Which Year of a Presidential Term Has Actually Paid Investors Best
Every US presidential term is four years long, and market folklore has assigned each of those years a personality: year one is the reform year, year two is the painful year, year three is the year the White House wants the economy humming, year four is the election. It is a tidy story. The useful question is whether the returns actually line up behind it.
They do — more cleanly than almost any other calendar pattern in US equities. Across 98 years of S&P 500 total returns, 1928 through 2025, the third year of the presidential term has averaged +17.8% and finished positive in 92% of cases. The second year — the midterm year — has averaged +7.1% and finished positive 58% of the time. Same index, same century, ten percentage points a year apart depending on nothing but where the calendar sat in the election cycle.
That matters right now for a specific reason. 2026 is a midterm year. The election is 3 November 2026. The moment it passes, the cycle rolls into year three.
This is a data piece, not a forecast. Every figure comes from one public dataset, the method is spelled out at the end, and the places where the pattern is weaker than it looks are stated plainly rather than buried.
What the presidential cycle actually is
The presidential (or "election") cycle theory dates to Yale Hirsch's Stock Trader's Almanac in the 1960s. The claim is simple: fiscal and monetary policy is not distributed evenly across a presidential term, so neither are equity returns.
The four positions, using the convention throughout this article:
- Year 1 — post-election year. The year after a presidential election (2025, 2021, 2017…). New administration, new agenda, often the least market-friendly half of it.
- Year 2 — midterm year. Congressional midterms (2026, 2022, 2018…). Policy uncertainty peaks; the president's party usually loses seats.
- Year 3 — pre-election year. (2027, 2023, 2019…) The year an administration most wants growth, employment and asset prices moving in the right direction ahead of the next campaign.
- Year 4 — election year. The presidential election itself (2028, 2024, 2020…).
We measure calendar-year total return — price change plus dividends reinvested — because that is what an investor actually earns and because it is what the source dataset publishes.
Every cycle position since 1928: the reference table
| Cycle position | Years | Average | Median | Positive | Worst | Best |
|---|---|---|---|---|---|---|
| Year 1 · post-election | 25 | +10.6% | +12.4% | 60% | −35.3% (1937) | +50.0% (1933) |
| Year 2 · midterm | 24 | +7.1% | +5.0% | 58% | −25.9% (1974) | +52.6% (1954) |
| Year 3 · pre-election | 24 | +17.8% | +22.5% | 92% | −43.8% (1931) | +46.7% (1935) |
| Year 4 · election | 25 | +11.9% | +15.9% | 84% | −36.5% (2008) | +43.8% (1928) |
| All years | 98 | +11.9% | +15.2% | 73% | −43.8% | +52.6% |
The ordering is unambiguous: year 3 > year 4 > year 1 > year 2. And it is not a quirk of the Depression years sitting at the front of the sample. Split the record three ways:
| Cycle position | Since 1928 | Since 1950 | Since 1980 |
|---|---|---|---|
| Year 1 · post-election | +10.6% · 60% up | +11.5% · 63% up | +18.3% · 83% up |
| Year 2 · midterm | +7.1% · 58% up | +8.3% · 63% up | +5.9% · 64% up |
| Year 3 · pre-election | +17.8% · 92% up | +20.8% · 100% up | +19.2% · 100% up |
| Year 4 · election | +11.9% · 84% up | +11.4% · 89% up | +10.0% · 83% up |
Year 3 does not merely survive the cut — it gets better. Since 1950 and since 1980, every single pre-election year finished positive. Year 2 is the only position that gets worse in the modern era, averaging just +5.9% since 1980.
