Form NSAR-BT Files Dataset

The Form NSAR-BT Files Dataset is the closed historical archive of transitional annual reports filed on Form NSAR-BT (and its amendment variant Form NSAR-BT/A) by registered management investment companies under Section 30 of the Investment Company Act of 1940. One record represents one accepted EDGAR submission — a single registrant's report covering the irregular stub period created when an open-end fund, closed-end fund, or other management company changes its fiscal year-end. The corpus is intrinsically bounded: it begins with the EDGAR opening in 1994 and ends at the SEC's June 1, 2018 rescission of the entire N-SAR family under Investment Company Reporting Modernization Releases 33-10231, 34-79095, and IC-32314, after which Form N-CEN absorbed the census-style annual reporting role. Each record packages a per-filing metadata.json side-car with the original EDGAR documents — primarily the legacy SGML-wrapped answer.fil carrying the fixed-position N-SAR questionnaire response, plus any TXT, HTML, or PDF exhibits that accompanied the submission.

Dataset APIs

Dataset Files

What This Dataset Contains

The dataset captures every accepted Form NSAR-BT and Form NSAR-BT/A submission across the full life of the form. NSAR-BT is the transitional-period variant of the management-company N-SAR: its content schema is identical to the regular annual NSAR-B, but the period of report is the irregular window between a registrant's prior fiscal year-end and a new fiscal year-end. The form is a numbered questionnaire of roughly ninety items with many sub-items — not a narrative document — and covers registrant identification, service-provider identity, distribution and sales-load arrangements, brokerage activity, sales and redemptions of fund shares, expense and income data, balance-sheet line items, average net assets, yields, governance, and insurance.

Because the form was rescinded effective June 1, 2018, the corpus is closed and runs from the 1994 EDGAR opening through mid-2018. The dataset is distributed as monthly ZIP containers; each container holds the accession folders for filings accepted that calendar month. File types found inside accession folders are TXT, JSON, FIL, HTML, and PDF. Image attachments from the original submissions are excluded from the dataset copy, and there are no XBRL instance documents because N-SAR predates and was never brought under the structured-data mandate.

Content Structure of a Single Record

What one record represents

One record is one accepted EDGAR submission of Form NSAR-BT or NSAR-BT/A. The unit of identity is the EDGAR accession number, materialized as a folder named with the 18-digit accession with dashes stripped (for example 000104106216001248, corresponding to canonical accession 0001041062-16-001248). The folder holds a per-filing JSON side-car and the original EDGAR documents that constituted the submission, less any image attachments. Folders are grouped under a year-month parent (e.g. 2016-08/) inside monthly ZIP containers, but the record itself is the accession folder and its contents — not the container and not the year-month grouping.

What the underlying filing is

Form N-SAR is the historical "Semi-Annual Report for Registered Investment Companies", a census-style questionnaire filed by registered investment management companies (open-end funds, closed-end funds, and small business investment companies) under Section 30 of the Investment Company Act of 1940 and Sections 13 and 15(d) of the Securities Exchange Act of 1934. NSAR-BT is the transitional-period variant: it is filed when a management company changes its fiscal year-end and must report on the irregular stub window between the prior fiscal year-end and the new one. Its content schema is identical to the regular annual variant (NSAR-B); only the period of report differs. NSAR-BT/A is a post-acceptance amendment that re-files or corrects a previously accepted NSAR-BT.

Content layers inside a single record

Each accession folder contains two layers of information:

1. A machine-readable JSON side-car (metadata.json) describing the EDGAR header for the submission and, where applicable, the SEC series and class taxonomy for the constituent funds and share classes inside the trust.
2. The original EDGAR documents for the submission. The canonical document is the N-SAR answer file (conventionally answer.fil), which carries the questionnaire body inside the historical SEC SGML wrapper around a fixed-position answer encoding. When the submission included exhibits (cover letters, auditor consents, signed amendment pages, attached schedules), those documents accompany answer.fil in their original formats.

