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
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 PythonNSE market calendarMCX trading calendarIndia trading holidays APIPython NSE trading calendarmarket open India PythonHow to check if NSE market is open in PythonPython library for Indian stock market holidaysMCX evening session calendar PythonMuhurat trading calendar APIaion-indian-market-calendar vs pandas_market_calendarsWhy 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 PythonPython library for Indian stock market holidaysMCX evening session calendar PythonMuhurat 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 PythonPython 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_EQUITYNFO -> NSE_EQUITY_DERIVATIVESFNO -> NSE_EQUITY_DERIVATIVESNIFTY -> NSE_EQUITY_DERIVATIVESBANKNIFTY -> NSE_EQUITY_DERIVATIVESNSE_CASH_FNO -> NSE_EQUITY_FNO_UNDERLYINGFNO_UNDERLYING -> NSE_EQUITY_FNO_UNDERLYINGBSE -> BSE_EQUITYBFO / BSE_FNO -> BSE_EQUITY_DERIVATIVESBSE_CASH_FNO -> BSE_EQUITY_FNO_UNDERLYINGSENSEX / 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 dayNone on a full holiday / no session day
SessionSegment contains:
marketopenclosekind — "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 PythonNSE market calendarMCX trading calendarIndia trading holidays APIPython NSE trading calendarmarket open India PythonHow to check if NSE market is open in PythonPython library for Indian stock market holidaysMCX evening session calendar PythonMuhurat trading calendar APIaion-indian-market-calendar vs pandas_market_calendarsWhy pandas_market_calendars fails for Indian exchangesnse trading calendar pythonindian stock market calendar pythonmcx trading hours pythonis market open today india pythonmarket calendar api indiapandas_market_calendars India alternativealgorithmic trading calendar guardrailsquantitative finance market session validationNSE holidaysIndian trading holidaysBSE trading calendarMCX evening sessionIndia 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_EQUITYNSE_EQUITY_FNO_UNDERLYINGNSE_EQUITY_DERIVATIVESNSE_CURRENCY_DERIVATIVESNSE_COMMODITY_DERIVATIVESNSE_INTEREST_RATE_DERIVATIVESNSE_CORPORATE_BONDSBSE_EQUITYBSE_EQUITY_FNO_UNDERLYINGBSE_EQUITY_DERIVATIVESMCX
- 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.