Year 3: the strongest year, and the two that broke it
Here is every pre-election year in the sample:
1931 −43.8% · 1935 +46.7% · 1939 −1.1% · 1943 +25.1% · 1947 +5.2% · 1951 +23.7% · 1955 +32.6% · 1959 +12.1% · 1963 +22.6% · 1967 +23.8% · 1971 +14.2% · 1975 +37.0% · 1979 +18.5% · 1983 +22.3% · 1987 +5.8% · 1991 +30.2% · 1995 +37.2% · 1999 +20.9% · 2003 +28.4% · 2007 +5.5% · 2011 +2.1% · 2015 +1.4% · 2019 +31.2% · 2023 +26.1%
Two down years in twenty-four. 1931 (−43.8%, the worst single year in the entire 98-year sample) and 1939 (−1.1%, the year Germany invaded Poland). Since 1939 there have been 21 consecutive positive pre-election years, averaging +20.3%.
That streak is the single most-quoted statistic in presidential-cycle commentary, and it deserves the asterisk it rarely gets: 1931 is in this bucket. The year 3 group contains both the highest hit-rate and the deepest hole in the record. An average of +17.8% with a −43.8% tail is a very different object from a steady +17.8%.
The three weakest survivors of the streak are also instructive: 2011 (+2.1%), 2015 (+1.4%) and 2007 (+5.5%). All three were positive, all three were nothing like +20%, and 2007 was followed immediately by 2008's −36.5%. "Positive" and "good" are not the same claim.
Year 2 is the weakest — and that is the year we are in
The midterm year is the cycle's problem child: +7.1% average, +5.0% median, positive only 58% of the time, and the only position that has deteriorated in the modern sample. Our companion reference on midterm election years and the stock market breaks that year apart in detail — the deeper-than-average intra-year drawdown, the September–October bottom zone, and what the result itself has and hasn't meant.
What that piece does not cover, and what completes the picture, is the handoff.
The Year 2 → Year 3 handoff
| Midterm year | That year | Following pre-election year |
|---|---|---|
| 1930 | −25.1% | −43.8% |
| 1934 | −1.2% | +46.7% |
| 1938 | +29.3% | −1.1% |
| 1942 | +19.2% | +25.1% |
| 1946 | −8.4% | +5.2% |
| 1950 | +30.8% | +23.7% |
| 1954 | +52.6% | +32.6% |
| 1958 | +43.7% | +12.1% |
| 1962 | −8.8% | +22.6% |
| 1966 | −10.0% | +23.8% |
| 1970 | +3.6% | +14.2% |
| 1974 | −25.9% | +37.0% |
| 1978 | +6.5% | +18.5% |
| 1982 | +20.4% | +22.3% |
| 1986 | +18.5% | +5.8% |
| 1990 | −3.1% | +30.2% |
| 1994 | +1.3% | +37.2% |
| 1998 | +28.3% | +20.9% |
| 2002 | −22.0% | +28.4% |
| 2006 | +15.6% | +5.5% |
| 2010 | +14.8% | +2.1% |
| 2014 | +13.5% | +1.4% |
| 2018 | −4.2% | +31.2% |
| 2022 | −18.0% | +26.1% |
Across all 24 pairs, the pre-election year was positive 22 times (92%) with a mean of +17.8%.
The conditional cut is sharper. In the 10 cases where the midterm year itself finished negative, the following pre-election year was positive 9 times out of 10 and averaged +20.7% — better than the unconditional average. 1930→1931 is the single exception, and it sits inside the Great Depression. Recent examples are striking: 2018 (−4.2%) → 2019 (+31.2%), and 2022 (−18.0%) → 2023 (+26.1%).
The mirror image is worth as much: after a strong midterm year, year 3 has often been mediocre. 1938 (+29.3%) → 1939 (−1.1%). 1986 (+18.5%) → 1987 (+5.8%). 2006 (+15.6%) → 2007 (+5.5%). 2010 (+14.8%) → 2011 (+2.1%). 2014 (+13.5%) → 2015 (+1.4%). The cycle looks less like a calendar guarantee and more like mean reversion wearing a calendar costume.
