Short answer: not an official one, but the data is public and machine-readable in two steps. NSE shows the disclosures that promoters, directors and designated persons make under SEBI's insider-trading rules (PIT Regulation 7) on its Insider Trading filings page. The page lists filings. Who traded, how many shares, at what value and what they hold afterwards is not in that list: it is in an XBRL file attached to each filing. On 16 November 2026 one request returned the 30 filings broadcast on 13 November, and one more request per filing returned the transactions.
This post shows both steps with Python that ran on 16 November 2026, the field names you need, and the things that go wrong.
Why people want this as data
Promoter buying, promoter selling and pledges are among the few signals that insiders must disclose within days of trading. Screening sites show them in a browser. If you want them in your own scanner, alert bot, research sheet or model, you need rows: one per transaction, with the person, their category, the quantity, the value and the holding before and after.
The timing matters too. Listed companies close their trading window from the end of each quarter until 48 hours after results, so disclosures bunch up in the weeks after results. In the week of 8-15 November 2026, after the September-quarter results, NSE carried 112 insider-trading filings holding 203 transactions: 131 buys, 57 sells and 15 pledge events, 132 of them by promoters or their group.
Step 1: the list of filings
The filings page fills its table from a JSON endpoint. Dates are DD-MM-YYYY.
import requests
API = "https://www.nseindia.com/api"
s = requests.Session()
s.headers.update({
# python-requests' default User-Agent gets no answer; a browser string works
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 "
"(KHTML, like Gecko) Chrome/141.0.0.0 Safari/537.36",
"Accept": "application/json, text/plain, */*",
})
r = s.get(f"{API}/corporates-pit-gg",
params={"index": "equities", "from_date": "13-11-2026", "to_date": "13-11-2026"},
timeout=60)
r.raise_for_status()
filings = r.json()["data"]
print(len(filings), "insider-trading filings broadcast on 13 November 2026")
for f in filings[:3]:
print(f["broadcastDateTime"], "|", f["symbol"], "|", f["regulation"], "|",
f["typeOfSubmission"], "|", f["xmlFileName"].rsplit("/", 1)[-1])
30 insider-trading filings broadcast on 13 November 2026
13-Nov-2026 23:28:54 | UNIECOM | Regulation 7 (2) | Original | IT_23091_WebXMLFile_20261113_232854688.xml
13-Nov-2026 23:09:05 | 360ONE | Regulation 7 (2) | Original | IT_17593_WebXMLFile_20261113_230905015.xml
13-Nov-2026 20:33:12 | 360ONE | Regulation 7 (2) | Original | IT_17593_WebXMLFile_20261113_203311922.xml
Thirty filings from 23 companies; 28 marked Original and 2 marked Revision. Each item has these fields:
| Field | Meaning |
|---|---|
appId |
NSE's id of the filing; use it to deduplicate |
symbol, companyName
|
the company |
regulation |
for example Regulation 7 (2), the continual disclosure |
typeOfSubmission, prevAppId, revisionRemark
|
Original or Revision, and what a revision replaces |
broadcastDateTime, exchdisstime
|
when NSE received and published it, India time |
xmlFileName |
the XBRL file with the transactions |
ixbrl |
the same filing as a readable HTML page |
Notice what is missing: no person, no quantity, no value. An older endpoint name, corporates-pit, still answered on 15 November 2026 but with an empty list, which is why code written before mid-2026 now finds nothing. NSE serves these filings from 3 May 2026 in the present format; 3,492 were listed by 15 November.
Step 2: the transactions inside each filing
Each XBRL file holds one or more transactions. Every fact carries a contextRef: MainI for the filing as a whole, then Disclosure1, Disclosure2 and so on, one per transaction. Group the facts by that reference and you have rows.
import xml.etree.ElementTree as ET
from collections import defaultdict
SHORT = {
"NameOfThePerson": "person",
"CategoryOfPerson": "category",
"SecuritiesAcquiredOrDisposedTransactionType": "type",
"ModeOfAcquisitionOrDisposal": "mode",
"SecuritiesAcquiredOrDisposedNumberOfSecurity": "quantity",
"SecuritiesAcquiredOrDisposedValueOfSecurity": "value_inr",
"SecuritiesHeldPriorToAcquisitionOrDisposalNumberOfSecurity": "held_before",
"SecuritiesHeldPostAcquistionOrDisposalNumberOfSecurity": "held_after",
"SecuritiesHeldPostAcquistionOrDisposalPercentageOfShareholding": "held_after_fraction",
"DateOfAllotmentAdviceOrAcquisitionOfSharesOrSaleOfSharesSpecifyFromDate": "trade_from",
}
def transactions(xml_bytes):
"""One dict per transaction: facts are grouped by contextRef (Disclosure1, Disclosure2, ...)."""