In practice, JSON is always metadata.json; FIL is always the N-SAR answer document; TXT, HTML, and PDF appear as supporting documents on filings that included exhibits beyond the bare answer file. Image files from the original submission are stripped. N-SAR pre-dates the structured-data mandate, so linkToXbrl is always empty in metadata.json and dataFiles is always an empty array.

metadata.json — fields and meaning

metadata.json is a single JSON object describing the EDGAR submission as a whole. Its stable keys are:

• formType — "NSAR-BT" for original transitional reports or "NSAR-BT/A" for amendments.
• accessionNo — canonical EDGAR accession with dashes (e.g. 0001041062-16-001248).
• linkToFilingDetails, linkToTxt, linkToHtml — URLs into sec.gov/Archives for the primary document, the full-submission .txt wrapper, and the EDGAR filing-index page respectively.
• linkToXbrl — present but empty for this form family.
• description — human-readable form description (e.g. "Form NSAR-BT - Transitional annual report for management companies").
• filedAt — ISO-8601 timestamp with timezone offset of the EDGAR acceptance moment.
• periodOfReport — YYYY-MM-DD end date of the transition period being reported.
• id — opaque 32-character hex identifier internal to the dataset.
• documentFormatFiles — array of every document in the submission, each entry carrying sequence (string; occasionally a single space for the complete-submission wrapper), size in bytes (string), documentUrl, type (form-type code, blank for the wrapper), and an optional description.
• dataFiles — XBRL/data-exhibit array; always empty for NSAR-BT.
• entities — array of filer-side entity records. Each entry includes companyName (with role suffix such as (Filer)), cik (zero-padded 10-digit), fileNo (Investment Company Act file number, e.g. 811-03447), irsNo, stateOfIncorporation, act (Securities Act code, 40 for the Investment Company Act), fiscalYearEnd in MMDD form, type (form type filed by this entity), and filmNo (EDGAR film number assigned at acceptance).
• seriesAndClassesContractsInformation — array of fund series belonging to the registrant trust. Each series entry carries a series ID of the form S followed by nine digits, a name, and a classesContracts array. Each class entry has a classContract ID of the form C followed by nine digits, a name, and a ticker. This is the SEC series/class taxonomy used for mutual-fund trusts that hold multiple funds, each with multiple share classes; the series and class identifiers join directly to the trailing series and class coordinates inside the answer file.

answer.fil — the N-SAR body

answer.fil is the canonical NSAR-BT response file. It uses the historical SEC SGML document wrapper around a fixed-position, line-based answer encoding. The wrapper is a <DOCUMENT> block with <TYPE>NSAR-BT, a <SEQUENCE> integer, a <FILENAME> (typically answer.fil), and a <TEXT> payload. Inside <TEXT> the body is a sequence of N-SAR answer lines, paginated by <PAGE> PAGE N markers (printed-page breaks with no semantic payload), and terminated by free-form SIGNATURE and TITLE trailer lines that attest the report.

Answer-line grammar

Every data line follows the fixed-column pattern NNN SCCCCCC VALUE:

• NNN (columns 1–3) is the N-SAR item number (e.g. 001, 022, 072), mapping directly to the numbered questions on Form N-SAR.
• A single space at column 4.
• S (column 5) is the sub-item letter, or a space when the item has no sub-items. Items that exceed twenty-six sub-items use two-letter extensions, e.g. 072AA, 072BB, 072CC continuing past 072Z.
• CCCCCC (columns 6–11) is a six-digit occurrence/series/class coordinate. By convention the first two digits encode the occurrence within a repeating sub-item (such as the n-th broker, n-th adviser, or n-th counterparty), the next two encode the series number within the trust, and the last two encode the share class. The value 000000 denotes a fund-wide answer; 000100 encodes "first series, no class breakdown"; 000101 encodes "occurrence 00, series 01, class 01"; alphabetic placeholders such as 00AA00 also appear for certain fund-wide tag positions.
• A single space, then the answer value. Values are typed by item: a Y/N flag, a free-text string (company name, address, broker name), an MM/DD/YYYY date, an integer (often right-aligned and zero-or-blank padded), or a decimal number.

Because per-series answers always carry the trailing series suffix, multiple funds in a multi-series trust are interleaved within the same item number rather than grouped into separate sections; reconstructing a per-fund view of the report requires demultiplexing on the last two digits of the coordinate and joining back to the seriesAndClassesContractsInformation entries in metadata.json.

The file ends with trailer lines using a 12-column label followed by free text:

          1
          SIGNATURE   ARTHUR RAMANJULU
        

          2
          TITLE       CFO
        

These are not numbered N-SAR items; they are the attestation block.

