The U.S. stock market is open for 251 full sessions in 2026, plus two half-days. So why does every textbook, risk model, and Bloomberg terminal say 252? Because 252 isn’t the literal count for any one year. It’s the long-run industry average (the actual range across years is 250–253). The convention got locked in decades ago, and every annualization formula in finance was built on top of it.
Below: the exact 2026 calendar with every closure, the three formulas you’ll actually use 252 for (annualized return, volatility, Sharpe), and the reason applying 252 to crypto, forex, or non-U.S. exchanges quietly biases half the back-tests on the internet.
At a glance: – 2026 full trading days (NYSE/Nasdaq): 251 – 2026 half-day sessions: 2 (Black Friday Nov 27, Christmas Eve Dec 24) – 2026 holiday closures: 10 – Convention used in formulas: 252 – For crypto annualization: 365 (markets run 24/7) – For spot forex annualization: 252 (closes weekends, runs 24h on weekdays)
The 2026 trading calendar
| Date | Day | Reason | Status |
|---|---|---|---|
| Jan 1, 2026 | Thursday | New Year’s Day | CLOSED |
| Jan 19, 2026 | Monday | Martin Luther King Jr. Day | CLOSED |
| Feb 16, 2026 | Monday | Presidents’ Day | CLOSED |
| Apr 3, 2026 | Friday | Good Friday | CLOSED |
| May 25, 2026 | Monday | Memorial Day | CLOSED |
| Jun 19, 2026 | Friday | Juneteenth | CLOSED |
| Jul 3, 2026 | Friday | Independence Day (observed) | CLOSED |
| Sep 7, 2026 | Monday | Labor Day | CLOSED |
| Nov 26, 2026 | Thursday | Thanksgiving | CLOSED |
| Nov 27, 2026 | Friday | Day after Thanksgiving | HALF-DAY |
| Dec 24, 2026 | Thursday | Christmas Eve | HALF-DAY |
| Dec 25, 2026 | Friday | Christmas Day | CLOSED |
Why 252? The math behind the convention
The number is plain arithmetic over a representative year:
- 365 days in a calendar year
- −104 weekend days (52 Saturdays + 52 Sundays)
- −9 holiday closures (the historical NYSE baseline)
- = 252 trading days
That math worked cleanly until 2021, when Juneteenth was added as a federal holiday. NYSE now closes for 10 days per year, which makes the modern arithmetic 365 − 104 − 10 = 251. The convention didn’t update; rewriting every annualization formula in finance was a worse problem to solve than living with a one-day discrepancy.
The literal count varies between 250 and 252 in any given year depending on:
- Whether it’s a leap year (one extra weekday)
- Which day of the week January 1 falls on (shifts the weekday count by ±1)
- How many holidays fall on weekends and get observed on adjacent weekdays
2026 lands at 251 because every NYSE holiday lands on a weekday (Independence Day on Saturday gets observed Friday July 3, so the closure still happens).
The reason “252” stuck is operational: financial formulas need one fixed factor. If volatility scaling used the literal count of each year, fund factsheets and risk reports would be subtly incomparable across years. So the industry locked in 252 decades ago, and every risk engine, options pricer, and performance system was built on top of that constant.
It’s why a trader will tell you “use 252” while an exchange ops team will tell you “this year has 251.” Both are right — they’re answering different questions.
How to use 252 in real calculations
Three formulas use 252 daily. Getting them wrong is one of the most common silent errors in performance reporting, and not just at the retail level. Institutional reports get the asset-class adjustment wrong all the time.
(1 + 0.0008)252 − 1 = 22.3% annualized return.
Note: this is the geometric form. For an arithmetic estimate, multiply: 0.08% × 252 = 20.2% — a small but persistent gap that grows with daily-return magnitude.
1.0% × √252 = 1.0% × 15.87 = 15.87% annualized volatility.
Why √252? Variance scales linearly with time; standard deviation is its square root. So if daily variance compounds 252× per year, daily volatility compounds by √252 (about 15.87×). This is why doubling the time horizon multiplies your vol by √2 — not 2.
(0.0005 / 0.008) × √252 = 0.0625 × 15.87 = 0.99 Sharpe ratio.
A Sharpe near 1.0 is decent; institutional quant funds target 1.5–2.0+ on diversified strategies. Anything reported above 3 should be assumed wrong until you’ve audited the calculation — it’s almost always due to overfitting, look-ahead bias, or wrong annualization factor.
- Forex (24/5): use 252 — same convention; markets close weekends.
- Crypto (24/7): use 365. Annualizing crypto vol with √252 gives a ~17% under-estimate.
- Commodity futures (CME): use 252 — close to equity convention despite extended hours.
- Intraday strategies: scale by √(252 × intraday-bars-per-day) to annualize bar-frequency volatility.
