It is not the market that is seasonal — the sectors are
Most seasonality work looks at the whole market: S&P 500, DAX, Nasdaq. A study published in May 2024 in the International Review of Financial Analysis goes one level deeper and asks about sector ETF seasonality — which industry delivers, and when. The answer is unusually clear-cut.
The number worth quoting: across the 25 years from 1999 to 2023, 8 of 9 US sector ETFs show statistically significant positive returns in April, November and/or December — and not a single one does so in six other months of the year. We recomputed all of it with our own data.
What the study actually did
Abbas Valadkhani and Barry O'Mahony examined the nine Select Sector SPDR ETFs tracking the S&P 500 in "Sector-specific calendar anomalies in the US equity market" (ScienceDirect, May 2024), covering January 1999 to December 2023.
The methodological trick: these nine ETFs have no overlapping constituents. Every S&P 500 stock sits in exactly one of them. That lets you isolate sector effects cleanly instead of blending them inside an index.
| Ticker | Sector |
|---|---|
| XLB | Materials |
| XLE | Energy |
| XLF | Financials |
| XLI | Industrials |
| XLK | Technology |
| XLP | Consumer Staples |
| XLU | Utilities |
| XLV | Health Care |
| XLY | Consumer Discretionary |
Two findings stand out. First, the positive calendar anomalies cluster in April, November and December. Second, there are six "muted months" (March, May, June, August, September, October) in which not one ETF shows a significant positive or negative anomaly — statistically, nothing reliable happens there at all.
The cross-check with SeasonAlpha data
We ran the same nine ETFs over the same window (1999–2023, n = 25 years per month) through our own methodology: normalized returns per calendar year, monthly return as the change within the month window, plus a t-test against zero.
Bold values are statistically significant (p < 0.05):
| ETF | April | November | December |
|---|---|---|---|
| XLB | +3.6% (p 0.021) | +3.2% (p 0.011) | +2.2% (p 0.056) |
| XLE | +4.7% (p 0.004) | +2.1% | +1.2% |
| XLF | +3.4% (p 0.013) | +1.4% | +1.2% |
| XLI | +3.2% (p 0.018) | +3.5% (p 0.002) | +1.4% |
| XLK | +1.7% (p 0.191) | +2.9% (p 0.050) | +0.5% |
| XLP | +1.6% (p 0.012) | +1.9% (p 0.001) | +0.8% |
| XLU | +2.5% (p 0.005) | +0.5% | +1.5% (p 0.062) |
| XLV | +2.2% (p 0.008) | +2.0% (p 0.027) | +2.1% (p 0.017) |
| XLY | +3.1% (p 0.014) | +2.9% (p 0.006) | +0.9% |
Our result confirms the study — and comes out slightly stronger: all nine ETFs have at least one significantly positive month among April, November and December.
April is the strongest single month
April dominates: 8 of 9 ETFs are significantly positive, averaging +2.9%. The one exception is Technology (XLK, p = 0.191) — the very sector most investors assume sets the seasonal tone.
November ranks second: 6 of 9 significant, averaging +2.3%. December, by contrast, clears the significance bar for exactly one ETF (XLV, p = 0.017). Its reputation as a rally month is weaker at the sector level than the broad index suggests.
The "muted months" hold up
The counter-test is what makes this convincing. The six muted months produce 54 individual tests (9 ETFs × 6 months), and in our data exactly one is significant (XLP in October, p = 0.035). Across 54 tests, chance alone would predict roughly one hit.
Put differently: for half the year, US sectors show no reliable calendar anomaly. That matters, because it suggests the April/November strength is not an artifact of an over-eager data search.
What the heatmap shows
The monthly heatmap colors every single monthly return by year — green for gains, red for losses. That reveals not just the average but the consistency of a month. Here is the technology sector:
Two things stand out in the November column. Across the last ten completed years (2016–2025) it is green in eight of ten cases, averaging +2.8%. The September column right next to it is the mirror image: −1.5% with a hit rate of only 50%.
One reading note: the top row is the current year, 2026. Months that have not happened yet appear without a value — that is not a data error.