The compounding gap
Run the arithmetic to its conclusion. Start with $100 in 1928 and hold the S&P 500 only during years matching one cycle position, sitting in cash the rest of the time:
| Strategy | $100 becomes |
|---|---|
| Only year 3 · pre-election | $3,596 |
| Only year 4 · election | $1,235 |
| Only year 1 · post-election | $781 |
| Only year 2 · midterm | $334 |
| Buy and hold every year | $1,157,001 |
An 11-fold spread between the best and worst quarter of the calendar, from the same index over the same 98 years. And the number that should stop anyone from acting on this too literally: buy-and-hold every year turned $100 into $1,157,001 — 322 times the best single-position strategy. The cycle tells you something about the distribution of returns. It does not tell you to sit out three years in four.
The dispersion the averages hide
Four averages computed from 24 or 25 observations each is a thin foundation, and a bar chart flatters it. Plot every individual year instead and the honest picture appears: heavy overlap between the groups, fat tails in both directions, and a year-3 bucket that owns the worst single observation in the whole dataset.
Year 1 deserves a note of its own, because it is the most misunderstood position. Over the full sample it looks mediocre — 10 down years out of 25, the highest count of any position. But since 1980 it has averaged +18.3% and finished positive 83% of the time. Nearly every one of its losing years is clustered before 1982: 1929, 1937, 1941, 1953, 1957, 1969, 1973, 1977, 1981 — then 2001, and nothing since. Whatever "year 1 is the bad year" once described, it has not described the last four decades.
Where the cycle stands in 2026
- 2025 was year 1. The S&P 500 returned +17.7%.
- 2026 is year 2, the midterm year. Through 30 July 2026, SPY is +8.8% year to date on a price basis — already ahead of the +5.0% midterm-year median, with five months and an election still to run.
- 2027 will be year 3.
So the setup does not match the classic template. The textbook midterm year is choppy and finishes single-digit or negative, setting up the year-3 rebound; 2026 has instead spent seven months grinding higher despite a hawkish Fed and a geopolitical risk premium — the same crosscurrents we tracked through July's CPI report and the AI momentum unwind.
That matters because of the conditional finding above. Year 3 has been strongest after weak midterm years (+20.7%, 9 of 10) and distinctly mediocre after strong ones (1939, 1987, 2007, 2011, 2015 all followed double-digit midterm years). If 2026 finishes where it currently sits, history's closer analogues are the second group, not the first.
Why the pattern might exist — and why it might not
The case for: administrations have both motive and instruments. Fiscal stimulus, deregulation and tax measures tend to be timed toward the back half of a term. The Fed is nominally independent, but the business cycle and the political cycle have historically rhymed. Uncertainty — which markets price directly — is highest before midterms and lowest once the composition of Congress is known.
The case against, which is stronger than most write-ups admit:
- The sample is tiny. Twenty-four observations per bucket. A single 1931 moves the year-3 average by more than 1.8 percentage points on its own.
- The pattern is public. It has been in the Stock Trader's Almanac since the 1960s. Widely-known calendar anomalies usually erode; that year 3 has not eroded is either evidence the driver is real or evidence the sample is too small to detect decay.
- Confounding is everywhere. 2023's +26.1% was an AI-driven megacap rally and a disinflation trade. Calling it a "pre-election year" return assigns credit to the calendar for something the calendar did not do.
- Cycle position is not a mechanism. It is a label on the x-axis. The mechanism, if there is one, is policy and rates — which is why the Fed rate-hike and rate-cut cycle references are the better companions to this table than any political narrative.
Treat the cycle as base rates and context, not as a signal.
Tracking the cycle without trading a calendar
Calendar effects are a starting hypothesis. What decides an actual position is what the tape and the fundamentals are doing right now.
- SimianX AI runs multi-agent analysis — technical, fundamental, news and a decision agent — on the same asset from one framework, so a seasonal prior gets tested against live evidence instead of standing in for it.
- Live Command Room — open a session on SPY or any single name and watch the agents work the actual levels.
- Autopilots re-run that analysis on your schedule and notify you when something you care about — a Fed decision, a breakout, an earnings print — actually fires.
- Market Pulse flags 52-week and 30-day extremes across equities and crypto as they print.
- AI Model Leaderboard — 30 models across 6 providers trading real positions, so model skill is measured on P&L rather than asserted.
