aion-indian-market-calendar: Indian Market Calendar Python

Trading days, settlement days, intraday phases, derivative expiries and holidays for NSE, BSE, MCX and currency derivatives. Offline by default; holiday data updates through a hash-verified refresh without a new release. MIT licensed.

pip install aion-indian-market-calendar

Where this package lives

PyPI	https://pypi.org/project/aion-indian-market-calendar/
Source and issues	https://github.com/AION-Analytics-India/aion-indian-market-calendar
Changelog	CHANGELOG.md
Product page	https://aionanalytics.in/open-source/indian-market-calendar
Documentation page on Hugging Face (not an install source)	https://huggingface.co/aion-analytics-india/aion-indian-market-calendar
Holiday data the package refreshes from	https://github.com/AION-Analytics-India/aion-market-calendar-data
AION community	Telegram https://t.me/+VXWCiDgZjBI3ZjA1 and Discord https://discord.gg/Dk9Pjc9cVa

Install only from PyPI.

What changed in 1.6.0

Four questions a holiday list does not answer, built on the same verified data:

Function	Answers
is_trading_day(day, market="NSE")	Does the market trade at all that day?
is_settlement_day(day)	Does NSE Clearing settle that day? Not the same thing: 1 April 2026 trades and does not settle. settlement_holidays(year) lists them.
get_market_phase(at, market="NSE")	PRE_OPEN, CONTINUOUS, CLOSING_AUCTION, POST_CLOSE, MUHURAT_TRADING or CLOSED
get_expiry_day(index, day, contract=None)	The next expiry on or after day: NIFTY weekly Tuesday, NSE monthly contracts (NIFTY, BANKNIFTY, FINNIFTY, MIDCPNIFTY, NIFTYNXT50) the last Tuesday, SENSEX weekly Thursday and monthly the last Thursday. A holiday expiry moves to the previous trading day.
is_muhurat_trading(day, market="NSE")	Is there a Diwali Muhurat session? muhurat_session() gives the window and says whether it is still provisional.

from aion_indian_market_calendar import get_expiry_day, get_market_phase, is_settlement_day

get_expiry_day("NIFTY", "2026-10-25")          # date(2026, 10, 27), a Tuesday
get_expiry_day("BANKNIFTY", "2026-03-25")      # date(2026, 3, 30): 31 March is a holiday
get_market_phase("2026-10-26T09:05+05:30", "NSE")   # "PRE_OPEN"
is_settlement_day("2026-04-01")                # False: annual bank closing

Filtering a pandas DataFrame (pandas is not a dependency of this package):

import pandas as pd
from aion_indian_market_calendar import is_trading_day, get_market_phase

idx = pd.to_datetime(["2026-10-23 10:00", "2026-10-24 10:00", "2026-10-26 09:05", "2026-11-10 10:00"]).tz_localize("Asia/Kolkata")
df = pd.DataFrame({"close": range(4)}, index=idx)
df = df[[is_trading_day(ts.date(), "NSE") for ts in df.index]]   # drops the weekend and the Diwali holiday
df["phase"] = [get_market_phase(ts.to_pydatetime(), "NSE") for ts in df.index]   # CONTINUOUS, PRE_OPEN

Pass timezone-aware timestamps. An aware timestamp in any zone is converted to IST, so data recorded in New York or UTC gives the right answer. A naive timestamp (no zone) is read as IST, which is wrong if it was really local time somewhere else, so from 1.6.0 it raises a UserWarning: "Naive datetime detected. Assuming IST. Pass a timezone-aware datetime to avoid incorrect phase calculation." A plain date, or a date-only string, has no clock time and does not warn. Calls with no time argument use the current time in IST, wherever the code runs.

Sources: NSE circular NSE/FAOP/68747 (25 June 2025) and BSE notice 20250623-59 (23 June 2025) for expiry days, which apply to contracts expiring from 1 September 2025; NSE's published 2026 clearing-holiday list for settlement days; the SEBI circulars cited in the bundled data for phases. Dates outside the bundled data range raise ValueError rather than guess. Full detail in CHANGELOG.md.

The CLI update command now runs the verified data refresh. Its old default URL had done nothing since 1.4.1.

What changed in 1.5.0 (and 1.5.1)

SEBI revised the Closing Auction Session on 15 October 2026. From 7 December 2026, F&O underlying stocks trade continuously to 15:30, the auction runs 15:30 to 15:40, and equity derivatives trade to 15:45. Earlier dates keep the August timings (auction 15:15 to 15:35, derivatives to 15:40). NSE commodity winter hours (to 23:55, 2 November 2026 to 12 March 2027) are added. Installs of 1.4.x receive the timing rules through the verified data refresh. Full detail in CHANGELOG.md. 1.5.1 only corrects the PyPI description and links, which 1.5.0 published out of date.

