How we verify completeness

A human cannot check tens of thousands of documents. So we run three independent checks — independent on purpose, so a failure in one doesn't quietly pass the others. Every discrepancy carries a written reason.

39/43

sections match SEBI's count

63,524

documents under check

unexplained gaps

1 SEBI's own count against ours

Every paged SEBI listing prints its own record total — “1 to 25 of 2,792 records”. We compare our figure to it, section by section. This is the check you can repeat yourself in a browser in ten seconds.

39 of 43 sections match exactly · 5 flagged for review

2 Document-ID accounting

Every SEBI document URL ends in a number from one global sequence (…_103521.html) — one sequence for everything SEBI publishes, so your documents are interleaved with press releases and speeches. Reporting raw missing numbers would be alarming and useless, so we also index SEBI's out-of-scope listings — not as deliverables, purely to attribute their IDs. What neither your sections nor those classes explain is the residual, and that is the number worth looking at.

ID window 1–103,531 (103,531 slots). 71,927 in your sections, 7,223 attributed to out-of-scope SEBI classes, 24,381 unattributed (76.5% accounted).

3 SEBI's own sitemap

sebi.gov.in/sitemap.xml is an enumeration SEBI publishes independently of its listing pages. Every in-scope URL in it must appear in our index — a second opinion that doesn't depend on us having walked the listings correctly.

Found 21 live documents that SEBI's own listings omit — every one resolves (HTTP 200)

These are published on sebi.gov.in and in SEBI's sitemap, but absent from the section listing they belong to — mostly historical amendment regulations, plus four mutual fund filings under a section SEBI reports as "No record". A vendor walking listing pages alone cannot see them, and would have no way of knowing.

Open items

Section	Finding	Explanation
Debt Offer Document Final filed with ROC	SEBI 604 · Atlas 605 (+1)	1 row(s) more than SEBI declares. SEBI's counter is known to undercount where the same document is listed under more than one month; we keep every distinct URL.
Orders of Chairperson/Members (incl. Adjudication)	SEBI 6,412 · Atlas 6,424 (+12)	12 row(s) more than SEBI declares. SEBI's counter is known to undercount where the same document is listed under more than one month; we keep every distinct URL.
Orders on Insider Trading	SEBI 46 · Atlas 50 (+4)	4 row(s) more than SEBI declares. SEBI's counter is known to undercount where the same document is listed under more than one month; we keep every distinct URL.
Recovery Proceedings	SEBI 19,013 · Atlas 18,935 (-78)	78 row(s) fewer than SEBI declares. Most often a listing row whose link points outside the document paths (for example a direct PDF under /sebi_data/), or a row SEBI counts but does not render. Needs row-level diff before sign-off.
Issues	SEBI 3 · Atlas 2 (-1)	1 row(s) fewer than SEBI declares. Most often a listing row whose link points outside the document paths (for example a direct PDF under /sebi_data/), or a row SEBI counts but does not render. Needs row-level diff before sign-off.

Conduct

SEBI's listings can't be paged with an ordinary script — the site ignores page parameters submitted directly and drives pagination through its own JavaScript. We run a real browser against the site, one page at a time, two seconds apart. robots.txt permits this; it restricts only /js and /css, neither of which we touch. No login, no rate-limit evasion.