Altaha Screener

A bulk deal is one client trading more than 0.5% of a company’s shares in a day. A block deal is a negotiated trade of at least ₹10 crore inside one of two fixed windows. Both are disclosed after the trade, not live: block deals appear within minutes of their window, bulk deals arrive through the day and complete after the close.

An order is flagged WOW when the value the company disclosed is at least a set share of its market capitalisation — shown on the page, and configurable. Plenty of order disclosures quantify a contract in megawatts or containers and never state a rupee figure at all; those are listed as value not disclosed, and are never ranked or given an assumed number. Nothing here is a recommendation: an order is an event with a size, and the size is all this claims.

Everything on this page is taken from the transcript itself. Forward-looking lines are quoted verbatim and attributed to the person who said them — a commitment put in management's mouth that they did not make would be the worst thing this page could print, so an analyst's question or the moderator's script is never quoted as guidance. Topic emphasis counts how often a subject came up per thousand words: it shows what the call was spent on, which is not the same as what was said about it. A written summary needs a language model; when none is configured the page says so instead of presenting the extraction as one.

Only stakes above 1% appear. A company names a public shareholder in its filing only once they cross 1% of its equity. A position below that can be several hundred crore in a large company and is completely invisible here. Every portfolio on this page is therefore a floor on what someone holds, never the whole of it.

It is quarterly, and it is late. Shareholding patterns are filed once a quarter and land up to 21 days after the quarter ends, so a position shown here can be four months old. It is what was disclosed, not what is held today, and it is not a reason to buy anything.

Names are matched by hand, not by a similarity score. Atul Auto’s register names VIJAY KEDIA and KEDIA SECURITIES PRIVATE LIMITED separately; Elecon’s writes Vijay Kishanlal Kedia. Getting that wrong would mean publishing a false statement about what a named person owns, so every association is curated, and every rolled-up total is shown broken into the entities that make it up. A holding in a family trust is counted but labelled as one — a trust someone is trustee of is not the same as shares they own outright.

A company leaving the list is not necessarily a sale. It has fallen below the 1% disclosure threshold or been sold, and the filing does not say which. This page does not guess.

A different source from the investor portfolios, with different limits. SEBI requires every asset manager to publish each scheme’s full portfolio every month. There is no 1% floor here — a fund’s entire book is disclosed, down to the smallest position — and every row carries an ISIN, so a holding is matched to a listed company by registered identifier rather than by name. Nothing on this page is guessed.

Monthly, and due by the tenth of the following month, so a position here can be about six weeks old. Fresher than the quarterly shareholding filings, still not live.

Coverage is partial and is stated. AMFI lists all 53 fund houses but publishes a directory, not a feed — each pack has to be fetched from that AMC’s own website, and a good many build their download list in JavaScript rather than serving a link. The number read is printed on the page. Fund houses not listed here have not been read; it does not mean they hold nothing.

Listed equity only. A pack also contains debt, government securities and trust units. Those are read but not shown, and anything whose ISIN is not in NSE’s equity list — another fund’s units, an unlisted or BSE-only holding — is listed separately rather than dropped, so the weights reconcile.

NSE publishes, for every stock every day, what share of traded volume was delivered — settled into somebody’s demat account — rather than bought and sold again before the close. Two stocks both up 60% score identically on plain momentum: one rose on 70%-delivery days and was accumulated, the other on 20%-delivery days and was churned by traders flat by 3:30pm.

Delivery is one of five signals. Each becomes a percentile against every other name ranked today, and the five are averaged with equal weight. Equal weighting is a finding, not laziness: weighting by each signal’s measured information coefficient was tested and produced a worse book, because those weights chase whichever signal last worked.