Item groups inside the questionnaire

The N-SAR item dictionary partitions cleanly into functional ranges:

• 000 — registrant header. Period-of-report date, CIK, amendment flag, multi-series flag.
• 001–007 — registrant identity and series enumeration: legal name, address, telephone, e-mail, file number, and one row per fund series with name and money-market flag.
• 008–015 — service-provider identity blocks: investment adviser, sub-adviser, sub-administrator, principal underwriter, fund accountant, custodian, transfer agent, and similar. Each block is a repeating tuple of company name, file number, city/state/zip, and a relationship/affiliation code.
• 018–021 — distribution arrangements: sales-load configuration, contingent deferred sales charges, and 12b-1 plan flags.
• 022 — broker/dealer brokerage-commissions table: one row per counterparty with sub-codes A (name), B (IRS number), C (purchases value), D (sales value).
• 024–027 — money-market, joint-account, and affiliated-transaction flags.
• 028 — sales and redemptions of fund shares, structured as a four-quarter by eight-row matrix per series with sub-codes A–H for shares sold/redeemed and dollar amounts (gross sales, redemptions, reinvested distributions, exchanges in/out).
• 030–045 — front-end and contingent sales loads, redemption fees, exchange privileges, and shareholder-services details.
• 048 — advisory-fee schedule with breakpoints per series.
• 049–069 — yes/no compliance flags spanning rule 12d3 transactions, fund-of-funds structures, master-feeder structures, distribution plans, code of ethics adoption, and similar policy disclosures.
• 070 — portfolio-transactions matrix indicating which security types the fund traded, with A–R rows (e.g. U.S. governments, municipals, corporates, equities, options, futures, repos) across two columns per series.
• 071 — aggregate purchases, sales, and portfolio turnover dollars per series.
• 072 — per-series income statement: dividends, interest, advisory fees, distribution-plan fees, transfer-agent fees, custodian fees, registration fees, professional fees, distributions paid, and similar, with sub-codes running A through EE.
• 073 — per-share distributions (ordinary income, capital gains).
• 074 — per-series balance-sheet line items (assets, liabilities, capital).
• 075 — average net assets per series.
• 076–086 — yields, fidelity bond carrier and coverage amount, errors-and-omissions insurance, audit-committee composition, independent public accountant identity, and related governance disclosures.

Included content

Each record packages the metadata.json side-car with the original EDGAR documents for the submission: always the N-SAR answer.fil answer document, plus any additional exhibits as TXT, HTML, or PDF documents (cover letters, auditor consents, attached schedules, signed amendment pages, amendment narratives) when the filer included them. The side-car preserves filer identifiers, file numbers, the period of report, document URLs back to EDGAR, and the series/class taxonomy needed to interpret the per-series coordinates inside the answer file.

Excluded or separate content

Image files from the original submission are removed from the dataset copy. There are no XBRL instance documents because N-SAR was never under the XBRL mandate; both linkToXbrl and dataFiles reflect this absence. The dataset does not redistribute SEC index files; navigation back to EDGAR is via the URLs preserved in metadata.json.

Schema variation across the corpus

Although the NSAR-BT item dictionary is fixed, the corpus spans roughly twenty-four years (1994 through mid-2018) and several content shifts surface across that span:

• Governance and insurance items added in later N-SAR rule amendments — master-feeder flags, several code-of-ethics flags, several insurance items in the high-07x–08x range — appear blank or N in pre-amendment reports and become populated after the corresponding rule changes.
• The advisory-fee breakpoint schedule in item 048 and the expanded portfolio-transactions matrix in item 070 likewise reflect later expansions of the questionnaire.
• The SEC series/class taxonomy (the S-prefixed series IDs and C-prefixed class IDs in seriesAndClassesContractsInformation) only entered EDGAR submissions after the SEC adopted the series/class identification regime in the mid-2000s; earlier filings carry no seriesAndClassesContractsInformation block, and the trailing series/class digits inside answer-file coordinates are still present but do not join to externally identified entities.
• Amendments (NSAR-BT/A) follow the same internal layout as originals; the amendment relationship is signaled only by the form-type code in metadata.json and by header items in the 000 block — there is no diff payload.

Data-format stability over time

NSAR-BT was filed under EDGAR's pre-modern document conventions for its entire life. The answer-file body is the same fixed-column SGML-wrapped text encoding from 1994 through 2018: the <DOCUMENT>/<TYPE>/<SEQUENCE>/<FILENAME>/<TEXT> wrapper, the <PAGE> pagination markers, the NNN SCCCCCC VALUE answer grammar, and the trailing SIGNATURE/TITLE block all remained stable. There was never an HTML or iXBRL transition for the answer document itself; HTML and PDF appear in the dataset only as occasional companion exhibits, never as a re-rendered form body. Structured consumption of the questionnaire therefore requires line-by-line parsing of answer.fil against the N-SAR item dictionary regardless of filing year, and the only natively machine-readable artifact in a record is the dataset's own metadata.json side-car.