Get those three right and you’ve covered most of what 252 is used for in practice. The rest is knowing when not to use it.
Trading days across major global exchanges
The 252 convention is U.S.-centric. Other exchanges run between 242 and 256 sessions per year because their national holiday calendars differ enormously. Use the right factor for the market you’re analyzing.
| Exchange | Region | 2026 trading days | Daily session | Calendar quirk |
|---|---|---|---|---|
| NYSE / Nasdaq | United States | 251 | 9:30 AM – 4:00 PM ET (6.5 h) | Reference benchmark for “252” convention |
| TSX | Canada | 250 | 9:30 AM – 4:00 PM ET (6.5 h) | Adds Family Day, Boxing Day; skips U.S. Thanksgiving |
| LSE | United Kingdom | 253 | 8:00 AM – 4:30 PM GMT (8.5 h) | 8 bank holidays + Christmas Day + Boxing Day |
| Euronext | France / NL / BE / IT / IE / PT | 256 | 9:00 AM – 5:30 PM CET (8.5 h) | Closes only on 6 pan-European holidays |
| Deutsche Börse / Xetra | Germany | 253 | 9:00 AM – 5:30 PM CET (8.5 h) | Adds German Reformation & Unity days |
| SIX Swiss | Switzerland | 250 | 9:00 AM – 5:30 PM CET (8.5 h) | Multiple cantonal-religious closures |
| TSE | Japan | 245 | 9:00 AM – 3:00 PM JST + 60-min lunch (5 h) | Up to 16 public holidays incl. Golden Week |
| HKEX | Hong Kong | 246 | 9:30 AM – 4:00 PM HKT + 60-min lunch (5.5 h) | Lunar New Year multi-day, Ching Ming, Buddha's Birthday |
| SSE / SZSE | Mainland China | 242 | 9:30 AM – 3:00 PM CST + 90-min lunch (4 h) | Spring Festival week + Golden Week (Oct) |
| ASX | Australia | 252 | 10:00 AM – 4:00 PM AEDT (6 h) | Australia Day, ANZAC, Queen's Birthday |
| NSE / BSE | India | 251 | 9:15 AM – 3:30 PM IST (6.25 h) | 14–15 holidays incl. Diwali (special evening session) |
| B3 | Brazil | 249 | 10:00 AM – 5:00 PM BRT (7 h) | Carnival multi-day closure (Feb–Mar) |
A common mistake in cross-market portfolio work: applying 252 to every leg. The error compounds fast. A Hong Kong-listed equity annualized with 252 instead of 246 overstates volatility by about 1.2%. A Shanghai equity annualized with 252 instead of 242 overstates volatility by 2%. Across a multi-market portfolio, the cumulative effect on reported risk-adjusted returns can run several Sharpe-decimal points off, which is enough to make a poor strategy look mediocre or a mediocre one look good.
The fix is one rule: for each instrument, use its native trading calendar.
Equities, futures, forex, crypto: when 252 doesn’t apply
Asset classes don’t trade for the same number of hours per year. The annualization factor has to match the actual time-in-market, or the resulting numbers are biased.
By asset class:
- U.S. equities and most futures: use 252. The convention.
- Spot forex (24/5): use 252 despite the longer daily session. FX is closed on weekends, so daily standard deviations calibrate against five days a week, just like equities.
- Crypto (24/7): use 365. Markets never close. Annualizing crypto vol with √252 understates risk by about 17%.
- Equity intraday strategies: scale by √(252 × bars-per-day). A 5-minute strategy on 78 bars/day uses √(252 × 78) ≈ 140 to annualize bar-level vol.
Most retail back-testing platforms quietly default to 252 across every asset class, including crypto. If your back-test reports a 2.5 Sharpe on a Bitcoin strategy, divide it by √(365/252) ≈ 1.20 before getting excited. That’s the platform-induced inflation, not your edge.
Half-day sessions: what they actually mean
Two days a year, U.S. equity markets close at 1:00 PM ET instead of 4:00 PM:
- Day after Thanksgiving (Black Friday)
- Christmas Eve (when it falls on a weekday)
In 2026 both occur, on November 27 and December 24 respectively. They count as full trading days for the purposes of the 252 convention but only contribute about 3.5 hours of active session each.
For most traders, half-days are unremarkable. For three groups, they require attention:
- Algorithmic traders. Schedules, time-decay calculations for options, end-of-day rebalances, and rollover logic all reference the standard 4:00 PM close. If your code uses a hardcoded close time, expect order rejections after 1:00 PM ET on these dates.
- Options traders. Time-to-expiration calculations and theta-decay assumptions break on shortened sessions. Check your platform’s session-aware time-decay model before holding positions across a half-day.
