EducationMarket structure~17 min readUpdated 29 September 2026
The short answer
Seasonality in forex is the study of recurring calendar-based flows — month-end rebalancing, quarter-end index resets, year-end tax effects, and the turn-of-the-month window. Some of these are real and mechanically driven. Most are weak edges at best, and most disappear when you account for transaction costs. The tradeable part is small and specific: month-end and quarter-end flows, the last day of the month plus the first few days of the next, and the tax-year effects in Japan and the US. Everything else is data-mined noise.
Why this lesson is honest about a weak edge
Seasonality is the easiest thing in trading to data-mine. If you look at enough months, days and currency combinations, you will find patterns. The question is whether they are real or whether they are artefacts of a small sample.
Floating exchange rates only go back to 1971 — roughly 55 years of data. That is 660 monthly observations. Split into twelve months and you have 55 data points per month. That is not enough to distinguish signal from noise with any confidence. This lesson teaches the seasonal patterns that have a mechanical driver behind them, and warns you about the ones that do not.
T
Written by the Trade To The Top team|Reviewed 29 September 2026
Seasonality research cross-checked against the BIS Quarterly Review, published research on the turn-of-the-month effect (Ariel 1987, Lakonishok & Smidt 1988, McConnell & Xu 2008), the Federal Reserve Bank of St. Louis FRED historical exchange rate data (1971–present), the Bank of Japan's published research on Japanese fiscal-year-end repatriation flows, the WMR 4pm London fix documentation, and the published FTSE, MSCI and S&P index rebalancing rules. The small-sample warning follows the standard statistical framework for multiple-comparison corrections.
Every trading book has a chapter on seasonality. Most of them are wrong. They take the twelve calendar months, look at 50 years of data, and report the patterns that show up. What they do not report is how many other patterns showed up that they chose not to write about, or how many of them failed out-of-sample. This lesson takes a different approach.
Key takeaways
Seasonality is a weak edge. It works as a bias filter, not as a system.
The turn-of-the-month effect is the most robust seasonal pattern in financial markets. Documented across decades and asset classes.
Month-end flows are real and mechanically driven: corporate hedging, index rebalancing, and the WMR 4pm London fix.
Quarter-end flows are the largest of the rebalancing events. Pension funds and index trackers reposition.
Year-end tax effects matter most in the US (tax-loss selling) and Japan (fiscal-year-end repatriation).
Currency-specific patterns (USD strength in Q4, JPY strength in Q1) exist but have small sample sizes.
Data mining is the trap. If you look at 12 months × 40 currencies × 60 years, you will find patterns. Most are coincidence.
Never trade seasonality alone. A seasonal bias without a technical setup is not a trade.
Backtest on out-of-sample data. The only seasonal patterns worth trusting are the ones that held before 2000 and still hold after 2010.
Regime changes kill seasonality. Rate cycles, central bank policy and structural shifts in flows can invalidate decades-old patterns.
Seasonality is the study of recurring, calendar-driven patterns in price and volume. It is not astrology and it is not magic. It is the observable consequence of institutional behaviour that runs on a schedule.
Three things produce seasonal patterns in FX:
Reporting and settlement cycles. Funds report performance monthly and quarterly. Reporting deadlines create rebalancing deadlines, which create flows.
Fiscal-year cycles. Japan's fiscal year ends in March. The US tax year ends in December. Both produce predictable repatriation and tax-loss-selling flows.
Options expiry and index events. Options expire on the third Friday of every month. Indexes rebalance quarterly. Both force mechanical flows at known times.
These are not statistical artefacts. They are real institutional behaviour, scheduled in advance, and visible in the price action. That is why seasonality is worth understanding — and why it is still a weak edge, because everyone else knows about them too.
The four mechanical flows
Four calendar events produce most of the seasonal behaviour you will see in forex. Learn these and you have covered 80% of the seasonal edge.
The four mechanical flows
01
Month-end corporate hedging.
Multinational corporations hedge monthly payroll, invoices and intercompany transfers. Flow concentrates in the last two business days of the month. Predictable, small, direction-agnostic.
02
WMR 4pm London fix.
The WM/Refinitiv benchmark fix runs at 16:00 London time. Asset managers benchmark their fills to this price. Flow is concentrated in the 5 minutes before the fix. The most-traded five minutes of the FX day.
03
Quarter-end index and pension rebalancing.
Index trackers rebalance to match quarterly index composition. Pension funds rebalance back to target allocations. Both are forced flows. The largest mechanical events of the calendar.
04
Year-end tax and window dressing.
US mutual funds sell losers in December for tax-loss harvesting. Japanese institutions repatriate capital before the March fiscal-year-end. Smaller than they used to be, but still visible.
THE FLOW CALENDAR · WHERE THE PRESSURE SITS
A single month mapped to the four mechanical flows
The flow calendar. Four events dominate: monthly options expiry, month-end hedging, the daily WMR fix, and quarterly rebalancing.