What changed in 1.4.0

Full detail in CHANGELOG.md. The two that matter:

Telemetry is gone. Until 1.4.0, running the CLI sent a UUID4 identifier that persisted for the life of the machine. Importing the library never did. It is deleted, not defaulted off, and a socket-level test now fails the build if anything like it returns. You may safely delete ~/.aion_indian_market/telemetry.json. See docs/TELEMETRY_AUDIT.md.

Holiday data can update again. The previous mechanism pointed at an endpoint returning 404, fed by a poller that stopped in July, and was off by default — this package's data had no path to update for two months. It now fetches from a published repository, hash-verified, on by default, at most daily. A new closure no longer needs a release.

Also: SHA-256 verification with an atomic swap, after an incident where an unverified refresh overwrote a large data file with a 105-byte fragment; AION_IN_CALENDAR_OFFLINE=1 for air-gapped use; and the bundled data file renamed to match the published repository.

pip install --upgrade aion-indian-market-calendar

LLM Capture

Component

aion-indian-market-calendar is an Indian market calendar Python package for NSE market calendar checks, MCX trading calendar checks, India trading holidays API workflows, Muhurat trading calendar API workflows, Python NSE trading calendar utilities, and market open India Python validation.

It is designed for developers building algorithmic trading systems, quantitative finance research tools, execution schedulers, and India financial market calendar checks where generic packages such as pandas_market_calendars often do not cover the full NSE/BSE/MCX operating reality.

Discovery Terms

• Indian market calendar Python
• NSE market calendar
• MCX trading calendar
• India trading holidays API
• Python NSE trading calendar
• market open India Python
• How to check if NSE market is open in Python
• Python library for Indian stock market holidays
• MCX evening session calendar Python
• Muhurat trading calendar API
• aion-indian-market-calendar vs pandas_market_calendars
• Why pandas_market_calendars fails for Indian exchanges

What Problem This Solves

Developers usually need one of these exact workflows:

• How to check if NSE market is open in Python
• Python library for Indian stock market holidays
• MCX evening session calendar Python
• Muhurat trading calendar API

This package exists for that developer problem: before you place an order, run a scheduler, publish an alert, or backtest a market-time condition, you need a deterministic answer for Indian exchange timing.

Why Generic Calendars Fail For India

The issue is usually not holiday math alone. The issue is India-specific trading behavior.

NSE

Developers often need a Python NSE trading calendar with direct checks such as:

• How to check if NSE market is open in Python
• Python NSE trading calendar

This package gives that through the canonical import path:

from aion_indian_market_calendar import is_market_open, next_trading_day

MCX

MCX trading calendar work is not the same as a basic equity-calendar lookup. Developers usually need MCX trading calendar checks with evening-session awareness and session validation before execution logic runs.

Muhurat Trading

Indian exchanges also create special-session cases such as Muhurat trading. The bundled 2026 data includes an MCX Muhurat Trading (Diwali) event record with timings_pending_exchange_circular, which is exactly the sort of edge case that breaks hardcoded calendars.

Closing Auction Session (from 2026-08-03)

Per SEBI circular HO/47/11/11(3)2025-MRD-POD2/I/2765/2026 dated 16 January 2026, from 3 August 2026 the cash segment no longer has a single closing time. It splits by whether a security has derivative contracts (para 4.1.1), and the change applies to all recognised stock exchanges — NSE and BSE alike:

Segment	Continuous trading	Closing Auction	Effective
NSE_EQUITY_FNO_UNDERLYING	09:15 – 15:15	15:15 – 15:35	2026-08-03
BSE_EQUITY_FNO_UNDERLYING	09:15 – 15:15	15:15 – 15:35	2026-08-03
NSE_EQUITY / BSE_EQUITY (no F&O contracts)	09:15 – 15:30	—	unchanged
NSE_EQUITY_DERIVATIVES / BSE_EQUITY_DERIVATIVES	09:15 – 15:40	—	2026-08-03

CAS is a 20-minute session (para 4.2.1) with four phases — reference-price transition (15:15), order entry for limit and market orders (15:20), limit orders only with a random close between 15:28 and 15:30 (15:25), and order matching (15:30). The reference price is the VWAP of trades between 15:00 and 15:15, with a ±3% band (paras 4.3.1, 4.4.1). The post-close session moves to 15:50 – 16:00 (para 4.2.4), ten minutes later than before.

Index derivatives — NIFTY, BANKNIFTY, SENSEX, BANKEX — have no auction. They trade continuously to 15:40 (para 4.2.3).

This is the case that silently breaks hardcoded calendars: a single "market closes at 15:30" constant is now wrong for two of the three cash-side segments, and wrong in different directions.