Where the study and the present diverge
Seasonal patterns are not carved in stone. Two shifts show up clearly when we swap 1999–2023 for the last 20 years (2006–2025):
December has evaporated. In that more recent window, not one of the nine ETFs is significant in December. Energy (XLE) is even slightly negative at −0.4%. Anyone betting on a sector-level "December effect" today is leaning mostly on older data.
July has caught up. The authors found July as an additional month for seven ETFs in their post-financial-crisis subsample. Our data supports that even more strongly: over 2006–2025, July is significantly positive for 8 of 9 ETFs (all but XLE), averaging +2.8% with hit rates between 70% and 90%. XLF posts a positive July in 90% of those years.
Energy generally marches to its own drum — a useful contrast to the tech heatmap:
Why these patterns exist at all
Calendar anomalies do not come from nowhere. Plausible drivers include:
- Earnings-season rhythm: April and October/November are quarterly reporting months, which concentrates positive surprises on the calendar.
- Tax-year effects: Loss harvesting in autumn and reinvestment afterwards support late-year strength.
- Capital inflows: Bonuses, pension contributions and year-end reallocations arrive in clusters.
- Sector specifics: Utilities (XLU) and staples (XLP) are rate-sensitive and defensive — their seasonality follows different cycles than energy or technology.
Important caveat: none of these is a law of nature. They explain why a pattern can be stable, not why it must hold next year.
Limits — and what this means in practice
Three constraints are non-negotiable:
- Averages hide dispersion. An average of +3.2% means individual years were down double digits. An 80% hit rate also means one year in five went the other way.
- Known anomalies decay. The vanished December effect is exactly that. The better known a pattern, the sooner it gets priced in.
- Costs and taxes matter. Rotating monthly across nine ETFs creates fees and taxable events that raw returns do not include.
On the front-running idea: Quantpedia notes in its write-up of the study that entering one month earlier — October for November — historically produced better results than naively waiting for the month to start. The logic makes sense if many participants know the same pattern. Our own data, however, shows no significance in October for any of the nine ETFs over 2006–2025, so front-running remains a hypothesis rather than a demonstrated effect.
For investors, the sober takeaway: seasonality provides context, not a signal. It can calibrate your sense of timing — for instance, knowing that a sluggish September in tech is historically normal.
To check the patterns across all 324 tickers yourself: the monthly heatmap and monthly cycle are interactive for any ticker, sector rotation compares industries side by side, and the scanner hunts for notable calendar effects across the entire universe.
Methodology and transparency
We calculate normalized returns based on adjusted closing prices: each year starts at 100 and daily returns compound on top. Monthly returns are the change within the respective month window, and significance comes from a t-test against zero (p < 0.05). How we validate data is laid out openly on our methodology page.
Conclusion
Sector ETF seasonality is one of the better-documented calendar patterns in the US market: April and November carry most of the seasonal return, while the summer and early-autumn months are statistically silent. Our cross-check confirms the peer-reviewed study in both directions — for the strong months and for the muted ones.
At the same time, the last 20 years show that patterns migrate. December has all but disappeared; July has become notably stronger. That is precisely why seasonality is worth re-checking regularly rather than memorizing once. The interactive heatmaps for every sector are on seasonalpha.ai.
Frequently Asked Questions
Which month is historically best for US sector ETFs?
April. Over 1999–2023, 8 of 9 Select Sector SPDR ETFs posted a statistically significant positive April return, averaging +2.9%. November follows with 6 of 9 significant ETFs and an average of +2.3%.
Does sector seasonality apply to November 2026?
Seasonality describes probabilities drawn from the past, not a forecast. Historically, November was significantly positive for six of the nine sectors, and for technology (XLK) it was green in eight of the last ten completed years. That is context — not a guarantee for any single year.
What are "muted months"?
That is the authors' term for the six months of March, May, June, August, September and October, in which no sector ETF shows a statistically reliable calendar anomaly. In our cross-check, exactly one of 54 individual tests was significant — roughly what chance alone would produce.
Can I build a strategy on sector seasonality?
Statistically, seasonality is a filter rather than a complete trading system. Rotation costs, taxes and the dispersion of individual years eat into the effect. Known anomalies also decay — the December effect, which disappears in the more recent data window, is a case in point.