Interpretation and extraction notes

Several practical points govern parsing and joining:

• The series/class coordinates in answer-file lines are positional. Reconstructing per-fund views requires splitting CCCCCC into occurrence, series, and class digits and joining the series/class portions to seriesAndClassesContractsInformation in the side-car.
• Item 072 is the densest payload: the per-series income statement runs through sub-codes A–Z and continues with two-letter extensions (AA, BB, ... up to roughly EE). Parsers must accept multi-character sub-item codes rather than a single-letter slot.
• Numeric values inside repeating per-series items always end with a two-digit series suffix; the absence of that suffix is the marker that the answer is fund-wide rather than per-series.
• <PAGE> lines must be skipped during parsing; they reflect the form's printed-page heritage and carry no payload.
• The SIGNATURE and TITLE trailer lines are not numbered items and use a different column layout (12-column label, then free text); treating them as malformed answer lines is a common parsing error.
• The complete-submission .txt wrapper sometimes appears in documentFormatFiles with sequence set to a single space rather than an integer; consumers should accept that as an EDGAR convention.
• NSAR-BT/A amendments do not carry diff information; comparing an amendment to its prior NSAR-BT for the same registrant requires a record-to-record join on filer CIK plus periodOfReport.
• Because NSAR-BT is filed only when a registrant changes its fiscal year, an individual registrant typically appears at most once in the corpus, and the per-accession record is effectively the entire view of that transition period.
• The accession-number folder name decomposes as 10-digit filer/agent CIK + 2-digit year + 6-digit sequence within that filer-year, which is useful for grouping all submissions made through the same filing agent without parsing the JSON.

Who Files or Publishes This Dataset, and When

Who files

The filer of an NSAR-BT submission is a registered management investment company under the Investment Company Act of 1940 — the registrant entity (trust, corporation, or limited partnership) itself, not its service providers. The population includes:

• Open-end management investment companies (mutual funds), filed at the registrant level, with series-level data captured inside the report.
• Closed-end management investment companies, including listed and unlisted closed-end funds and interval funds.

Form N-SAR was not used by:

• Unit investment trusts (separate reporting regime; not management companies).
• Face-amount certificate companies (a distinct 1940 Act class).
• Business development companies (not registered under the 1940 Act).
• Private funds relying on Section 3(c)(1) or 3(c)(7), and unregistered foreign funds.

Investment advisers, sub-advisers, principal underwriters, administrators, custodians, transfer agents, and independent accountants are identified within the report but are not filing parties.

Statutory and rule framework

Form N-SAR (and the BT variant) was required under a layered authority set:

• Section 30 of the Investment Company Act of 1940 (15 U.S.C. 80a-29) — periodic reporting authority for registered investment companies.
• Sections 13 and 15(d) of the Securities Exchange Act of 1934 (15 U.S.C. 78m, 78o(d)) — periodic reporting for issuers with registered securities or effective registration statements; N-SAR served this function for registrant-issuer funds.
• Rule 30a-1 — annual N-SAR filing obligation for management investment companies.
• Rule 30b1-1 — semi-annual N-SAR filing obligation.
• Rule 30b2-1 — shareholder reports filing mechanics intersecting the N-SAR framework.

The "BT" suffix is the transitional annual variant of the management-company N-SAR — the Rule 30a-1 annual filing adapted to an irregular, less-than-twelve-month period.

Triggering event

NSAR-BT is generated by a single trigger: a registered management investment company changing its fiscal year-end. The change creates a transition period running from the day after the prior fiscal year-end through the new fiscal year-end. Because that period is shorter than a full fiscal year, the registrant cannot use:

• N-SAR-A (regular semi-annual interim),
• N-SAR-B (full fiscal-year annual), or
• N-SAR-AT (semi-annual transitional, used when only a semi-annual cycle falls within the transition).

NSAR-BT is filed in lieu of the annual N-SAR-B for that irregular cycle. Absent a fiscal-year-end change, no NSAR-BT exists. Common drivers of the underlying year-end change include complex-wide calendar alignment, tax-year changes, and reorganizations, mergers, or adviser changes.

Timing

Annual N-SAR filings — including NSAR-BT — were due within 60 calendar days after the close of the reporting period. For NSAR-BT, the clock runs from the last day of the transition period (the new fiscal year-end). The 60-day deadline was uniform across management-company variants and did not vary by registrant size or public float.