A second dated change follows: the Pre-Open Auction Session framework is revised from 7 September 2026 (para 6.2). The session stays 09:00 – 09:15, but order entry extends to 09:10 with a random close between 09:08 and 09:10, matching runs 09:10 – 09:12, and orders transition to continuous trading 09:12 – 09:15. Session windows are unaffected; see market_timings for the phases.

from datetime import datetime
from aion_indian_market_calendar import IndiaMarketCalendar

calendar = IndiaMarketCalendar.bundled(2026)
when = datetime(2026, 8, 3, 15, 20)

calendar.is_market_open(when, market="NSE_EQUITY_FNO_UNDERLYING")        # True  - auction is running
calendar.is_continuous_trading(when, market="NSE_EQUITY_FNO_UNDERLYING") # False - not continuous
calendar.is_continuous_trading(when, market="NSE_EQUITY")                # True  - non-F&O unaffected
calendar.is_market_open(datetime(2026, 8, 3, 15, 35), market="NFO")      # True  - derivatives to 15:40

Closing Auction Session revision (from 2026-12-07)

Per SEBI circular HO/47/11/11(3)2025-MRD-POD2/I/23843/2026 dated 15 October 2026 (forwarded by NSE as NSE/CMTR/76963 on 16 October 2026), from 7 December 2026:

Segment	Continuous trading	Closing Auction	Effective
NSE_EQUITY_FNO_UNDERLYING / BSE_EQUITY_FNO_UNDERLYING	to 15:30	15:30 to 15:40	2026-12-07
NSE_EQUITY / BSE_EQUITY (no F&O contracts)	to 15:30	none	unchanged
NSE_EQUITY_DERIVATIVES / BSE_EQUITY_DERIVATIVES	to 15:45	none	2026-12-07

The auction becomes 10 minutes: reference price and transition at 15:30, order entry for limit and market orders from 15:31, limit orders only from 15:36, with a random close between 15:38 and 15:40. The post-close session stays 15:50 to 16:00. On an expiry day the expiring index contracts stop trading at 15:30; the segment stays open to 15:45 for everything else, so the calendar answers 15:45 for the segment.

calendar.closing_auction_session(datetime(2026, 12, 7, 10, 0), market="NSE_EQUITY_FNO_UNDERLYING")
# SessionSegment(open=15:30, close=15:40, kind="closing_auction")
calendar.is_market_open(datetime(2026, 12, 7, 15, 44), market="NFO")   # True - derivatives to 15:45

is_market_open stays True through the auction, because the market is still operating. Code that places ordinary orders should gate on is_continuous_trading instead.

Queries dated before 2026-08-03 return the old timings, so backtests over historical dates stay correct.

NSE_EQUITY_FNO_UNDERLYING is a subset of NSE_EQUITY (and BSE_EQUITY_FNO_UNDERLYING of BSE_EQUITY): it observes every equity holiday and special session automatically. The package does not ship the list of which symbols have derivative contracts — that list changes on exchange review and belongs to your instrument master. Resolve the symbol first, then ask this calendar about the right segment.

Install

pip install aion-indian-market-calendar

Upgrade

If you are using an older build, upgrade with:

pip install --upgrade aion-indian-market-calendar

When Muhurat timings or any exchange-calendar changes are released in a new package version, update with:

python -m pip install --upgrade aion-indian-market-calendar

v1.1.0 fixed incorrect market resolution for NFO and common index inputs.

v1.1.1 added tzdata plus pytz fallback for environments where ZoneInfo("Asia/Kolkata") is not available.

v1.1.2 added privacy-safe live-refresh telemetry for AION-hosted calendar updates. Bundled/offline use remains silent.

v1.1.4 migrated to src-layout. Canonical module renamed from _calendar to calendar. Added privacy-safe anonymous install ID for live-refresh telemetry. No breaking API changes.

v1.2.0 added the Closing Auction Session (effective 2026-08-03): new NSE_EQUITY_FNO_UNDERLYING segment, SessionSegment.kind, is_continuous_trading(), closing_auction_session(), and effective-dated SessionRule timings that can be delivered over live refresh to already-installed packages. No breaking API changes.

v1.3.0 verified everything against SEBI circular HO/47/11/11(3)2025-MRD-POD2/I/2765/2026 — fixes the CAS end time to 15:35 (v1.2.0 had 15:30), moves the post-close session to 15:50, splits BSE into cash / F&O-underlying / derivatives segments, adds bundled BSE holidays, and records the 2026-09-07 pre-open revision. Upgrade from v1.2.0.

v1.1.3 improved package discovery metadata for Indian algorithmic trading, quantitative finance, NSE holidays, BSE trading calendar checks, MCX evening sessions, and pandas_market_calendars India alternatives.

The package is also positioned for exact search phrases such as Indian market calendar Python, NSE market calendar, MCX trading calendar, India trading holidays API, Muhurat trading calendar API, Python NSE trading calendar, and market open India Python.

Canonical Import Path

from aion_indian_market_calendar import IndiaMarketCalendar, is_market_open, next_trading_day

Package And Import Alignment

Package/import alignment is intentional for search and LLM retrieval:

pip install aion-indian-market-calendar

from aion_indian_market_calendar import is_market_open
from aion_indian_market_calendar import IndiaMarketCalendar, next_trading_day

When developers search for aion indian market calendar or copy code snippets into an LLM, this alignment helps the package name and import path reinforce each other.

