How Midterm Elections Affect Your Portfolio: What 19 Midterms Since 1950 Actually Did to the S&P 500

Score: the "sell everything before the election" instinct is the bogus part. We ran the numbers on every midterm since 1950. The midterm year itself is genuinely the roughest year of the presidential cycle — the average one includes a dip of about 18%. But in the 12 months after the vote, the S&P 500 finished higher 19 times out of 19, by about 15.5% on average, no matter which party won or lost. That is a pattern, not a promise — 19 elections is a tiny sample — but it points the opposite way from the scary headlines. The boring move has been the right one: don't trade the election.

The claim making the rounds

The vote is Tuesday, November 3. All 435 House seats are on the ballot, along with 35 Senate seats and 39 governorships. Republicans go in holding the White House and both chambers — a unified government — and the usual script says the president's party is about to have a bad night: it has lost House seats in 18 of the last 20 midterms, going back to 1946.

Which is exactly when the portfolio advice shows up. Every cycle, some version of the same headline makes the rounds: get defensive, raise cash, wait for the election to pass. The fear is understandable. Elections are loud, the outcome genuinely matters for policy, and nobody likes watching a portfolio statement during a coin-flip week.

So we checked the record instead of the vibes. Everything below is the S&P 500's price change (dividends excluded, which makes every figure here a little conservative), computed from daily index closes. Where our numbers overlap with the big research shops — Carson Group, Truist, Bank of America, Fidelity — we'll say so; they land in the same place.

First, the part that's real: midterm years are the rough ones

The fear isn't invented. The midterm year — year two of a president's term — has historically been the weakest of the four:

Year of the presidential cycle	Average S&P 500 calendar-year gain, 1951–2022
Year 1	+7.9%
Year 2 (midterm year)	+3.7%
Year 3	+16.8%
Year 4 (presidential election year)	+7.3%

And the ride inside a midterm year is worse than the full-year number suggests. Measuring each year's worst peak-to-trough slide on a closing basis, midterm years since 1954 averaged a −17.7% drawdown, against −12.4% in all other years. Carson Group's Ryan Detrick, running the same study his way, puts the average midterm-year correction at 17.5% — the deepest of the cycle. Midterm years averaged a +3.7% calendar gain against +10.7% for every other year in our count. Weak year, deep dips: that part of the story checks out.

Two things soften it. The pain has tended to come before the vote, not after: the fourth quarter of midterm years (September close to December close) averaged +6.6% in our data, positive in 16 of 19, and Carson finds October and November are historically the two best months of a midterm year (+3.0% and +2.7% on average since 1950). And 2026 has been a live example of the pattern so far — the index set a record close in early October, then wobbled into the vote. None of that tells you what happens next week. It tells you the chop you're seeing is the normal part, not the alarming part.

Then the vote happens

Here is the number the fear headlines leave out. We measured the S&P 500 from the last close before each Election Day to the close one year later, for every midterm from 1950 through 2022:

Nineteen elections. Nineteen positive years. Average gain +15.5%, median +15.1%. The best was +33.2% after 1954. The worst was +1.3% after 1986 — and even that window contains the October 1987 crash, one of the worst single days in market history, and still finished ahead. For scale, run the identical early-November-to-early-November window across every year since 1950, election or not, and the average gain is +8.9%. The post-midterm year has run at close to double the normal pace.

This isn't a cherry we picked. It's one of the most replicated stats in market research:

• Carson Group (since 1946): positive 12 months after 20 of 20 midterms, averaging about +15%.
• Truist (Keith Lerner, since 1946): average +14.4%, best +34% after 1954.
• Bank of America: up in every 12-month post-midterm period in the postwar data, about +14% on average.
• Fidelity (since 1938): positive 95% of the time, averaging about +14% — their one miss is in the pre-1950 data, which is why their percentage isn't perfect.

It's also why year three of the cycle, the year that starts after a midterm, tops the table above at +16.8%. The post-midterm window is where the cycle's strength has lived.

Does it matter who wins? The gridlock question

This is where the bar-room version of the research — "gridlock is good for stocks" — outruns what the data can actually carry.