Common mistake
Treating month-end as a single-day event. Month-end flows build over the last three to five days of the month. The peak is on the last business day, but the pressure starts the prior Wednesday or Thursday. If you are trying to trade against month-end flow, you need to know it is coming before the actual day.
The turn-of-the-month effect
The turn-of-the-month effect is the most robust seasonal pattern in financial markets. It is the observation that the last trading day of the month plus the first three trading days of the next month have historically produced returns significantly above the average for the rest of the month.
The effect was first documented in equities (Ariel, 1987; Lakonishok & Smidt, 1988) and has been replicated across bonds, commodities, and currencies. It has held for over a century of US equity data and has been confirmed internationally.
THE TURN-OF-THE-MONTH EFFECT · AVERAGE DAILY RETURN BY DAY OF MONTH
Historical average return across major equity indices · the four-day window dominates
The turn-of-the-month effect. The four-day window produces returns far above the rest of the month. The effect carries over into FX, especially in risk-on pairs.
The effect is real, but three caveats matter:
The magnitude has declined. After publication in 1987, the effect was widely traded, and the average return fell by roughly half. It is still positive, but no longer spectacular.
It is concentrated in specific markets. Strongest in equities, moderate in high-yield FX pairs, weak in safe-haven pairs.
It is directional in risk-on regimes. In a risk-off month, the effect can invert. The mechanism is portfolio inflow at the start of the month, which does not happen when flows are leaving.
The turn-of-the-month effect is the only seasonal pattern worth relying on. Everything else is noise with a good story.
Quarter-end and year-end
Quarter-end rebalancing is the largest mechanical flow event of the calendar. Pension funds, index trackers, and asset managers all have to be at target weights by the last business day of March, June, September and December. Those positions are not adjusted freely; they are mechanically forced.
What that means for FX:
Quarter-end flow
Mechanism
Typical FX effect
Equity rebalancing
Trackers buy/sell to match index weights
Currency of the market with the best quarter gets bought
Pension rebalancing
Funds reset to target allocations (60/40, risk parity)
Losers get bought, winners get sold
Corporate repatriation
Multinationals convert foreign cash back to home currency
Home currency gets bought
Regulatory reporting
Banks window-dress balance sheets before reporting
Reserve currencies strengthen
Year-end is the same phenomenon, amplified. December has three additional effects:
US tax-loss selling. Funds sell losers in November and early December to lock in tax losses. The pattern reverses in January — the "January effect."
Window dressing. Mutual funds sell their losers before year-end reporting and buy recent winners. This is cosmetic, not strategic, but it moves prices.
Japanese fiscal-year-end repatriation. Japan's fiscal year ends on 31 March. Japanese institutions bring capital home in the weeks before, which strengthens the yen against foreign currencies.
Common mistake
Over-reading the January effect in FX. The January effect is a well-documented equity-market pattern (small caps outperforming in January). Its FX equivalent is much weaker. Do not assume the January effect translates into a reliable currency bias. Use it as one input among many, not as a standalone signal.
Currency-specific patterns
Beyond the mechanical flows, there are documented currency-level seasonal patterns. Treat these with caution. The sample size is small and the magnitude is modest.
Pattern
Description
Confidence
USD strength in Q4
Historically the dollar has been stronger in October and November than in spring and summer.
Moderate
JPY strength in Q1
Japanese fiscal-year-end repatriation flows (March) support the yen.
Moderate
AUD and NZD strength in Q1
Commodity currencies have historically done well in January and February.
Weak
GBP weakness in September
UK fiscal year-end (April) is often preceded by tax-related outflows, but September weakness is documented.
Weak
EUR strength in Q1
Some evidence of EUR strength in the first quarter tied to European corporate flows.
Weak
"Sell in May" (equities)
The May–October period has historically underperformed November–April in equities. FX effect is indirect.
Moderate (equities)
Two things to note. First, moderate confidence is still low confidence. A pattern with a 60% hit rate in a 55-year sample is not far from random. Second, these patterns are averages. A month can be 5 standard deviations from the average and still be within the range of the data. Do not bet a large position on a 5-basis-point edge.
SEASONAL RETURNS HEATMAP · MAJOR CURRENCIES BY MONTH
Averages are tiny · the point of the chart is how small the edges are
Seasonal returns by currency and month. Every edge is small. None of them would survive transaction costs on their own.
The data-mining trap
Here is the uncomfortable truth about seasonality. If you look at enough combinations, you will always find patterns. Every calendar has 12 months, 7 days of the week, 52 weeks, and dozens of holidays. Multiply that across 40+ currencies and 55 years of data, and you have hundreds of thousands of possible combinations.
What that means practically:
At a 95% confidence level, 1 in 20 combinations will look significant by pure chance. With thousands of combinations, dozens will look significant. Almost none of them are real.
The patterns that are published are the ones that survived. The ones that failed were never written about. This is called publication bias.
Out-of-sample testing is the only test that matters. A pattern that shows up in the 1971–2000 data and still shows up in the 2001–2025 data is worth considering. A pattern that only shows up in one period is not.