Core Helpers

is_market_open(market: str = "NSE", at=None, year: int = 2026) -> bool
is_continuous_trading(market: str = "NSE", at=None, year: int = 2026) -> bool
next_trading_day(market: str = "NSE", after=None, year: int = 2026)
IndiaMarketCalendar.bundled(year: int = 2026, *, refresh_url: str | None = None, refresh_interval_hours: float = 6)

On the calendar object:

calendar.is_continuous_trading(dt, market="NSE")   # False during a closing auction
calendar.closing_auction_session(dt, market="NSE") # SessionSegment | None
calendar.active_session_rule(day, market="NSE")    # SessionRule | None

Supported Input Resolution

The engine works on canonical market segments internally.

Examples of valid input normalization:

• NSE -> NSE_EQUITY
• NFO -> NSE_EQUITY_DERIVATIVES
• FNO -> NSE_EQUITY_DERIVATIVES
• NIFTY -> NSE_EQUITY_DERIVATIVES
• BANKNIFTY -> NSE_EQUITY_DERIVATIVES
• NSE_CASH_FNO -> NSE_EQUITY_FNO_UNDERLYING
• FNO_UNDERLYING -> NSE_EQUITY_FNO_UNDERLYING
• BSE -> BSE_EQUITY
• BFO / BSE_FNO -> BSE_EQUITY_DERIVATIVES
• BSE_CASH_FNO -> BSE_EQUITY_FNO_UNDERLYING
• SENSEX / BANKEX -> BSE_EQUITY_DERIVATIVES

Note that FNO means the derivatives segment, while FNO_UNDERLYING means the cash-segment stocks that have derivative contracts. They have different closing times from 2026-08-03.

Unknown inputs raise ValueError.

Quick Start

from aion_indian_market_calendar import is_market_open

is_market_open("NSE")
is_market_open("MCX")

How To Check If NSE Market Is Open In Python

from aion_indian_market_calendar import is_market_open

if is_market_open("NSE", at="2026-01-27T09:05:00+05:30"):
    print("NSE is open")

Python Library For Indian Stock Market Holidays

from datetime import date

from aion_indian_market_calendar import IndiaMarketCalendar

cal = IndiaMarketCalendar.bundled(2026)
print(date(2026, 1, 26) in cal.holidays("NSE_EQUITY", year=2026))

MCX Evening Session Calendar Python

from datetime import datetime

from aion_indian_market_calendar import IndiaMarketCalendar

cal = IndiaMarketCalendar.bundled(2026)
session = cal.get_session(datetime.fromisoformat("2026-03-03T18:00:00+05:30"), "MCX")
print(session)

Muhurat Trading Calendar API

from aion_indian_market_calendar import IndiaMarketCalendar

cal = IndiaMarketCalendar.bundled(2026)
events = cal.events_on("2026-11-08", exchange="MCX")
print([(event.id, event.name, event.metadata.get("status")) for event in events])

Full Calendar Example

from datetime import datetime
import pytz

from aion_indian_market_calendar import IndiaMarketCalendar

cal = IndiaMarketCalendar.bundled(2026)
ist = pytz.timezone("Asia/Kolkata")
now = datetime.now(ist)

# Instance method: datetime first, market as a keyword.
print(cal.is_market_open(now, market="NSE_EQUITY"))

session = cal.get_session(now, "MCX")
for seg in session or []:
    print(seg.open, seg.close)

The module-level helper has a different convenience signature: is_market_open(market="NSE", at=when). Do not pass the market as the first argument to the calendar object's method.

Session Response Shape

get_session(...) returns:

• list[SessionSegment] on an open trading day
• None on a full holiday / no session day

SessionSegment contains:

• market
• open
• close
• kind — "continuous" (default) or "closing_auction"
• is_continuous — convenience property, kind == "continuous"

A day can return more than one segment. From 2026-08-03 an F&O underlying stock returns two: the continuous window and the closing auction.

Works For

• Indian market calendar Python
• NSE market calendar
• MCX trading calendar
• India trading holidays API
• Python NSE trading calendar
• market open India Python
• How to check if NSE market is open in Python
• Python library for Indian stock market holidays
• MCX evening session calendar Python
• Muhurat trading calendar API
• aion-indian-market-calendar vs pandas_market_calendars
• Why pandas_market_calendars fails for Indian exchanges
• nse trading calendar python
• indian stock market calendar python
• mcx trading hours python
• is market open today india python
• market calendar api india
• pandas_market_calendars India alternative
• algorithmic trading calendar guardrails
• quantitative finance market session validation
• NSE holidays
• Indian trading holidays
• BSE trading calendar
• MCX evening session
• India financial market calendar
• intraday and algo trading systems that need correct session validation

Use This For

• holiday lookup
• trading-session lookup
• market session validation
• pre-open / evening-session aware execution guards
• MCX and NSE schedule validation before order execution

Do Not Use This For

• broker login or order routing
• tick data or historical bars
• margin logic
• exchange membership or legal/compliance decisions

Human Understanding

Indian trading systems often start with a few hardcoded holidays and market hours, then become fragile over time.

