How many trading days are in a year? 2026 counts for 19 markets

The number of trading days in a year is 251 for the US stock market in 2026, and again in 2027: 261 weekdays minus 10 holidays. Finance formulas assume 252, a convenient number near the long-run US average. Elsewhere, how many trading days are in a year depends on local holidays: across 19 stock markets, 2026 runs from 242 in China, Japan and Thailand to 256 in France and Spain.

Early October 2026 shows the spread. China’s market is closed from 1 to 7 October for National Day and reopens on Thursday 8 October, so its October has 17 sessions against 22 in the US. South Korea and Taiwan close on Friday 9 October, Japan, Brazil and Argentina on Monday 12 October and Thailand on Tuesday 13 October, while US stocks keep their usual stock market hours.

How many trading days in 2026, market by market

2026 is 52 weeks plus one extra Thursday, which is why both trading weeks come to 261 days. The bars show what each market subtracts:

France and Spain close on five weekdays: New Year’s Day, Good Friday, Easter Monday, 1 May and Christmas Day. Germany and Italy also close on 24 and 31 December, which France and Spain trade as early closes: that is the whole gap between 256 and 254.

India’s 245 counts regular weekday sessions. Its 16 closures include a special one for a state election on 15 January, and a Sunday session on 1 February sits outside the weekday calendar. All counts are scheduled, as every holidays response notes: "Holiday calendars are indicative and may change."

Real sessions have matched the schedule so far in the US, China, Japan and Saudi Arabia: from 1 January to 6 October 2026, daily bars land exactly on the calendar’s 191, 181, 184 and 187 sessions. For China and Saudi Arabia, the market hours pages list the dates: China’s 2026 market holidays and Saudi Arabia’s Sunday-to-Thursday calendar.

Why China and Japan trade 14 fewer days than France

China and Japan reach 242 in opposite ways. China’s three long breaks, 6 sessions at the lunar new year, 3 for Labour Day and 5 for National Day, add up to 14: the whole gap to France. Japan never closes for more than three sessions in a row, but it closes in every month except June. The faint cells show where each market loses its days:

• Lunar new yearTaiwan closes 7 sessions in a row (12 to 20 February), China 6 (16 to 23 February), South Korea and Hong Kong 3. Taiwan’s February has 12 sessions.
• Golden Week and an autumn runJapan closes 29 April and 4 to 6 May for Golden Week, and 21 to 23 September, where a bridge day joins two public holidays. No other break runs past two sessions.
• EasterFrance, Germany, Italy, Spain and Sweden close on Good Friday and Easter Monday, leaving 20 sessions in April. The US closes only on Good Friday.
• Eid and a Sunday-to-Thursday weekSaudi Arabia closes 5 sessions twice, 17 to 23 March and 24 to 28 May, on a Sunday-to-Thursday week. Its May has 16 sessions, as does Turkey’s.
• Holidays on a weekendThey cost nothing. Turkey has four in 2026. Germany gains two trading days in 2027, 256 against 254, because 1 May and 25 December fall on Saturdays.
• Early closes still countSweden has 5 half days, Hong Kong 3 morning-only sessions, Turkey 3 half days, and the US, France and Spain 2 early closes each. All are trading days.

How many trading days in a month: the US in 2026 and 2027

A US month has 19 to 22 trading days in 2026, 20.9 on average. Finance code uses 21, which keeps 12 × 21 = 252, yet only four of the twelve months have exactly 21. By quarter, 2026 runs 61, 62, 64 and 64; the US market holiday calendar lists the closures behind them.

2027 also has 251, with holidays in new places. Juneteenth, Independence Day and Christmas fall on weekends, so the market closes on Friday 18 June, Monday 5 July and Friday 24 December instead. Friday 31 December 2027 is a trading day, and the one early close is 26 November.

Why 252 trading days per year, and why the US runs closer to 251

A Gregorian year averages 260.89 weekdays (365.2425 × 5 ÷ 7). Before 2022 the US market had nine regular holidays, each costing a weekday except New Year’s Day on a Saturday, not made up on the Friday. That leaves about 252 sessions a year; Juneteenth, a market holiday since 2022, brings it to about 251. Daily bars for US500, the US 500 large-cap benchmark, show how often a real year hits 252:

Single years move around that average with the calendar: a year has 260 to 262 weekdays, depending on the day it starts and whether it is a leap year, and 2022 began on a Saturday. Two one-off closures cost a session each: there is no bar for Wednesday 5 December 2018 or Thursday 9 January 2025.