Amendments: NSAR-BT/A

The "/A" suffix denotes an amendment to a previously filed NSAR-BT, submitted by the same registrant. Amendments are corrective and event-driven, not periodic. Typical causes:

• corrections to financial or operational data items,
• corrections to identified service providers (adviser, sub-adviser, underwriter, administrator, custodian, accountant),
• corrected or supplemented attachments (auditor's internal control report, accountant's consent),
• responses to SEC Division of Investment Management staff comments, or
• recalculations of items such as portfolio turnover, expense ratios, or share sales and redemptions.

Rescission and replacement

The SEC's Investment Company Reporting Modernization rulemaking (Release Nos. 33-10231 / 34-79095 / IC-32314, October 13, 2016) rescinded Form N-SAR in full — every management-company variant (A, B, AT, BT) and their amendments — and replaced it with Form N-CEN (annual structured census report) and Form N-PORT (monthly portfolio holdings).

Rescission took effect June 1, 2018. After that date, no NSAR-BT filings are made; a fiscal-year-end change in a post-June-2018 cycle is reported through Form N-CEN's transition-period mechanics. The dataset closes at this rescission boundary.

How This Dataset Differs From Similar Datasets or Filings

Form NSAR-BT belongs to the N-SAR family of investment company periodic reports filed under Section 30 of the Investment Company Act of 1940. The most informative comparisons are the other N-SAR variants, the post-2018 successor (N-CEN), and a few adjacent fund-disclosure forms that cover the same registrants but different subject matter. NSAR-BT is the transitional annual variant: triggered only by a fiscal-year-end change, it is small, sparsely distributed, and not interchangeable with the regular periodic filings of the same registrants.

Form NSAR-B (regular annual report)

The direct cycle counterpart. Form NSAR-B and NSAR-BT share the same annual item set, the same SGML/positional answer.fil payload, and the same legal basis. The only difference is the reporting period: NSAR-B covers a full twelve-month fiscal year; NSAR-BT covers the shortened transition period created when a registrant changes its fiscal year-end. NSAR-B is the continuous annual series; NSAR-BT is the one-off bridge filed in the year of the change.

Form NSAR-AT (semi-annual transitional)

The closest sibling in trigger logic: both are event-driven by a fiscal-year-end change. They differ on which interval within the transition is being reported and on item depth. Form NSAR-AT carries the lighter semi-annual item set; NSAR-BT carries the full annual item set, including audit-related responses. Tracking fiscal-year changes typically requires both.

Form NSAR-A (regular semi-annual report)

The semi-annual companion to NSAR-B, filed mid-fiscal-year on the regular cycle. Same answer.fil format, but a narrower item set (no audit items, no full-year metrics) and no connection to fiscal-year changes. Not a substitute for NSAR-BT. Form NSAR-A covers this cycle.

Form NSAR-U (Unit Investment Trusts)

The N-SAR variant for UITs rather than management companies. Same SGML/positional layout, but a different filer population and a UIT-specific item set (no investment adviser items, no management-company turnover concept). NSAR-BT is restricted to management companies (open-end and closed-end funds) in transition years, so Form NSAR-U is adjacent rather than overlapping.

Form NSAR-BT/A (amendment variant)

Amendments to previously filed NSAR-BT reports, using the same item set and format. Treat NSAR-BT/A as a versioning layer on top of NSAR-BT: within an accession lineage, amended responses may supersede the original.

Form N-CEN (post-2018 successor)

N-CEN replaced the entire N-SAR family in June 2018 and is the modern functional equivalent for annual census-type items. Two differences dominate:

1. Format. N-CEN is structured XML against a defined schema; NSAR-BT uses the legacy SGML/positional answer.fil payload, which must be parsed against the N-SAR item map.
2. Transition handling. N-CEN has no separate transitional form. A transition period is signaled by internal report-type fields within the standard N-CEN filing, so no post-2018 form code corresponds to "NSAR-BT."

Any analysis crossing the June 2018 boundary must join NSAR-BT and N-CEN as a single conceptual series despite the format mismatch.

Adjacent fund-disclosure forms

Same registrants, different subject matter. Complements, not substitutes.

• Form N-Q (2004–2019). Quarterly position-level portfolio holdings (succeeded by N-PORT). NSAR-BT contains aggregate operational metrics (turnover, sales, redemptions, expense ratios), not security-level positions.
• Form N-PX. Annual proxy voting record, vote-by-vote. Overlaps with NSAR-BT only in filer population.
• Form N-CSR. Certified semi-annual and annual shareholder reports: audited financial statements, MD&A, and Sarbanes-Oxley certifications. Narrative and accounting-focused, where NSAR-BT is structured and operational. A registrant may file both for a transition period; they answer different questions.