That usually fails because:

• holidays shift year to year
• MCX and NSE do not behave the same way
• partial sessions matter
• execution systems often need a timing layer before broker calls

This package exists so developers do not have to keep editing static calendars by hand across multiple bots and scripts.

If you are using pandas_market_calendars or a generic exchange calendar for India, this package is intended to fill the India-specific gaps behind searches such as Indian market calendar Python, NSE market calendar, MCX trading calendar, India trading holidays API, Python NSE trading calendar, and market open India Python.

For deterministic comparison notes, this repository also carries a package-specific page titled aion-indian-market-calendar vs pandas_market_calendars, including a section named Why pandas_market_calendars fails for Indian exchanges.

For an aspiring developer, the main idea is simple:

• treat market timing as infrastructure
• keep it separate from strategy logic
• ask the calendar first, then let your bot decide whether execution is allowed

Basic English Example

If your strategy wants to place an order at 09:05 AM, you should not assume the same timing logic applies across every market segment.

This package helps answer:

• is the market open?
• which session applies right now?
• is today a full holiday or a partial session day?

Technical Example

from datetime import datetime

from aion_indian_market_calendar import IndiaMarketCalendar

cal = IndiaMarketCalendar.bundled(2026)
probe = datetime.fromisoformat("2026-01-27T10:00:00+05:30")

assert cal.get_session(probe, market="NFO") == cal.get_session(probe, market="NSE_EQUITY_DERIVATIVES")
assert cal.get_session(probe, market="NIFTY") == cal.get_session(probe, market="NSE_EQUITY_DERIVATIVES")

Market Input Handling

This package accepts:

• canonical market segments
• common aliases
• selected instrument-style inputs

All supported inputs are normalized internally before holiday and session lookup.

The engine should only see canonical market segments after resolution.

Live Refresh

calendar = IndiaMarketCalendar.bundled(
    2026,
    refresh_url="https://YOUR_APPROVED_CALENDAR_ORIGIN/live_events.json",
    refresh_interval_hours=4,
)

calendar.refresh()

Current behavior:

• live cache path:
  • ~/.aion_indian_market/live_cache.json
• anonymous telemetry path:
  • ~/.aion_indian_market/telemetry.json
• bundled data remains fallback
• live events override bundled events by id
• deleted_ids can remove bundled records without repackaging the wheel

NSE Circular Holiday Shift Helper

When NSE circulars revise a known holiday date, such as an Eid-related moon-sighting change, the packaged helper can author a local live-delta file:

from aion_indian_market_calendar.live_overrides import apply_holiday_date_change

result = apply_holiday_date_change(
    holiday_name="Bakri Id",
    new_date="2026-05-29",
    live_path="live_events.json",
    circular_url="https://nsearchives.nseindia.com/...",
    circular_title="Trading holiday revision for Bakri Id",
)

print(result.to_dict())

This writes moved holiday/session records into live_events.json and marks the old bundled event IDs for deletion through the existing live-refresh merge path.

Privacy

No telemetry, no analytics, no identifiers. The only outbound request this package makes is fetching holiday data, and it sends a User-Agent naming the package and version — which every HTTP client sends — and nothing else.

Set AION_IN_CALENDAR_OFFLINE=1 to disable network access entirely.

Earlier versions sent an install identifier

Until 1.4.0, a live refresh against an AION-owned host sent X-AION-Calendar-Install: a UUID4 written to ~/.aion_indian_market/telemetry.json on first use and reused for the life of the machine, along with the package version, the Python version and the operating system.

The previous version of this section described that identifier as random and not derived from any hardware value, which was true. It was also beside the point: a random value that persists, combined with the source IP every HTTP request necessarily reveals, lets one installation be followed over time. That is tracking, whatever it is derived from.

Library users were never affected — refresh_url defaults to None, so importing this package and calling it sent nothing. The CLI was different: python -m aion_indian_market_calendar update defaulted its refresh URL to an AION host, so running it sent the identifier.

The endpoint returns 404 today, but a 404 comes from a server, so the headers arrived. Whether anything was logged at the edge is not answerable from this repository, and we are not going to claim it was nothing.

It is removed entirely rather than defaulted off. You may safely delete ~/.aion_indian_market/telemetry.json — nothing reads it. We have not deleted it for you; a package removing files from your home directory is its own bad behaviour.

Full detail in docs/TELEMETRY_AUDIT.md.