Correction for multiple comparisons is mandatory. A seasonal edge that requires looking at 12 months to find should be treated as though the confidence level were divided by 12.
Seasonality as a filter
RoleSECONDARY
WeightLOW
UseBIAS ONLY
ExampleConfirms a chart setup
WORKSAugments a real edge
Seasonality as a system
RolePRIMARY
WeightHIGH
UseSTANDALONE
ExampleTrade every "strong" month
FAILSNot a real edge
How to use seasonality honestly
Four rules. Follow all four or skip seasonality entirely.
The four rules of seasonal trading
01
Only use patterns with a mechanical driver.
Month-end, quarter-end, WMR fix, options expiry, tax-year-end. If you cannot explain why the pattern exists, do not trade it. Mechanism first, statistics second.
02
Out-of-sample test every pattern.
Split the data. Check the pattern in the first half and confirm it in the second half. If it fails in either, discard it. Half the patterns will fail this test.
03
Use seasonality as a tiebreaker, not a trigger.
Only act when a chart setup already exists and seasonality aligns. Seasonal bias with no setup is not a trade. Setup first, seasonality second.
04
Size down, not up.
Because the edge is weak, the position should be smaller than your standard setup. Reduce by half. If the trade is seasonal-only, do not take it at all. Weak edge = small size.
Worked example — the same setup, two calendar contexts
Setup
H4 bullish order block on EURUSD
Entry
1.0850
Stop
1.0830
Target
1.0920
Risk
20 pips
Reward
70 pips
R:R
3.50 : 1
Scenario A — Setup occurs mid-month, no seasonal alignment.
No calendar flow in either direction. Standard execution. Full size.
Result: Whatever the setup produces, standard variance applies.
Scenario B — Setup occurs during the turn-of-the-month window.
Calendar bias confirms direction. Reduce size by 25% because the seasonal edge is weak.
Result: Slightly better expected outcome, smaller position, similar expected R.
Scenario C — Setup occurs three days before quarter-end rebalancing against the flow.
Seasonal bias opposes the trade. Skip it, or halve size.
Result: Same setup. Calendar context changes the decision.SEASONALITY IS A TIEBREAKER, NOT A TRIGGER.
When this fails
When this fails
When the regime has changed. A rate-hike cycle can invalidate decades-old seasonal patterns overnight. Check the current regime before applying historical seasonality.
When the sample is small. A pattern that relies on 15 years of data is not a pattern. It is a coincidence with a story. Require at least 30 years of out-of-sample confirmation.
When the pattern has been widely published. Once a pattern is on every trading blog, it is arbitraged away. The turn-of-the-month effect lost half its magnitude after 1987 for exactly this reason.
When the cost of trading exceeds the edge. A 3-basis-point seasonal edge is wiped out by the spread. Subtract transaction costs before deciding whether a pattern is real.
When the market is in crisis. In a crisis, calendar flows are overwhelmed by volatility. Seasonality does not survive regime breaks.
If you remember nothing else: the mechanical flows are real, the statistical patterns are mostly noise, and seasonality is a tiebreaker, not a system.
In one box
Seasonality = calendar-based flows. Real, but weak.
Log your seasonal trades in R. Our free trading journal lets you tag entries by calendar context — turn-of-month, quarter-end, month-end — so you can see whether seasonality is actually improving or hurting your results over time. Most traders discover it does not help.
5 questions · immediate feedback · retake any time
Question 01 of 05
What is the turn-of-the-month effect?
Correct: C. The turn-of-the-month effect is a documented pattern where the last trading day of the month plus the first three trading days of the next produce returns significantly above the monthly average. It was first documented in equities in 1987.
Question 02 of 05
Which of these is NOT one of the four mechanical flows discussed in the lesson?
Correct: B. Daily central bank interventions are not scheduled and not part of the four seasonal flows. The four mechanical flows are: month-end corporate hedging, the WMR 4pm London fix, quarter-end rebalancing, and year-end tax and window-dressing effects.
Question 03 of 05
Why is data mining a problem in seasonality research?
Correct: D. With 12 months, 40+ currencies, and 55 years of data, there are hundreds of thousands of possible combinations. At a 95% confidence level, 1 in 20 will look significant by chance. Most "published" seasonal patterns are data-mined artefacts.
Question 04 of 05
How should seasonality be used in a trading plan?
Correct: A. Seasonality is a weak edge. It works as a tiebreaker when a chart setup already exists, but not as a standalone trigger. Size should be reduced, not increased, when the seasonal edge is the primary reason for the trade.
Question 05 of 05
Why did the turn-of-the-month effect decline after 1987?
Correct: C. Once a seasonal pattern becomes widely known, it is traded by enough participants that the effect is arbitraged away. The turn-of-the-month effect lost roughly half its magnitude after being published in 1987. This is a general rule for seasonal patterns.
Interest rates, growth, inflation and current accounts. The four pillars of currency valuation — and why fundamental data moves price in a way the chart alone cannot.