Boundary summary

NSAR-BT is uniquely defined by the simultaneous combination of three properties:

1. Annual-level N-SAR item set (full year-end items, including audit-related questions), not a semi-annual subset.
2. Event-triggered by a fiscal-year-end change, not filed on a recurring cycle, which makes the dataset small and sparsely distributed across registrants.
3. Legacy SGML/positional answer.fil format, requiring the N-SAR item map to parse and structurally incompatible with the XML-based N-CEN that replaced it in 2018.

For continuous panels of fund operational data: use NSAR-A and NSAR-B for regular cycle years, NSAR-AT and NSAR-BT (plus their /A amendments) for transition years, and join to N-CEN after June 2018. Within that map, NSAR-BT is specifically the transition-year annual bridge.

Who Uses This Dataset

Because Form NSAR-BT was triggered only by a fiscal-year-end change, the audience is narrow and each role targets specific item-number bands inside answer.fil.

Fund-Industry Historians

Strategy and archive groups inside fund complexes reconstruct the operating history of legacy mutual fund and closed-end series across mergers, sponsor reorganizations, and complex-wide calendar alignments. They rely on Item 072 (advisory fee, 12b-1, expense lines), Item 074 (net assets, capital share activity), and the registrant block to trace fund lineage through sponsor changes.

Academic Finance Researchers

Researchers studying expense ratios, turnover, flows, and share-class economics use NSAR-BT to extend pre-2018 panels and capture odd-period observations that commercial databases drop. Core fields are portfolio turnover, the sales/redemption/exchange matrices broken out by share class, and Item 072 expense components. The transition-period framing also supports event studies on whether fiscal-year-end changes coincide with adviser turnover, board reconstitution, or sponsor M&A.

Fund Administration and Compliance

Administrators, transfer agents, and compliance staff at fund sponsors pull NSAR-BT for archive lookups when defending or reconstructing prior filing positions. Service-provider identification items (principal underwriter, custodian, transfer agent, auditor, adviser), fidelity bond and E&O coverage items, and affiliation disclosures support regulator inquiries that reach into the pre-2018 archive.

Litigation Support and Forensic Accounting

Forensic accountants on Section 36(b) excessive-fee suits and breach-of-fiduciary-duty matters use the advisory fee schedule, breakpoint disclosures, fee waivers/reimbursements, and the relationship between average net assets and reported fee revenue (Item 072 plus the adviser block) to build damages models and comparative fee analyses. The transition window often isolates a short period aligned with the corporate actions named in a complaint.

Fund-Data Vendors

Data engineering teams at fund-data vendors and quantitative research platforms ingest NSAR-BT to bridge standard annual N-SAR and the post-2018 N-CEN regime. Income-statement and balance-sheet items prorate annualized expense ratios, the sales/redemption matrices maintain continuous flow series, and CIK plus series/class identifiers stitch entities across the fiscal-year-end change.

RIA, Sub-Adviser, and Transfer-Agent Due Diligence

Due-diligence teams at advisers and institutional gatekeepers verify historical adviser, sub-adviser, underwriter, and transfer-agent relationships before onboarding or extending a service contract. The adviser, sub-adviser, and underwriter identification blocks plus affiliated-person items document the service-provider chain for Form ADV preparation, RFP responses, and fiduciary files.

SEC Examiners and Policy Economists

Examination staff and policy economists treat the dataset as the authoritative archive of transitional reporting under the rescinded N-SAR regime. It supports examination workpapers tracing a fund's regulatory history, retrospective rulemaking analysis on the frequency of fiscal-year-end changes, and granularity comparisons between pre-2018 N-SAR items and current N-CEN data points.

Asset-Management M&A and Corporate Development

Corporate development teams evaluating acquisitions of legacy fund families, advisers, or fund-administration books pull historical operating metrics across reorganization-driven fiscal-year-end changes. Items 072 and 074 (expense components, turnover, net assets, flows) plus the adviser and underwriter blocks feed valuation models and integration plans that must span the irregular reporting period.

LLM and RAG Engineering Teams

Engineering teams building retrieval-augmented systems and fund-domain assistants index NSAR-BT to extend coverage into the pre-2018 N-SAR vocabulary. Structured item numbers, registrant metadata, and consistent layout let them align transition-period filings with N-CEN and N-PORT records so queries about a fund's pre-2018 operations, service providers, or expense history return grounded answers.

Specific Use Cases

Each use case below ties directly to the transition-period framing of NSAR-BT and to specific item bands in answer.fil.