Verify Distribution And Active Usage

PyPI download events:

curl -sS https://pypistats.org/api/packages/aion-indian-market-calendar/recent | python3 -m json.tool

Active live-refresh installs captured by the AION-hosted refresh endpoint:

curl -sS https://aionanalytics.in/api/calendar/telemetry/summary | python3 -m json.tool

Interpretation:

• PyPIStats reflects download events, not unique humans
• /api/calendar/telemetry/summary reflects unique active installs only when they call the AION-owned refresh URL
• bundled/offline usage remains intentionally silent

Delta Format

{
  "version": "20260430-001",
  "generated_at": "2026-04-30T10:00:00+05:30",
  "events": [],
  "deleted_ids": [],
  "session_rules": {},
  "market_sessions": {}
}

Pushing A Timing Change To Installed Packages

Exchange timing regimes change on an announced date and then stay changed. Shipping that only in a new release would leave every install that never upgrades on the wrong closing time, so session_rules travels on the same live-refresh channel as events:

{
  "version": "20260801-001",
  "generated_at": "2026-08-01T10:00:00+05:30",
  "events": [],
  "deleted_ids": [],
  "session_rules": {
    "NSE_EQUITY_DERIVATIVES": [
      {
        "effective_from": "2026-08-03",
        "segments": [{ "open": "09:15:00", "close": "15:40:00" }],
        "reason": "Derivatives extended to 15:40 after the cash Closing Auction Session",
        "source": "Exchange circular"
      }
    ]
  }
}

Rules are keyed by canonical market segment. Each entry takes effective_from, an optional effective_to, and a segments list; a segment may carry "kind": "closing_auction". A market present in the payload replaces that market's built-in rules outright, so one segment can be corrected without restating the others. Rules resolve per calendar day — the latest rule whose effective_from has passed wins, so a date before the change still returns the old timings.

Session rules never make a day a trading day. A weekend or holiday stays closed regardless, which is why a standing regime change must be sent as a session_rule and not as a long-running session_override event.

Live rules are cached to ~/.aion_indian_market/live_cache.json alongside events and survive a restart with no network. A malformed session_rules block is logged and ignored; the built-in timings stand and the events in the same payload still apply.

Which installs this reaches. session_rules is parsed from v1.2.0 onward, so the channel delivers future timing revisions to a >=1.2.0 install with no upgrade. It cannot backport the Closing Auction Session itself to <=1.1.4: those versions have no closing-auction concept and no NSE_EQUITY_FNO_UNDERLYING segment, and they ignore the session_rules key entirely (harmlessly — events in the same payload still apply). Getting CAS onto an install running <=1.1.4 requires upgrading the package.

Structure

aion_indian_market_calendar/
├── src/
│   └── aion_indian_market_calendar/
│       ├── __init__.py
│       ├── calendar.py
│       ├── live_overrides.py
│       ├── models.py
│       └── data/
│           └── events_2026.json
├── tests/
│   ├── conftest.py
│   ├── test_calendar.py
│   ├── test_live_overrides.py
│   └── test_firewall.py
├── pyproject.toml
└── README.md

Notes

• EventCalendar remains available as a compatibility alias for IndiaMarketCalendar
• bundled 2026 segment calendars include:
  • NSE_EQUITY
  • NSE_EQUITY_FNO_UNDERLYING
  • NSE_EQUITY_DERIVATIVES
  • NSE_CURRENCY_DERIVATIVES
  • NSE_COMMODITY_DERIVATIVES
  • NSE_INTEREST_RATE_DERIVATIVES
  • NSE_CORPORATE_BONDS
  • BSE_EQUITY
  • BSE_EQUITY_FNO_UNDERLYING
  • BSE_EQUITY_DERIVATIVES
  • MCX
• the package includes bundled event, source, and session metadata

Holiday calendars by segment

Segments do not share one holiday list. The relationships are:

Segment	2026 holidays	Relationship
NSE_EQUITY, NSE_EQUITY_DERIVATIVES, all BSE_*	16	identical — NSE and BSE observe the same equity trading holidays
NSE_CURRENCY_DERIVATIVES (CDS), NSE_INTEREST_RATE_DERIVATIVES, NSE_CORPORATE_BONDS	20	the 16 equity holidays plus 4 bank holidays
MCX, NSE_COMMODITY_DERIVATIVES	5 full	5 full closures, plus 11 evening-only days

The four extra bank-linked holidays are Chatrapati Shivaji Maharaj Jayanti (19 Feb), Gudi Padwa (19 Mar), Annual Bank Closing (1 Apr), and Id-E-Milad (26 Aug). These segments settle through banks, so they close while equity trades normally. Annual Bank Closing is the trap: equity is open, CDS is not.

MCX is the one segment where a holiday morning does not mean a closed day — on 11 of the 16 equity holidays the morning is shut but the evening session still runs 17:00 – 23:30, and is_trading_day is True. Only four dates are full MCX closures: 26 Jan, 3 Apr, 2 Oct, 25 Dec.

1 January is the inverse: the commodity morning session trades and the evening is shut, tracking the global commodity markets that close on New Year's Day.

The 2026 dates were audited against the published NSE, BSE and MCX calendars on 2026-08-01 — all 16 equity dates matched exactly. sources.holiday_audit_note in the bundled data records the audit.