Start with the post-midterm windows themselves. Split our 19 by the result: 14 midterms produced a divided government (the president's party missing at least one chamber), 5 left the president's party in full control. The divided outcomes averaged +15.7% over the next 12 months and were positive 14 of 14. The unified outcomes averaged +14.9% and were positive 5 of 5. Both flavors worked. And with only five unified outcomes since 1950, no honest person can claim the 0.8-point gap between them means anything at all.

The longer-run configuration tables get quoted more selectively. An analysis of S&P 500 returns by government makeup from 1926 to 2023 found:

Government, 1926–2023	Years	Average annual S&P 500 return
Democratic president, divided Congress	15	+16.6%
Unified Republican	13	+14.5%
Unified Democratic	36	+14.0%
Republican president, divided Congress	34	+7.3%

Read that table carefully and it argues against the simple gridlock story. The best and worst rows are both divided governments — what separates them is the president's party, not the split. And the buckets are really buckets of decades: the Republican-president/divided row holds the 2007–09 financial crisis and several debt-ceiling standoffs, while the Democratic-president/divided row holds most of the late-1990s boom. Change the period and the answer moves: Janus Henderson, using 50 years of Bloomberg data to October 2026, found split control returned 9.24% a year against 8.73% for unified — nearly a wash — while Bank of America, going back to 1928 with dividends included, found divided Congresses ahead, 13.5% to 11.4%. Three honest studies, three different "winners."

The conclusion the data actually supports is narrower and less exciting: every configuration has averaged solidly positive returns, and no study has isolated a gridlock effect from the economic weather that happened to come with it. The market's problem has never been which party wins. It's the not-knowing — and the not-knowing ends Tuesday night either way.

What actually changes after the vote

If the winner barely matters to the index, what does change? Mostly, one variable: how much is undecided.

The uncertainty premium unwinds. Options markets charge extra for protection into big national votes, and that premium deflates once results are known — a result academic research (Kelly, Pástor and Veronesi, studying elections across 20 countries) has documented for years. Stocks don't need good news after an election. They need settled news.

Policy risk becomes specific instead of general. Before the vote, every sector with a Washington problem trades under a cloud. After it, the mechanics are simple arithmetic: if control splits, any major new tax or spending law needs votes from both parties, so the default is the status quo; if one party holds all three levers, the reconciliation playbook — the route the 2017 tax cuts took — is available. That's not a prediction of either outcome. It's just which doors are open under each.

The sectors to watch are the ones whose earnings run through Washington: health care (drug pricing and insurance subsidies), energy (permits and regulation), defense (appropriations sail under both parties), financials (the regulatory temperature), and tech — where chip policy has been one of the few genuinely bipartisan projects going (the Micron deep-dive on this site walks through a fab both parties showed up to support). What moves these stocks after Tuesday isn't the index pattern. It's specific bills, rules, and budgets becoming guessable again.

Gridlock's real cost shows up on a schedule, too: a split government turns routine chores — funding the government, raising the debt ceiling — into recurring brinkmanship, as Bank of America notes. That's episodic headline risk, not a reason to be out of the market. But it's the honest downside of the "gridlock is good" slogan.

The honest caveats

Pattern posts usually stop before this section. Don't skip it, because it's where the 19-for-19 stat earns its limits.

• Nineteen is a tiny sample. Twenty, if you count from 1946. Those windows bundle in rate cuts, recession recoveries, wars, and a tech boom. Nobody has proven why the pattern exists — uncertainty ending, midterm years starting from a beaten-down base, and presidents pushing for a strong economy before their own re-election are all plausible, and none is established.
• The streak is window-sensitive. Measured November-to-November it's perfect since 1950. Shift the window and it breaks: Lincoln Financial found a −1.5% result from July 2002 to June 2003. U.S. Bank, extending the study back to 1900 through the Dow era, found the post-midterm year was not positive every time (though it still averaged +12.4%).
• Some wins were near-misses. 1986: +1.3%. 2010: +4.5%. 2014: +4.2%. A "perfect record" built partly on coin-flips is not a law of nature.
• Positive at month 12 doesn't mean calm in between. After the 2018 midterm, the market kept falling for seven more weeks and bottomed on December 24. The 2006 window closed in November 2007 — a month after the market peaked, with the financial crisis just outside the frame. The stat describes where the index ended up a year later. It says nothing about the ride, and nothing about month 13.
• These are price returns. Dividends are excluded (including them would raise every figure), and nothing is adjusted for inflation, which would shrink the 1940s and 1970s results.
• This election could be the first miss. The streak will either extend to 20 or end at 19, judged from November 2027. The data can't tell you which, and neither can we.