1. Bridging fund-operating panels across fiscal-year-end changes

Data engineering teams at fund-data vendors splice NSAR-BT records into otherwise broken NSAR-B panels so expense ratios, turnover, and flow series remain continuous through a registrant's calendar shift. The workflow parses Item 072 (advisory fee, 12b-1, transfer-agent, custodian, professional fees), Item 071 (purchases, sales, turnover dollars), and Item 075 (average net assets), demultiplexes the trailing two-digit series suffix on each line, and joins to seriesAndClassesContractsInformation in metadata.json to reattach S- and C-IDs. The output is a per-series, per-class operational time series with the stub period prorated to a comparable annual basis.

2. Reconstructing share-class flow histories

Academic researchers and product analysts rebuild gross sales, redemptions, reinvestments, and exchange-in/exchange-out histories at the share-class grain by extracting Item 028's four-quarter by eight-row matrix (sub-codes A through H). Each row line carries occurrence + series + class digits in the CCCCCC coordinate, so the parser pivots on those positions to produce a class-level flow table. This fills the transition-window gap that commercial flow databases typically drop and supports event studies tying fiscal-year changes to flow disruption.

3. Building damages models for Section 36(b) excessive-fee litigation

Forensic accountants supporting 36(b) and breach-of-fiduciary-duty matters compute realized advisory-fee rates by dividing Item 072 advisory-fee lines by Item 075 average net assets per series, then compare against the Item 048 advisory-fee schedule with breakpoints. The adviser and sub-adviser identification blocks (Items 008-009) and affiliated-person flags fix the contractual counterparty during the transition window. Because NSAR-BT isolates a short, fiscal-year-end-aligned interval, it often pinpoints fee behavior around the exact corporate action named in the complaint.

4. Auditing fidelity bond and E&O coverage histories

Compliance staff and SEC examiners reviewing pre-2018 insurance arrangements pull the fidelity-bond carrier and coverage amount items along with the errors-and-omissions insurance items in the high 07x-08x band, plus the independent public accountant identity items, to verify continuity of coverage across a registrant's fiscal-year change. Cross-referencing the entities block (CIK, file number 811-xxxxx, state of incorporation) with the insurance items produces a defensible workpaper trail for examination requests reaching into the rescinded-form archive.

5. Mapping service-provider chains for due-diligence files

Due-diligence teams at RIAs, gatekeepers, and acquirers pull Items 008-015 (investment adviser, sub-adviser, sub-administrator, principal underwriter, fund accountant, custodian, transfer agent) for each fund series in the trust. Each block resolves to company name, file number, location, and a relationship/affiliation code, giving a snapshot of the full service-provider chain at the moment a sponsor restructured its calendar. The output feeds Form ADV historical disclosures, RFP responses, and pre-acquisition red-flag reviews on legacy fund families.

6. Detecting fiscal-year-end changes correlated with sponsor events

Policy economists and corporate-development analysts use the corpus itself as an event index: every NSAR-BT filing marks a fiscal-year-end change. Joining filedAt, periodOfReport, the fiscalYearEnd field in entities, and the registrant header in Item 000 produces a clean event panel. Researchers then test whether transition filings cluster with adviser turnover (Items 008-009), board reconstitution flags (in the 049-069 governance band), or complex-wide calendar alignments after sponsor M&A.

7. Grounding pre-2018 fund-domain RAG systems

LLM and RAG engineering teams index each accession folder by chunking answer.fil along item-number boundaries and attaching the metadata.json series/class taxonomy as structured metadata. Item 072 lines become expense-question chunks, Item 074 lines become balance-sheet chunks, and the Items 008-015 service-provider blocks become entity-relationship chunks. The result lets a fund-domain assistant answer questions about a specific fund's pre-2018 transition period operations and join those answers to N-CEN and N-PORT records on the post-2018 side of the format break.

Dataset Access

The Form NSAR-BT Files dataset can be accessed in three ways: through a JSON index API that lists all containers and their metadata, as a single full-dataset archive, or as individual monthly container files.

Dataset Index JSON API: https://api.sec-api.io/datasets/form-nsarbt-files.json

Returns dataset-level metadata and the full list of container files. Each container entry includes its size, record count, last updated timestamp, and direct download URL. Use this endpoint to monitor which containers were refreshed in the most recent run and decide which ones to download incrementally. This endpoint does not require an API key.