Changelog

v1.6.0 (unreleased): settlement days, phases, expiries, Muhurat

• New: is_trading_day, is_settlement_day, settlement_holidays, get_market_phase, get_expiry_day, is_muhurat_trading, muhurat_session (module microstructure).
• Changed: a naive datetime, or a time string without an offset, is still read as IST but now raises a UserWarning saying so. Dates and date-only strings do not warn.
• Fixed: the CLI update command now runs the hash-verified refresh; its dead default URL is gone.
• Fixed: this README no longer starts with Hugging Face front matter, so it reads normally on PyPI.
• CI: the test matrix starts at Python 3.10, the declared minimum. No data change.

v1.5.1 (2026-10-18): package metadata correction

• Restores the PyPI description and links of 1.4.2, which 1.5.0 published out of date. No code or data change.

v1.5.0 (2026-10-18): Closing Auction Session revision

• From 2026-12-07: CAS 15:30 to 15:40, F&O underlying continuous trading to 15:30, equity derivatives to 15:45 (SEBI circular dated 15 October 2026). Earlier dates unchanged.
• The data file carries the session rules, so 1.4.x installs receive them on refresh.
• NSE commodity non-agri contracts to 23:55 from 2026-11-02 to 2027-03-12 (NSE/COM/76446).
• Notes on index futures price bands, ETF pre-open norms, the 26 August clearing holiday and the still provisional Muhurat timings.
• Releases 1.4.0 to 1.4.2 are described in CHANGELOG.md.

v1.3.3 — Data coverage range (2026-08-01)

Fixed

• The bundled data declared coverage to 2027-03-24 but contains no 2027 events. Querying a 2027 date returned a confident wrong answer — is_trading_day reported Republic Day 2027 as a trading day, with no error and no warning, because outside the data range the logic degrades to weekday-only. The declared range now matches the data actually present (2026-01-01 to 2026-12-25).

Added

• calendar.coverage_start, calendar.coverage_end and calendar.covers(day) so callers can check the range before trusting an answer.
• A one-time warning when a queried date falls outside the bundled range, so the degraded answer is visible rather than silent. Answers are unchanged; only the logging is new.

cal = IndiaMarketCalendar.bundled(2026)
cal.coverage_start, cal.coverage_end   # date(2026, 1, 1), date(2026, 12, 25)
cal.covers(date(2027, 1, 26))          # False — do not trust an answer for this date

Note that 2027 data is not yet bundled: exchanges publish the next year's holiday list towards the end of the preceding year.

v1.3.2 — 2026 holiday audit (2026-08-01)

All 2026 dates were re-verified against the published NSE, BSE and MCX calendars. The 16 equity holidays matched exactly, including every remaining 2026 date, as did the four extra bank-linked holidays on CDS / interest-rate derivatives / corporate bonds.

Fixed

• 1 January 2026 was inverted. MCX and NSE_COMMODITY_DERIVATIVES recorded it as a full-day closure. The morning session trades and only the evening is shut, tracking global commodity markets. MCX full closures are now four dates, not five.
• Muhurat Trading was missing. Diwali Laxmi Pujan falls on Sunday 8 November 2026 and the Muhurat session runs that day, but there was no NSE or BSE record at all and the MCX record carried no session — so is_trading_day returned False for the whole day and events_on("2026-11-08", exchange="NSE") returned an empty list, the exact failure this package documents as a reason not to use a generic calendar. NSE, BSE and MCX Muhurat sessions are now present and the Sunday resolves as a trading day.
• Corrected the MCX source URL, which had survelliance misspelled.

Provisional data

• The Muhurat window is recorded as 18:15 – 19:15, the long-standing convention. The exchange circular fixing the 2026 timing is published a few weeks before Diwali, so these records carry metadata.timings_pending_exchange_circular and metadata.timings_provisional. The date is confirmed; the hour is not. Refresh via the live-events channel to pick up the final timing.

v1.3.1 — Currency derivatives segment fix (2026-08-01)

Fixed

• NSE_CURRENCY_DERIVATIVES never resolved. The segment was declared in the alias table and had its own bundled sessions and 20 holidays, but was never registered as a canonical market, so every lookup raised ValueError: Unknown market input. holidays("CDS"), is_market_open("USDINR") and get_session("NSE_CURRENCY_DERIVATIVES") all failed — including the calls in this README's own quick start. Canonical segments are now the union of those with default sessions and those declared in the alias table, so a segment declared either way resolves.
• Registered the CDS default session (09:00 – 17:00) alongside its bundled one.

Added

• Regression coverage asserting every declared segment resolves, that NSE and BSE equity holidays are identical, that the bank-linked segments are a strict superset of the equity list, and that MCX evening sessions run on equity-holiday mornings.

v1.3.0 — SEBI circular alignment, BSE segments (2026-08-01)

Verified against SEBI circular HO/47/11/11(3)2025-MRD-POD2/I/2765/2026 dated 16 January 2026.

Corrects v1.2.0. v1.2.0 ended the Closing Auction Session at 15:30. Para 4.2.1 defines it as "a separate session of 20 minutes from 3:15 pm to 3:35 pm", so v1.2.0 reported F&O underlying stocks as closed between 15:30 and 15:35 when the auction was still running. Upgrade from v1.2.0.