rows = defaultdict(dict)
for el in ET.fromstring(xml_bytes).iter():
ctx, name = el.attrib.get("contextRef", ""), el.tag.split("}")[-1]
if ctx.startswith("Disclosure") and name in SHORT and el.text:
rows[ctx][SHORT[name]] = el.text.strip()
return [rows[k] for k in sorted(rows, key=lambda c: int(c[10:]))]
xml = s.get(filings[0]["xmlFileName"], timeout=60).content
for t in transactions(xml):
print(t)
{'category': 'Promoter and Director', 'person': 'Rohit Kumar Bansal', 'held_before': '128074', 'quantity': '100', 'value_inr': '8430', 'type': 'Buy', 'held_after': '128174', 'held_after_fraction': '0.0011', 'trade_from': '2026-11-11', 'mode': 'Market Purchase'}
{'category': 'Promoter and Director', 'person': 'Rohit Kumar Bansal', 'held_before': '128174', 'quantity': '2886', 'value_inr': '244933', 'type': 'Buy', 'held_after': '131060', 'held_after_fraction': '0.0012', 'trade_from': '2026-11-12', 'mode': 'Market Purchase'}
{'category': 'Promoter and Director', 'person': 'Rohit Kumar Bansal', 'held_before': '131060', 'quantity': '6340', 'value_inr': '543254', 'type': 'Buy', 'held_after': '137400', 'held_after_fraction': '0.0012', 'trade_from': '2026-11-13', 'mode': 'Market Purchase'}
One filing, three open-market purchases on three days by a promoter-director of Unicommerce eSolutions. The filing-level facts (MainI) add the ISIN (ISINCode), the BSE scrip code, the regulation and RevisedFilling.
The gotchas
- The list has no numbers. Everything a screen shows (person, quantity, value, holding) needs the second request. Thirty filings are thirty-one requests.
-
Percentages are fractions.
0.0011is 0.11% of the company, not 0.0011%. Multiply by 100. - One filing, several transactions. Do not treat a filing as a trade. The 112 filings of 8-15 November held 203 transactions.
-
Pledges are not sales.
SecuritiesAcquiredOrDisposedTransactionTypealso carries pledge creation, pledge release and revocation; read it together withModeOfAcquisitionOrDisposal(market purchase, off market, ESOP, gift, inter-se transfer). -
Revisions. A
Revisionfiling restates an earlier one (prevAppId). Keep both and mark the old one, or your totals double. - Go slowly. The file archive refuses a client that asks too fast, for some minutes and for every NSE address at once. Three files at a time and a few per second has been safe for us; fifty a second was not.
-
Element names carry NSE's spelling.
SecuritiesHeldPostAcquistionOrDisposal...is spelled that way in the file. Copy the names, do not retype them. - Takeover disclosures are a different list. Acquirers crossing 5%, and holders above 5% moving 2% or more, file under the Takeover Regulations (Regulation 29); NSE lists those separately, with the figures in the list itself.
What NSE offers officially
NSE publishes an RSS feed for insider trading among its 23 feeds; when I measured it on 10 November 2026 it held 16 items, about one hour of filings, without the transaction figures. NSE's licensed data is sold by NSE Data & Analytics; its Paid Corporate Data page lists the products and their yearly fees. NSE publishes Terms of Use for its website; read them before you build on the endpoints above.
BSE publishes its own table of the same disclosures at bseindia.com, with a download of the table.
If you would rather not maintain it
I maintain the two steps above as a hosted Actor on Apify, NSE Corporate Announcements API. With reports: ["insiderTrading"] it returns one row per transaction, with numbers as numbers and percentages as percentages, and with onlyNew a scheduled run delivers each transaction once.
from apify_client import ApifyClient
client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("data_daemon/nse-corporate-announcements").call(run_input={
"reports": ["insiderTrading", "takeoverDisclosures"],
"daysBack": 7,
"categories": ["Market Purchase"], # open-market buys only; leave empty for everything
"maxItems": 0,
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
print(row["date"], row["symbol"], row["personName"], row["personCategory"],
row["transactionType"], row["quantity"], row["valueInr"], row["holdingAfterPercent"])
2026-11-14 RELTD Anuradha Ravindra Kulkarni Promoter Group Buy 5523 801382 0.07
2026-11-14 RBZJEWEL Rajiv Nitin Mehta Director Buy 2140 404211 0.02
2026-11-14 HAVELLS QRG Investments and Holdings Limited Promoter Buy 374060 433890897 41.69
Those are three of the 123 open-market purchases among the 203 transactions of 8-15 November 2026.
It costs $3 per 1,000 rows: the week of 8-15 November 2026 (203 insider transactions and 77 takeover disclosures) was $0.84, and a daily feed is about $3.50 a month. Apify's free plan covers that. A ready-made run of last week's trades is on the example page.
Since 16 November 2026 the same Actor also returns quarterly results as numbers (reports: ["financialResults"]: revenue, profit, EPS, balance sheet and cash flow from each results filing's XBRL file). The same series covers NSE announcements, corporate actions and bulk deals.
Data is NSE's public disclosure record, shown for information; it is not investment advice.
Top comments (0)