          1
          {
        

          2
            "datasetId": "1f13365b-9ae0-69ad-af76-b134f0367c83",
        

          3
            "datasetDownloadUrl": "https://api.sec-api.io/datasets/form-nsarbt-files.zip",
        

          4
            "name": "Form NSAR-BT Files Dataset",
        

          5
            "updatedAt": "2026-04-15T12:09:34.065Z",
        

          6
            "earliestSampleDate": "1994-01-01",
        

          7
            "totalRecords": 3202,
        

          8
            "totalSize": 8463487,
        

          9
            "formTypes": ["NSAR-BT", "NSAR-BT/A"],
        

          10
            "containerFormat": "ZIP",
        

          11
            "fileTypes": ["TXT", "JSON", "FIL", "HTML", "PDF"],
        

          12
            "containers": [
        

          13
              {
        

          14
                "downloadUrl": "https://api.sec-api.io/datasets/form-nsarbt-files/2018/2018-05.zip",
        

          15
                "key": "2018/2018-05.zip",
        

          16
                "size": 412875,
        

          17
                "records": 87,
        

          18
                "updatedAt": "2026-04-15T12:09:34.065Z"
        

          19
              }
        

          20
            ]
        

          21
          }
        

Download Entire Dataset: https://api.sec-api.io/datasets/form-nsarbt-files.zip?token=YOUR_API_KEY

Downloads the complete dataset as a single ZIP archive containing all Form NSAR-BT and NSAR-BT/A filings from January 1994 through June 2018. This endpoint requires an API key passed via the ?token= query parameter.

Download Single Container: https://api.sec-api.io/datasets/form-nsarbt-files/2018/2018-05.zip?token=YOUR_API_KEY

Downloads a single monthly container ZIP file instead of the full archive. Each container typically holds one calendar month of filings, which is useful for incremental downloads or targeted historical retrieval. This endpoint requires an API key passed via the ?token= query parameter.

Frequently Asked Questions

What form does this dataset cover?

The dataset covers Form NSAR-BT — the transitional annual report filed by registered management investment companies under Section 30 of the Investment Company Act of 1940 — together with its amendment variant Form NSAR-BT/A. NSAR-BT is the transitional-period sibling of the regular annual NSAR-B and shares NSAR-B's full annual item set.

What does one record in this dataset represent?

One record is one accepted EDGAR submission of Form NSAR-BT or NSAR-BT/A, identified by its EDGAR accession number. Each record is materialized as a folder containing a metadata.json side-car plus the original EDGAR documents — primarily the legacy SGML-wrapped answer.fil carrying the questionnaire response, plus any TXT, HTML, or PDF exhibits that accompanied the submission.

Who is required to file Form NSAR-BT?

Registered management investment companies — open-end mutual funds and closed-end funds (including listed, unlisted, and interval closed-end funds) — file NSAR-BT when they change their fiscal year-end. Unit investment trusts, face-amount certificate companies, business development companies, and private funds do not use the form.

What triggers a Form NSAR-BT filing?

A single trigger: a registered management investment company changing its fiscal year-end. The change creates a transition period shorter than a full fiscal year, and NSAR-BT is filed in lieu of the regular NSAR-B annual report for that irregular cycle. The filing was due within 60 calendar days after the close of the transition period.

What time period does the dataset cover?

The corpus runs from the EDGAR opening in 1994 through the June 1, 2018 rescission date. The SEC rescinded the entire N-SAR family — including NSAR-BT — under Investment Company Reporting Modernization Releases 33-10231, 34-79095, and IC-32314 (October 13, 2016), with rescission taking effect June 1, 2018. No NSAR-BT filings are made after that date; transition periods in post-June-2018 cycles are reported through Form N-CEN's transition-period mechanics.

How is NSAR-BT different from NSAR-B and N-CEN?

NSAR-B is the regular twelve-month annual report; NSAR-BT carries the same item set but covers the irregular stub period created by a fiscal-year-end change. N-CEN is the post-June-2018 successor to the entire N-SAR family and is structured XML against a defined schema, whereas NSAR-BT uses the legacy SGML/positional answer.fil payload that must be parsed against the N-SAR item map. N-CEN has no separate transitional form — transition periods are signaled by internal report-type fields within the standard N-CEN filing.

What file formats are inside the dataset?

The dataset is distributed as monthly ZIP containers. Inside each accession folder the file types are TXT, JSON, FIL, HTML, and PDF. JSON is always metadata.json; FIL is always the N-SAR answer.fil answer document; TXT, HTML, and PDF appear when the original submission included exhibits beyond the bare answer file. Image attachments from the original submissions are excluded, and there are no XBRL instance documents because N-SAR was never under the XBRL mandate.