Americans voted on November 3 in the 2026 midterm election, and Democrats won control of both the House and the Senate. The president cannot run again. For investors the question is simple: what has the S&P 500 done after midterms like this one?
At RegimeSignal we build models that predict the market’s turns: separate Bear, Correction, Pullback and Recovery signals for the S&P 500, which have averaged 84% accuracy when a signal fires and have called 8 of 8 bear markets since 1950. The election calendar is not one of those signals. But it is the question every investor is asking this week, so we measured it the same way we test everything else: every midterm since 1930, from the closing price on election day. Each finding below carries a label saying how much weight it can bear, and we publish the claims that did not hold up next to the ones that did.
SIGNIFICANT, BORDERLINE From the election-day close, the S&P 500 was higher a year later in 21 of 24 midterms, with a median gain of +13.2%. The same calendar dates in non-midterm years produced a median of +8.1%, higher 64% of the time. The pattern holds whichever party gains seats.
NOT STATISTICALLY SIGNIFICANT The week after the vote has been higher in 16 of 24 midterms, median +1.3%, against 55% higher and a median of +0.2% for the same week in other years. The direction is consistent, but the gap is small enough to be chance (z = +1.0). From election day to year-end, midterm years look no different from other years.
NOT STATISTICALLY SIGNIFICANT Five midterms were held under a president barred by the 22nd Amendment from another term. The S&P 500 rose after every one of them over the next week (average +1.4%), month (+2.6%) and year (+9.2%). The calendar year that followed is where they differ from other midterms: it averaged +6.6%, up in 4 of 5, against +15.9% for the post-midterm years when the president could run again.
| Midterm and president | Congress after the vote | Week | Month | 12 months | Next year |
|---|---|---|---|---|---|
| 1998 · Clinton | Republican | +1.6% | +3.5% | +22.0% | +19.5% (1999) |
| 1958 · Eisenhower2026 SETUP | Democratic | +2.8% | +1.9% | +11.1% | +8.5% (1959) |
| 2006 · G.W. Bush2026 SETUP | Democratic | +0.8% | +1.8% | +6.7% | +3.5% (2007) |
| 1986 · Reagan2026 SETUP | Democratic | +0.4% | +2.8% | +1.9% | +2.0% (1987) |
| 2014 · Obama | Republican | +1.4% | +3.0% | +4.5% | −0.7% (2015) |
Sorted by the next calendar year, largest first. Week, month and 12 months are 5, 21 and 252 trading days from the election-day close. 1974 is excluded: President Ford took office in August 1974 and could run in 1976.
2027 opens with a Republican president who cannot run again and Democrats holding both chambers. That has happened three times: after the 1958, 1986 and 2006 midterms. The S&P 500 rose every time, at a week, at a year and over the next calendar year. The gains were modest: +8.5% in 1959, +2.0% in 1987 (a year that included the October crash) and +3.5% in 2007.
Three cases cannot carry a forecast, and we do not present them as one. They are context: the closest historical matches to this year, and all three were positive but below the post-midterm average.
HIGHER CONVICTION The first half of the year after a midterm (Year 3 of the term) has averaged +10.0% and been higher 92% of the time across 24 cycles, against +4.1% for any half-year. This is the one political-calendar finding that clears a two-standard-deviation bar in both halves of the record. The full calendar year has averaged +14.0%, higher in 19 of 24. The lame-duck cases above are the reason to temper that number this time.
Several claims repeated every election season did not survive testing on S&P 500 data back to 1929:
| Claim | What the data says |
|---|---|
| Divided government beats unified | Divided 82.6% positive, unified 84.6%: no difference |
| A Republican Congress is best for stocks | True since 1950 (+14.7% a year vs +7.0% Democratic), but the order breaks when the record starts in 1929; a chance test gives about 1 in 10 |
| A midterm flip moves the market | Chamber changes hands 83.3% positive, no change 82.6% |
| Sell on election day | The 5 and 10 days after are slightly negative on average, never significant |
The most-quoted midterm statistic, the gain from the midterm-year low to a year later, is a look-ahead figure: the low is only known after the fact. We do not use it.
The vote matters less to the S&P 500 than the calendar it sits in. The year after a midterm has been one of the strongest stretches of the presidential cycle, and the 12 months after midterm day have been higher in 21 of 24 cases. This year adds a qualifier: under a president who cannot run again, the year after has still been positive, but by about half as much.
None of these findings is a signal, and none of them would have warned of 1987 or 2007 on its own. What protects a portfolio is knowing when a real turn is coming. Over Dec 2000 to Oct 2025, acting only on our Bear and Correction alerts would have ended with +126.3% more wealth than holding the S&P 500, with a worst drawdown of −12.7% against −50.8% for the index. And if a softer Year 3 is ahead, stock selection matters more than usual: our Top Stock Picks have returned 15.8% a year against 10.8% for the S&P 500 over 20 years, and our Core Stock Picks 20.1% against 14.7% over 15 years, ahead in 80% of those years.
The full election study, with the term clock for every president since 1929 and every null result, is on the Election Cycle page of the platform and in this week’s free Weekly Market Outlook. You can see whether the Bear or Correction signals are firing today, free for 15 minutes with no signup, at regimesignalai.com.
Method. S&P 500 daily closing prices (price only, no dividends), 1928 to October 30, 2026. Midterm windows run from the close on election day (or the prior close when markets were shut); week, month and 12 months are 5, 21 and 252 trading days. Calendar-year returns 1929 to 2025. Party control is the majority seated that January (Party divisions of United States Congresses); 2001–02 counts as split. “Higher conviction” holds against its baseline at a two-standard-deviation bar in both halves of the record; “significant, borderline” clears a 5% chance test once, uncorrected for the other cells tested; “not statistically significant” could be chance. Election result: CBS News, CNN, November 4, 2026. RegimeSignal calculations.
See if our Bear or Correction signals are firing today →Free 15-minute preview · no signup · no credit cardMore studies: Free Stock Market Research · new every Tuesday.
Cronus Market Intelligence, LLC is a publisher of market research and is not a registered investment adviser. Signal and stock-pick figures are historical model results (Dec 2000 to Oct 2025 for the alerts; 20 and 15 years for the picks), before costs and taxes, not live trading results. This article is informational only; it is not investment, tax or legal advice, is not personalized, and is not a recommendation or solicitation to buy or sell any security. Historical patterns and small samples may not repeat. Past performance does not indicate future results. RegimeSignal™ is a trademark of Cronus Market Intelligence, LLC.
RegimeSignal™ · Cronus Market Intelligence, LLC · Santa Barbara, California