- Liquidity-sensitive strategies. Volume on half-days runs 30–60% below average. Spreads widen. Slippage on large orders increases materially. If you’re size-constrained on a normal day, you’re more size-constrained here.
The day after Thanksgiving is also notorious for unusually thin order books in single-stock options — institutional desks are minimally staffed, and price action can be erratic against very small flow. If you don’t have to trade these days, plan around them.
When 252 wasn’t 252: the historical exceptions
The 252 convention assumes the market runs on a predictable schedule. Several times a generation, it doesn’t.
- The 1968 paperwork crisis. NYSE closed on Wednesdays for several months as back-office systems failed under settlement-volume strain. The whole year’s session count fell well below 250.
- 9/11/2001. U.S. markets closed September 11–14 (Tuesday through Friday) and reopened September 17. Four trading days lost. The week represented the longest unscheduled closure since the Great Depression.
- Hurricane Sandy, 2012. NYSE and Nasdaq closed October 29–30. Two trading days lost. The first weather-driven multi-day closure since 1888.
- Day of mourning closures. Markets close occasionally for state funerals and national days of mourning — most recently for President George H.W. Bush (December 5, 2018) and Jimmy Carter (January 9, 2025).
- COVID-19, 2020. Despite the most extreme market volatility since 2008 — including four trading-halt circuit-breaker triggers in March — the schedule itself stayed normal. No closures.
Most institutional risk models built after 2020 added explicit allowances for cyber-incident closures and infrastructure disruptions. A multi-day unplanned closure no longer sits at the edge of the imagination; it sits in the tail of the risk plan.
How trading days affect different strategies
How much the trading-day count actually affects you depends on what kind of trader you are.
Day traders are most exposed. Every closure is a missed opportunity day, and holiday weeks tend to run on lower volume with less directional follow-through, which favors mean-reversion over breakout setups. Most active day traders drop to half-position size on the day before a closure and skip the half-days entirely.
Swing traders holding through closures carry gap risk. News flow accumulates while markets are shut and gets priced in at the next open, which is why stop-losses sized to intraday volatility get jumped reliably on Mondays after a major weekend headline. The “weekend effect” is old enough to have its own name; it extends to any extended closure.
Long-term investors don’t care about the day count itself but do care about the calendar. Earnings dates, dividend record dates, FOMC meetings, and index-rebalance days all sit on specific trading days, and over a multi-year holding period the alignment matters for tax-lot timing and rebalancing.
Systematic and algorithmic strategies are the most calendar-sensitive of all. Almost every quant model hardcodes 252 sessions for return scaling, vol scaling, and time-decay. When the literal count diverges by one or two days (as in 2026), the bias is small enough to ignore. When the divergence is bigger — leap-year alignment edge cases, multi-day closures, mid-year hours changes — recalibration is the right move.
Earnings seasons and the trading calendar
Four times a year, the U.S. equity market goes through earnings season — roughly a six-week window beginning two to three weeks after each calendar-quarter end. The trading-day implications are:
- Volume concentration. ~40% of annual single-stock volume occurs in the four earnings seasons. The remaining ~60% is spread across the rest of the year.
- Volatility regime shifts. Implied volatility on individual names rises into earnings, collapses on the print, and stays compressed for two to three weeks before the next quarter’s run-up. Cross-sectional dispersion roughly doubles during peak-print weeks.
- Calendar collisions. When a holiday falls inside earnings season — which happens almost every year (e.g., Good Friday in Q1, Independence Day in Q2, MLK Day in Q1) — the typical earnings cadence compresses or shifts. Some firms move report dates forward; others release before market open or after close in unusual windows.
For options traders specifically, knowing the exact 2026 trading-day count between report dates matters because theta-decay calculations are session-based, not calendar-based. A holiday inside an earnings position eats into your premium without giving you a day of underlying drift.
Bottom line
The “252 trading days” you’ve seen in every textbook is a long-run convention, not a literal count. In 2026, the NYSE and Nasdaq run for 251 full sessions plus 2 half-days, with 10 holiday closures. The convention persists because every annualization formula in finance — return compounding, volatility scaling, Sharpe ratios — depends on a fixed factor that holds across years.
Three rules to keep your math honest:
- Use 252 for U.S. equities, futures, and forex. It’s the convention, and the numbers calibrate against it.
- Use 365 for crypto. Markets run 24/7. Anything else under-states risk.
- Use the local count for non-U.S. exchanges. A Hong Kong equity annualized with 252 is wrong by about 1.2%. Across a portfolio, the errors compound.
Get those three right, and your annualized numbers will match the institutional desks. Get them wrong, and you’ll consistently overstate Sharpe and understate volatility — which is exactly how most published “alpha” disappears in production.
