Who Owns These Instruments

Section 1A different question, asked from the other side

The rest of this series teaches where the inputs live: coin counts, senior claims, share counts, and the filings that pin them. This page asks something else, from the other side of the capital structure.

Not what the company owes. Who holds the paper.

That question has a real answer in the United States and a much thinner one nearly everywhere else. Form 13F, Schedule 13D, Schedule 13G and the Section 16 forms together make up the richest holder-side regime on the roster. They are also, in ways that stay invisible until they have cost you something, partial. Four limits do most of the damage.

Section 2The lag is a quarter and change

A 13F is a snapshot of the last day of a calendar quarter, filed up to forty-five days later. That is the design, not a shortfall in it. There is no continuous reporting, no intra-quarter disclosure, and no obligation to say anything when a position opens or closes between snapshot dates.

The consequence compounds across the calendar. For roughly six weeks of every quarter the current institutional ownership picture is a snapshot from a quarter that has already ended. For the two weeks before each deadline it is a partial tape that fills in as stragglers file. A manager who sold the whole position the day after quarter end still appears in the table at full size. A manager who bought the whole position that same day does not appear at all.

Section 3There are no short positions anywhere in this family

This is the limit that does the most damage when it is forgotten.

Form 13F reports long positions only. Short stock is not reported. Written options are not reported. There is no line anywhere in the information table where a short appears.

The trap is specific and it catches people constantly. A row can carry Put in the put or call column. That row does not mean the manager is short. It means the manager holds puts, a long position in puts, reported against the market value of the underlying. Long puts and short stock are economically related. They are not the same disclosure, and only one of them is reported.

The beneficial-ownership schedules do not fill the gap either. Schedules 13D and 13G report voting and dispositive power over a class of registered equity, and beneficial ownership has no negative branch. A filer above the threshold may be hedged into a net position of almost anything, and the schedule will not say.

Section 4Not everything is a reportable security

Form 13F requires reporting only of securities on the SEC's Official List of Section 13(f) Securities, published quarterly. If an instrument is not on that list, no manager files anything about it, whatever the size of the position.

For a treasury issuer with a multi-series preferred stack this is not a footnote. A company can have several preferred series outstanding with only one of them on the list. The convertible notes appear on the list one at a time, each in the quarter it joined, while whole preferred series never appear at all.

State the consequence precisely, because the loose version is wrong

The loose version is that no holder data exists for the off-list series. That is false, and it is falsifiable in about a minute, which makes it a costly thing to publish. Off-list rows do exist. Managers sometimes report securities the form tells them not to report. The instruction is not permissive: securities that are not on the Official List should not be reported on Form 13F. EDGAR accepts the row and nothing flags it.

The accurate statement is about requirement, not existence:

The distinction changes what may be done with the data. A sweep on a listed instrument returns many managers and can support a floor estimate against shares outstanding. A sweep on an off-list instrument can return a single filer, holding a position because that filer chose to file a row nobody required. Present that as institutional ownership and you have built a statistic out of one firm's compliance habit. An off-list position that is identical across consecutive quarters while only its market value moves is what such a sample tends to look like.

Section 5Three aggregation traps

Three distinct ways one tape produces a wrong number.

One manager, several rows

A single filer routinely reports the same instrument on multiple lines: different investment discretion, different internal managers, different sub-advised sleeves. Reading only the first row you find understates the position. Assuming one row per filer is wrong on its face. Group by filer, then sum within the filer.

Originals and amendments both sitting in the index

A restated 13F does not remove the original from EDGAR. Both remain, both are full-text indexed, and a naive sweep counts both. The overstatement is usually small, which is exactly why it survives review: it is the kind of error rather than the size of it that matters, and nothing warns you.

The mechanism sits on the cover page. Amendments come in two flavours, both declared in the amendment-type field. A restatement replaces the entire prior information table. A new-holdings amendment adds rows the original omitted. Treating a restatement as additive double counts the whole filing, and treating a new-holdings amendment as a replacement discards real rows.

Share-class confusion

The one that hurts. The same manager can appear twice in one issuer's record, in two different filings, with two different numbers, on two different instruments: a beneficial-ownership schedule on the common stock, and a 13F row on a preferred series. Those are different securities with different economics and different positions in the claims stack. They are not additive, they are not comparable, and a percentage computed on one cannot be carried to the other.