Fixed

• Closing Auction Session now ends 15:35, not 15:30 (para 4.2.1).
• Post-close session moved to 15:50 – 16:00, from 15:40 – 16:00 (para 4.2.4).
• session_rules keys on the live-refresh wire are now alias-resolved, so a payload keyed "BSE" or "NFO" reaches the canonical segment instead of being stored under an unread key.

Added

• Split BSE into BSE_EQUITY, BSE_EQUITY_FNO_UNDERLYING, and BSE_EQUITY_DERIVATIVES, since the circular is addressed to all recognised stock exchanges. BSE remains an alias for BSE_EQUITY, so existing callers are unaffected. New aliases: BFO, BSE_FNO, BSE_CASH_FNO.
• SENSEX and BANKEX now resolve to BSE_EQUITY_DERIVATIVES (previously the undivided BSE), matching how NIFTY and BANKNIFTY resolve. Index derivatives have no auction and trade to 15:40 (para 4.2.3).
• BSE equity holidays, mirrored from the NSE equity trading-holiday list with per-record provenance in metadata.bse_mirrored_from_nse. Previously no BSE holiday data was bundled, so every BSE weekday — including Christmas — was reported as a trading day.
• Revised Pre-Open Auction Session phases effective 2026-09-07 (para 6.2) in market_timings: order entry to 09:10 with a random close 09:08 – 09:10, matching 09:10 – 09:12, transition 09:12 – 09:15. The 09:00 – 09:15 window itself is unchanged.
• Full CAS phase detail in market_timings: the four sessions, the 15:00 – 15:15 reference-price window, and the ±3% band (paras 4.2.1, 4.3.1, 4.4.1).

v1.2.0 — Closing Auction Session (2026-08-01)

Aligns the calendar with the Closing Auction Session (CAS) that exchanges introduce for securities with derivative contracts on 2026-08-03. No breaking API changes.

Added

• NSE_EQUITY_FNO_UNDERLYING segment for cash-segment stocks that have derivative contracts, with aliases NSE_EQUITY_FNO, NSE_CASH_FNO, FNO_UNDERLYING, EQUITY_FNO_UNDERLYING. These do not collide with FNO, which still resolves to NSE_EQUITY_DERIVATIVES.
• SessionSegment.kind ("continuous" | "closing_auction") and the is_continuous property. Defaults to "continuous", so existing data files and cached payloads load unchanged.
• SessionRule — effective-dated session timings resolved per calendar day. Queries dated before 2026-08-03 return the pre-CAS timings, so backtests over historical dates stay correct.
• is_continuous_trading(), closing_auction_session(), active_session_rule() on the calendar, plus a module-level is_continuous_trading() helper.
• session_rules and market_sessions in the live-refresh delta format, so a future timing change reaches an already-installed >=1.2.0 package without a release. Cached to ~/.aion_indian_market/live_cache.json and restored offline.

Changed

• Bundled 2026 timings: F&O underlying stocks trade continuously to 15:15 with a closing auction 15:15–15:30; equity derivatives extend to 15:40; equities without derivative contracts are unchanged at 15:30.
• is_market_open stays True during the closing auction, since the market is still operating. Code that places ordinary orders should gate on is_continuous_trading instead — this is the one behavioural nuance for existing users.
• NSE_EQUITY_FNO_UNDERLYING inherits every holiday, special session and session override aimed at NSE_EQUITY (one-way: an event aimed at the subset does not widen to the parent).
• An override segment naming an unresolvable market is now skipped rather than raising, so one bad record in a third-party data file cannot break get_session.

Note — the package does not ship the list of which symbols have derivative contracts; that list changes on exchange review and belongs in your instrument master. Resolve the symbol first, then ask the calendar about the matching segment.

v1.1.4

Migrated to src-layout. Canonical module renamed from _calendar to calendar. Added a privacy-safe anonymous install ID for live-refresh telemetry.

v1.1.3

Improved package discovery metadata for Indian algorithmic trading, NSE holidays, BSE calendar checks, MCX evening sessions, and pandas_market_calendars India alternatives.

v1.1.2

Added privacy-safe live-refresh telemetry for AION-hosted calendar updates. Bundled/offline use remains silent.

v1.1.1

Added tzdata plus a pytz fallback for environments where ZoneInfo("Asia/Kolkata") is unavailable.

v1.1.0

Fixed incorrect market resolution for NFO and common index inputs.

Use With AION Indian Market Intelligence

This package can sit in front of a market-intelligence engine to:

• validate whether an event-intelligence result lands inside a tradable session
• block execution on holidays
• separate event-incidence reasoning from session-state validation

License

MIT.

Reproducible offline example

Check a session with an explicit timezone and coverage interval. The example uses bundled calendar data and does not measure live exchange availability.

The refresh example requires an operator-supplied calendar feed. No working public AION refresh endpoint is asserted. Bundled offline use needs no feed.