So what do you actually do with this?

The boring answer, which is the one the data keeps paying for:

Don't sell the election. Selling into midterm fear has historically meant selling during the weakest stretch of the cycle, right before its strongest. If your plan was fine in September, a vote count doesn't change it.

Don't bet on the streak, either. "Up 19 of 19" is an average of 19 very different years, not a signal with a trigger. Loading up because a pattern says so is just market timing wearing a history costume — and the caveats above are exactly the years that would have hurt.

Keep your schedule, not the news cycle's. Contributions, rebalancing, and cash needs should run on your calendar. If you're the type who checks what your holdings actually pay you while you wait out the noise, run them through the yield calculator — real numbers, no election required.

Separate the trade from the tax return. If Tuesday's result eventually becomes actual law that touches you — a tax bracket, a credit, a subsidy you use — that's a planning question for when there's a signed bill to read, not a trade to make on election night.

The vote matters. It decides real policy for real people, and the market will have opinions about it by Wednesday morning. But 75 years of record say the opinion that costs investors the most is the one that talks them out of their plan the week of the election. The pattern says stay put. The caveats say don't get cocky about it. Both fit in the same portfolio.

Get posts like this by email: Subscribe to BogusAPY.

Follow BogusAPY on X: @BogusAPY.

Check it yourself: links and sources

All market figures were checked November 1, 2026. Historical averages are price changes in the S&P 500 (dividends excluded) unless a source states otherwise.

• BogusAPY's own computation (the 19-for-19 table, midterm-year drawdowns, cycle years, Q4 figures, and the divided/unified split): S&P 500 daily closes, Yahoo Finance ^GSPC history, January 1950–October 2026. Post-midterm return = last close before Election Day to the close one year later. Cross-checked line-by-line against an independent computation of the same windows (Dave Manuel, "Stocks After Midterm Elections: 80 Years of Data," Oct 21, 2026) — same score, all 20 windows since 1946 positive, averages within ~1 point.
• Carson Group research via The Wealth Advisor (Oct 5, 2026) — Ryan Detrick: ~+15% average and no negative 12-month post-midterm period in 20 cycles since 1946; average midterm-year intra-year correction of 17.5% since 1950; October (+3.0%) and November (+2.7%) the best months of midterm years.
• Truist (Keith Lerner) and Bank of America figures — +14.4% average since 1946 (Truist); up every postwar post-midterm year at ~+14%, Q4 midterm total returns +6.5% vs +0.2% for Q1–Q3, and divided Congress 13.5% vs unified Congress 11.4% average annual total return since 1928 (Bank of America Capital Market Outlook, June 8, 2026) — as compiled in the October 2026 midterm research roundups linked here.
• Fidelity research, as reported by The Motley Fool (Sep 26, 2026): S&P 500 positive in the 12 months after 95% of midterm elections since 1938, averaging about +14%.
• Government-configuration averages, 1926–2023 (Democratic president/divided +16.6%, unified Republican +14.5%, unified Democratic +14.0%, Republican president/divided +7.3%): Retirement Researcher's analysis of S&P 500 returns by presidential and congressional control, as reported October 2026.
• Janus Henderson, "Maintaining perspective during the U.S. midterms" (Oct 2026) — Bloomberg data, 50 years to Oct 23, 2026: split control 9.24% annualized price return vs 8.73% unified; volatility 18.5% vs 15.4%.
• U.S. Bank Asset Management (post-midterm average +12.4% back to 1900, not positive every time) and Lincoln Financial (July 2002–June 2003: −1.5%) — as compiled in the Dave Manuel study above.
• Seat changes (president's party: House seats lost in 18 of 20 midterms since 1946): American Presidency Project, UC Santa Barbara, via the Dave Manuel study. 2026 ballot (435 House seats, 35 Senate seats, 39 governorships) and current control (Republican Senate 53, House 218): wire and major-outlet reporting, October 2026.

Figures checked November 1, 2026. Historical averages describe what happened in past windows; they are not a forecast for this election or any other.

Not financial advice.