Ownership Arcs — Canonical Reference

One-line definition: Ownership Arcs applies no selection rule and filters nothing: it is a viewing surface over insider, SAST and bulk/block disclosures as the exchanges broadcast them, ordered newest first. It shows who filed what and when. It ranks nothing and forecasts nothing.

1. What it is

Three statutory disclosure streams, shown as they arrive:

• Insider dealing — SEBI Prohibition of Insider Trading Reg 7(2) disclosures: named principals of a company reporting their own transactions in its shares.
• SAST — Substantial Acquisition of Shares and Takeovers Reg 29 (crossing-threshold acquisitions and disposals) and the Reg 31/32 pledge listings.
• Bulk and block deals — the exchange's own same-evening publication of large negotiated and on-market trades.

This lens has no rule. It does not screen a universe, it does not rank, it does not score, and it does not decide which names deserve a reader's attention. A company appears on it because somebody was legally obliged to file, and did. That is the entire admission criterion, and it belongs to the regulator rather than to us.

Where this stands: this is a reading surface over primary filings. Its value is coverage, timing honesty and provenance — being able to see the disclosure and know exactly when it became public — not selection.

2. Our variation vs. the standard technique

There is no standard technique to vary, because there is no technique. What is deliberate here is a set of refusals, and they are the substance of the lens:

• We do not infer intent from a filing. A principal buying is a recorded fact. Why they bought is not in the filing, and nothing on this surface pretends to know it.
• We date by disclosure, never by transaction. Both the insider and SAST streams are keyed on the exchange broadcast date — the moment the market could first have known — not the date the deal happened. Using the transaction date would silently backdate knowledge and leak look-ahead into anything built on top (§6).
• We do not aggregate the three streams into a score. They are different regimes with different obligations, thresholds and clocks. A single "ownership score" over them would be an invention, so there isn't one.
• We count, and we say what we counted. The tiles above the tape are counts over a stated window with a stated definition (§3), not verdicts.

3. How it works (methodology)

The tape. Each stream renders newest-first over a reader-chosen window, with optional reader-chosen narrowing by symbol, by disclosure class, or by a minimum transaction value. 🔴 Every one of those is a control the reader operates. None is applied by default, and none is a house opinion about which filings matter.

The tiles. Above each tape sit counted summaries. Where a tile names a cohort — for example principals who were net buyers and transacted recently — that cohort is a mechanical definition applied to filings that already exist, stated on the tile itself. It is a description of a filing pattern, not a prediction and not a ranking.

What is deliberately absent. No target, no rating, no conviction number, no ordering by attractiveness, and no aggregation across the three regimes.

4. Status, validation & honesty fence

LIVE — and there is nothing here to validate, which is the honest status rather than a gap:

• No selection rule means no backtest is applicable. A viewing surface over filings makes no claim that could be tested for return. This page therefore records no performance number, and none should ever be added without a construct that actually selects something and its own pre-registered study.
• A disclosure is not a signal. That a principal bought, that an acquirer crossed a threshold, or that a block traded is a fact about the past. Reading it as an instruction to act is the reader's inference and not this lens's claim.
• Pledge disclosures are especially easy to over-read. A pledge is a financing fact. It is not a solvency verdict, and this surface does not present it as one.
• Absence is not evidence. A company with no filings in the window has disclosed nothing in that window — it has not been cleared, screened out, or judged.
• No buy/sell/hold, no target, no return claim appears anywhere on this surface.

5. Where it lives (code· routes· DB· timers)

• Routes: /dash/insider (the insider view code)· /dash/sast (the sast view code). Lens keys insider, sast, shp.
• Ingest: the insider events code → insider_events· the sast events code → the Reg 29 and pledge event tables. Bulk/block arrive via the deals feed.
• Receipt: registered in the evidence receipt code as ownership-arcs. Its declared data class covers the bulk/block half; the insider and SAST halves resolve their source and clock through the feed manifest, which is where those are actually written down.
• Manifest: the feed manifest code — the single place each feed's source organisation and clock rule are recorded.

6. Data & provenance

All three streams are primary-source (Guardrail #8) — exchange and regulator filings, no vendor:

🔴 Two provenance facts a reader must carry:

1. The clock is the broadcast, not the deed. A transaction may have happened days before it was disclosed. Every date shown is when the market could know, which is the only date that can honestly be used. 2. Bulk/block has no dated back-archive. The feed is captured going forward. History that was never captured does not exist and cannot be reconstructed — so a thin early record is a collection gap, not a quiet period.

7. Terminology canon

• Ownership Arcs — the umbrella name for the three disclosure surfaces. It is a surface, never a screen, a shortlist or a strategy.
• Insider (here) — a designated person filing a legally required disclosure. It carries no suggestion of wrongdoing; the filing is the compliance act.
• SAST Reg 29 — threshold-crossing acquisition/disposal disclosure. Reg 31/32 — pledge creation/release listings.
• Bulk deal / block deal — exchange-defined large-trade categories; the exchange sets the definition, not us.
• Disclosure date — the broadcast date. 🔴 Not the transaction date. Do not conflate them, and never re-key a table on the transaction date.
• Flagged (tile) — a counted cohort under a stated mechanical definition. Not a rating.

Reader-facing definitions: the metrics glossary.

8. Decision & session history

• Feeds built as primary-source ingests under the standing data-sourcing rule, each with its clock rule recorded in the feed manifest at the time it was added.
• Disclosure-dating chosen over transaction-dating at ingest, deliberately, so nothing downstream can inherit look-ahead.
• 2026-08-31 — this page created so the Evidence Receipt has a truthful rule to quote for L3. It records that no selection rule exists, because authoring one would have put an invention on a member-facing receipt.

9. Open items / frozen work

• Bulk/block back-archive — not obtainable free and dated; the record starts when capture started. Treat early sparseness as a gap.
• No cross-stream aggregation is planned. Anyone proposing an "ownership score" must first say what it would mean across three different statutory regimes and record that as a decision.
• Any future ranked construct built on these feeds needs its own pre-registered, leak-free study in the strategy ledger. None exists today.

10. Sources of truth

• Feeds, sources and clock rules: the feed manifest code — the canonical record.
• Definitions: the metrics glossary.
• Code: the insider view code· the sast view code· the insider events code· the sast events code.
• Index: the README· the origins notes·../../the project's running record.