that happened to line up with the calendar. The season itself does not explain it: the same window in non-midterm years averaged 7.3%, below the 9.4% of all windows.

Price only. No dividends, no inflation, no fees, no taxes. With dividends every return here would be higher; after inflation they would be lower, most of all in the high-inflation 1970s.

It describes the past, not a signal. Nothing here says what the S&P 500 will do after November 3, 2026. Treating a 19-for-19 record as a reason to buy or sell is a market timing decision, and the dips above show what the year in between can look like.

How to cite this

Whitman Trading (Michael Whitman). “How Did the Stock Market Do After Every Midterm Election Since 1950?” whitmantrading.com, November 2, 2026. https://whitmantrading.com/research/stocks-after-midterm-elections/ Data: S&P 500 (^GSPC) daily closes from Yahoo Finance, price only.

Embed the chart on your own page by pasting this snippet; please keep the source link:

<figure>
  <img src="https://whitmantrading.com/img/charts/m109-midterm-12m-returns.svg" width="720" alt="S&P 500 return in the 12 months after each US midterm election, 1950 to 2022: all 19 positive, average +15.4%.">
  <figcaption>Source: <a href="https://whitmantrading.com/research/stocks-after-midterm-elections/">Whitman Trading, S&amp;P 500 after every midterm election since 1950</a></figcaption>
</figure>

Questions people ask

Do stocks go up after midterm elections?

Since 1950 they have every time on a 12-month view: the S&P 500 was higher a year after all 19 midterms, by 15.4% on average. Over three months it was 17 of 19. That is a record of the past, from a small sample, not a forecast.

Are midterm years bad for the stock market?

The 12 months before a midterm averaged 3.3% and rose in 12 of 19 years, below the 9.4% average of all 12-month windows. Some years were very weak, like 1974 (down 30.7%), and some strong, like 1954 (up 28.9%).

Does it matter which party wins the midterms?

This study does not split the results by party. With 19 elections, any split leaves groups of a handful of years, too few to separate politics from recessions, rates and bear markets that happened at the same time. All 19 years, whichever party gained, ended higher at 12 months.

What will the stock market do after the 2026 midterms?

Not yet known. The 2026 vote is on November 3, 2026, and its 12-month result will exist in November 2027. This page will add the row then.

Related

The full list of datasets behind this site’s figures, with downloads, is on the research page, and the rules for recomputing every number are on how this site is made. For what holding an index fund through windows like these involves, the index fund page uses the same S&P 500 series over much longer periods.

This page is educational, not financial advice. Test every idea on your own charts before risking money.