Rolling windows miss 252 too. The 365 days ending at the 6 October 2026 close hold 251 US500 bars, so a "one-year" series of daily returns ending that day has 250 returns, not 252.

How many trading days in a year for forex and crypto

TickerLayer’s forex and crypto calendars have no holidays to subtract, so their counts are plain calendar arithmetic:

What the day count does to annualized volatility

Annual volatility = daily volatility × √N, with N the trading days in a year. For a 1% daily standard deviation:

Across stock markets the result moves by only 0.44 points, so 252 stands in fairly for any of them. Mixing calendars costs far more: 252 on a 365-day asset understates annual volatility by 16.9%, and 365 on a stock overstates it by 20.4%. Stock beta annualizes with √252, and the Sharpe ratio annualizes the same way: √252 for weekday markets, √365 for crypto.

Count trading days in Python from the holiday calendar

GET /markets/holidays returns a year of closures and early closes for one market. schedule decides the count (closed or early_close); meta.completeness and an empty holidays list tell you when not to count:

{
  "market": "US",
  "year": 2026,
  "timezone": "America/New_York",
  "holidays": [
    {
      "date": "2026-07-03",1
      "name": "Independence Day observed",2
      "schedule": "closed",3
      "is_open": false,4
      "observed": true,5
      "memo": null6
    },
    {
      "date": "2026-11-26",
      "name": "Thanksgiving Day",
      "schedule": "closed",
      "is_open": false,
      "observed": true,
      "memo": null
    },
    {
      "date": "2026-11-27",
      "name": "Post-holiday early close",
      "schedule": "early_close",
      "is_open": true,
      "observed": true,
      "memo": "Early close"
    }
  ],
  "meta": {
    "calendar_version": "v1-static-2026-2027",
    "start": "2026-01-01",
    "end": "2026-12-31",
    "data_notice": "Holiday calendars are indicative and may change.",
    "completeness": "full"7
  }
}

The counter hands the closed dates and the market’s week to NumPy’s business-day counter, so weekend holidays cost nothing and early closes stay sessions. It needs pip install numpy requests, a TICKERLAYER_API_KEY environment variable and one request per market.

"""Count a stock market's scheduled trading days in a year from GET /markets/holidays.

Usage: python count_trading_days.py 2026 US CN SA
"""
import os
import sys

import numpy as np
import requests

API = "https://api.tickerlayer.com"
KEY = os.environ.get("TICKERLAYER_API_KEY")
WEEK = {"SA": "Sun Mon Tue Wed Thu"}  # every other market here trades Monday to Friday


def count(market, year):
    r = requests.get(
        f"{API}/markets/holidays",
        params={"asset": "stocks", "market": market, "year": year},
        headers={"x-api-key": KEY},
        timeout=15,
    )
    r.raise_for_status()
    cal = r.json()
    rows = cal["holidays"]
    # An empty list would count every weekday as a session, so refuse to count it.
    if cal["meta"]["completeness"] != "full" or not rows:
        raise ValueError("no full holiday list for this year")
    week = WEEK.get(market, "Mon Tue Wed Thu Fri")
    closed = np.array([h["date"] for h in rows if h["schedule"] == "closed"], dtype="datetime64[D]")
    early = sum(h["schedule"] == "early_close" for h in rows)
    # Early closes stay trading days, and a holiday outside the week costs nothing.
    days = np.busday_count(f"{year}-01-01", f"{year + 1}-01-01", weekmask=week, holidays=closed)
    return int(days), int(np.is_busday(closed, weekmask=week).sum()), early


if __name__ == "__main__":
    if not KEY:
        sys.exit("Set TICKERLAYER_API_KEY first.")
    args = sys.argv[1:]
    year = int(args.pop(0)) if args and args[0].isdigit() else 2026
    markets = [m.upper() for m in args] or ["US", "CN", "SA"]
    print("market  year  days  closed  early")
    for market in markets:
        try:
            days, shut, early = count(market, year)
        except (requests.RequestException, ValueError) as err:
            print(f"{market:<6}  {year}  not counted: {err}")
            continue
        print(f"{market:<6}  {year}  {days:>4}  {shut:>6}  {early:>5}")

market  year  days  closed  early
US      2026   251      10      2
CN      2026   242      19      0
SA      2026   249      12      0

market  year  days  closed  early
US      2027   251      10      1
DE      2027   256       5      0

np.busday_count takes any two dates, so the same call counts part of a year: from 8 October to 31 December 2026 the US calendar has 59 sessions left and China’s 61. For open or closed right now, /markets/status is covered in the exchange calendar API guide, and the holidays reference lists every parameter.

Six ways to miscount trading days