Worth checking on any issuer with a preferred stack: the five percent regime may have produced nothing at all on the preferred series, because the schedules in the record all report the common class.

Section 6Filing quality is not guaranteed

Managers make mistakes and EDGAR accepts them. A common defect on an equity instrument is an amount coded as a principal amount rather than as shares. The reported value usually makes the intent obvious, but the machine-readable field says something the filer did not mean.

That single defect breaks pipelines in two opposite directions. A pipeline that filters strictly on the shares code silently drops those rows. A pipeline that ignores the field silently mixes units. Neither failure raises anything.

Section 7The family: four forms, four clocks

Where to look

company browse, by form type
  /cgi-bin/browse-edgar?action=getcompany&CIK=<CIK>&type=<FORM>&output=atom

full-text search, which reaches inside an information table
  efts.sec.gov/LATEST/search-index?q=%22<CUSIP>%22&forms=13F-HR&startdt=&enddt=

document archive, for a known accession
  /Archives/edgar/data/<CIK>/<accession-no-dashes>/<filename>

All three want a descriptive user-agent header. Full-text search covers a long back history of filings and indexes the information-table documents themselves, which is what makes instrument-level search possible.

Section 8Reading an information table

Every row is one holding. The fields, in schema order, and where each one bites.

Checking whether an instrument is reportable

The check is mechanical and takes about a minute.

The identifier resolves the instrument by itself

A CUSIP is nine characters: a six-character issuer base, a two-character issue identifier, and a check digit. The issue identifier is assigned per security. Every distinct instrument an issuer has outstanding gets its own. It is not a class code, it is not shared across a family of preferreds, and resolving it does not require inference from coupon, convertibility or price.

Two rows carrying the same CUSIP are the same instrument, whatever the filers typed in the free-text title field. The description shown on the official list is a display label rather than the identifier, and a loose label does not make the identifier ambiguous.

Where two similar instruments cannot be told apart from filer titles, leave the mapping unresolved and say so. An unresolved identifier is a stated gap. A guessed one is a wrong join waiting to be inherited by every surface downstream.

What a single filer can demonstrate at once

A convertible-securities manager holding both an issuer's notes and its convertible preferred shows the whole vocabulary in one document: shares against principal amount in the same table, several rows on the same common identifier that must be summed, an option position reporting under the underlying's identifier with the underlying share count in the amount field and a PUT or CALL designation beside it, and a preferred row mislabelled as common in the free-text field while the identifier stays correct.

Long puts in such a table are long puts. They are not a short position, they are not evidence of one, and they do not net against the common rows in any way the filing discloses.

Section 9Sweeping EDGAR, and the limits of a sweep

An aggregate holder figure comes from a scripted sweep. The method is worth stating so it can be reproduced and attacked.

Limits, stated plainly

Where a quarter is still open, say so. An incompleteness that is stated is a fact about the data. An incompleteness that is not stated becomes a fact about the issuer in the reader's head.

Section 10Venue transparency, and why holder tables appear for some companies only

The US regime described above is the richest holder-side disclosure on the roster, and it is not the norm. Two contrasts from this site's own coverage explain why a holder table appears on some company pages and not others.

Growth market venues are not regulated markets. National major-shareholding rules generally attach to the regulated market, so on a growth multilateral trading facility the statutory flagging thresholds do not reach the issuer at all. What exists instead is narrower and points the other way: the exchange's own issuer rules can place the obligation on the company, and start it at a level far above the statutory thresholds. Below that level there is nothing to read. The same reasoning runs on the Swedish venue covered in the NGM guide. Confirm the threshold and the obligation against the current rule book for the venue in question, not against a practitioner summary.

The same logic runs inside a single issuer. A US-listed company generates holdings tables, beneficial-ownership schedules and insider forms, and several of its own preferred series can still generate no holder data at all, because they are not reportable securities. The regime is not company level. It is instrument level.

The line a surface should carry

For each instrument, one line, with its sources. Reportable or not. If reportable: the holdings reported for the quarter ended on a stated date, across a stated number of managers, deduplicated to the latest filing per manager, described as a floor, against shares outstanding from the issuer's own periodic report. If not reportable: that it is not a Section 13(f) security, that no reporting is required and no population exists, and any off-list rows cited individually by accession and never as a percentage.

Where none exists, say so rather than leaving the instrument off the surface. A missing row and a zero position look identical to a reader, and only one of them is true.