For the rest of the calendar map, our S&P 500 seasonality reference covers the month-by-month pattern that sits underneath the four-year cycle.
Frequently asked questions
Which year of the presidential cycle is best for stocks?
Year 3, the pre-election year. Since 1928 it has averaged +17.8% with 92% of years positive; since 1950 every pre-election year has finished positive, averaging +20.8%.
Which year is worst?
Year 2, the midterm year: +7.1% average, +5.0% median, positive 58% of the time — and just +5.9% average since 1980.
What happens to stocks after a midterm election?
The following pre-election year was positive in 22 of 24 cases (92%), averaging +17.8%. After a negative midterm year it was positive 9 times in 10 and averaged +20.7%.
Has a pre-election year ever been negative?
Twice in 24: 1931 (−43.8%) and 1939 (−1.1%). None since 1939 — 21 straight positive pre-election years.
Does it matter which party holds the White House?
Not in a way this dataset can settle. With roughly a dozen terms per party since 1928, party splits produce samples too small to separate from the business cycle. Our presidential election year reference works through the party and incumbency splits directly.
Where are we in the cycle right now?
2026 is a midterm year (year 2). The election is 3 November 2026, and 2027 will be a pre-election year (year 3).
Should I actually trade this?
The compounding table argues against it. Holding only year-3s turned $100 into $3,596 since 1928; holding every year turned it into $1,157,001. The cycle is useful as a base rate for what a normal year in each position looks like — not as an instruction to be in cash 75% of the time.
Methodology and limitations
Returns are S&P 500 annual total returns (price appreciation plus reinvested dividends) from the NYU Stern historical returns dataset maintained by Aswath Damodaran, covering 1928–2025 — 98 complete calendar years. That dataset is the standard academic reference for long-run US asset returns and is free to inspect, so every number here is reproducible.
Cycle position is assigned arithmetically from the calendar year: year 4 (election) is any year divisible by 4; year 1 is the following year, and so on. The 2026 year-to-date figure is a separate measurement — SPY price return from its 31 December 2025 close of $681.92 to its 30 July 2026 close of $741.69, taken from SimianX's own daily candle store. It excludes dividends and is therefore roughly 0.7–1.0 points below the equivalent total-return figure, and it is a partial year, not a year.
Four limitations that should travel with any use of this table:
Sample size. 24–25 observations per bucket. These are descriptive statistics, not significance tests, and no p-value is implied.
Fat tails. The averages sit on distributions containing −43.8% and +52.6%. The median is the more honest central estimate for every bucket, and the dot plot above is the more honest summary than any of the bars.
Regime overlap. The strongest era for year 3 (1950 onward) is also the era of the postwar expansion, the Volcker disinflation, and four decades of falling rates. Separating "pre-election year" from "falling-rate regime" is not possible with 24 data points.
Survivorship of the framing. Presidential-cycle theory became famous because the pattern was strong in the sample that produced it. Some of the apparent strength is selection.
None of this is investment advice. It is a reference table with its workings shown.
Related Reading
- Midterm Election Years and the Stock Market: 1950–2026
- S&P 500 in Presidential Election Years: 1928-2024 Data
- S&P 500 Seasonality: The Best & Worst Months 1950–2026
- Every S&P 500 Bear Market Since 1929: Duration and Recovery
- How Long Every Bear Market Took to Recover: 1929-2022
- Every Fed Rate-Hike Cycle Since 1983: S&P 500 & Chip Stocks
- Every Fed Rate-Cut Cycle Since 1980: Stocks, Bonds & Gold
- The 20 Worst Days in S&P 500 History and What Came Next
- Browse all SimianX market reference tables
References
- NYU Stern — Damodaran historical annual returns on stocks, bonds and bills
- Wikipedia — the S&P 500 index, composition and history
- Investopedia — the presidential election cycle theory explained
- Stock Trader's Almanac — Yale Hirsch's original four-year cycle research
- Federal Reserve — FOMC calendar and policy statements
- US House of Representatives — history of midterm seat changes by term
