Private Offering Notice Feed

Incremental feed of new Regulation D offering notices (Form D; add Form D/A amendments with formTypes). Each run downloads the filings itself from EDGAR and writes them to your Apify dataset.

1. What it does

The Actor reads the SEC daily form index for each business day since your cursor, keeps the Form D rows (and Form D/A rows if formTypes includes D/A), downloads each new primary XML, and pushes the filings that match your filters.

What you get

• One row per Form D accession, or per Form D/A too if you add D/A to formTypes, with issuer name, CIK, state, industry group, exemption codes, offering amount, amount sold, date of first sale, and links to the filing and its XML.
• Each matching filing is delivered once per state key. The delivered list keeps only the newest quarter's filings; filings from an older quarter, for example in a backfill, are not remembered.

It remembers what it has already delivered. Each charged filing is written to state before the run continues, and that state is saved again if the run is blocked, fails, or the platform migrates or aborts it. Once a week it can report filings that disappeared from the current quarter's index.

2. Source and attribution

The data comes from EDGAR on sec.gov. The SEC's privacy page says: "Information presented on sec.gov is considered public information and may be copied or further distributed by users of the web site without the SEC's permission. Please consider appropriate citation to the SEC as the source."

Every filing and every removal tombstone includes:

Source: U.S. Securities and Exchange Commission, EDGAR (sec.gov).

The same page says: "'SEC,' the EDGAR logo, and the names EDGAR, EDGARLink, and EDGARLink Online are the SEC's registered trademarks. You may not use them in a trade name, trademark, or domain name of an SEC- or EDGAR-related business without a license from the SEC. You may refer in text to the existence of EDGAR and the EDGAR system without a license, so long as you are not creating the impression that your business is affiliated with or approved by the SEC."

Ample Labs is not affiliated with, endorsed by, or sponsored by the U.S. Securities and Exchange Commission. "SEC" and "EDGAR" are used only to describe the data source.

3. Quick start

Run it once a day, at an off-minute in the morning US Eastern time, for example 06:17. The daily indexes are updated at night, starting around 22:00 America/New_York, and EDGAR disseminates filings Monday through Friday.

Leave the dates empty. The first run reads the latest completed business day. If that day's index is not published yet, the first run steps back up to 5 business days and reads the newest day that has one. Later runs continue from the cursor saved for stateKey.

Give each schedule, and each different filter setup, its own stateKey. The default is default. Dedupe is stored per key, so two setups that share a key can hide filings from each other. Use resetState once if you want that key to start over, then turn it off.

Use one schedule and do not start a second run while one is still going. See section 8.

4. Inputs and exemption codes

Input	Default	What it does
dateFrom	empty	Start of a backfill. Leave empty to continue from the cursor. One run scans at most maxBackfillDays business days (default 1, maximum 30). The same dates resume after the last finished day. Clear the dates when the backfill is done.
dateTo	empty	Last day to scan. Empty means the latest business day whose index should already be published. A span longer than maxBackfillDays finishes on later runs with the same dates.
formTypes	D	D is a new notice. D/A is an amendment.
industryGroups	empty	EDGAR industry group names, such as Biotechnology or Pooled Investment Fund. Empty keeps all.
excludePooledFunds	false	Drop industry group Pooled Investment Fund.
states	empty	Issuer state or country codes, such as CA or NY.
minAmountSold	empty	Minimum total amount sold, in dollars.
minTotalOffering	empty	Minimum total offering amount, in dollars. Indefinite passes.
exemptions	empty	Keep filings that use at least one of these codes.
firstSaleOnly	false	Keep only filings with a date of first sale.
includeRelatedPersonNames	false	Add related-person names, and sales-compensation recipient names and CRD numbers. Those may identify a person. The associated broker-dealer name and CRD are the firm's record and are included either way.
includeStreetAddress	false	Add the issuer street and phone. Personal data.
maxResults	100 in the Console, 1000 if omitted	The Console form starts at 100. A call that leaves the field out still stops after 1000 pushed filings. The unfinished day resumes next run.
maxBackfillDays	1	Business days to scan this run. Default 1. Maximum 30. A longer range, or a cursor more than this many business days behind, is left for a later run. The run warns and does not scan the waiting days in one run. The status says how many days were scanned.
stateKey	default	Names the saved cursor. Use one key per schedule or filter setup.
resetState	false	Drop the saved cursor and delivered list for this key and start fresh.

There is no input for request rate. The cap is fixed in the code.

Federal exemption codes are the item values in Form D XML technical specification §4.22:

Exemption	Code
Rule 504(b)(1) (not (i), (ii), or (iii))	04
Rule 504(b)(1)(i)	04.1
Rule 504(b)(1)(ii)	04.2
Rule 504(b)(1)(iii)	04.3
Rule 506(b)	06b
Rule 506(c)	06c
Securities Act §4(a)(5)	4a5
Investment Company Act §3(c)	3C
§3(c)(1), (2), (4), (5), (6), (7), (9), (10), (11), (12), (13), (14)	3C.1, 3C.2, 3C.4, 3C.5, 3C.6, 3C.7, 3C.9, 3C.10, 3C.11, 3C.12, 3C.13, 3C.14

Industry group names are the EDGAR industryGroupType values, including Agriculture, Commercial Banking, Insurance, Investing, Investment Banking, Pooled Investment Fund, Other Banking and Financial Services, Business Services, Coal Mining, Electric Utilities, Energy Conservation, Environmental Services, Oil and Gas, Other Energy, Biotechnology, Health Insurance, Hospitals and Physicians, Pharmaceuticals, Other Health Care, Manufacturing, Commercial, Construction, REITS and Finance, Residential, Other Real Estate, Retailing, Restaurants, Computers, Telecommunications, Other Technology, Airlines and Airports, Lodging and Conventions, Tourism and Travel Services, Other Travel, and Other.

5. Output dictionary

The Output tab lists every field. Personal-data fields stay off unless you turn them on.

Field	What it is
accessionNumber	EDGAR accession number. This is the dedupe key.
issuerName	Issuer name as filed.
formType	D or D/A.
filedAt	Filing date, YYYY-MM-DD.
totalOfferingAmount	Dollars, or null when the offering is Indefinite (then isIndefinite is true).
isIndefinite	True when the filing gives the total offering amount as Indefinite.
federalExemptions	Exemption item codes.
issuerState	Issuer state or country code.
eventType	new, backfill, removed, or no-new-filings.
daysScanned	On the status row. Business days fully scanned, possibly empty.
cursorDate	On the status row. Cursor after the run, or null.
stopReason	On the status row. completed, spending-limit, sec-blocked, index-not-ready, or time-margin.
runAt	On the status row. Time the row was written.

From 0.2.0, totalOfferingAmount and totalRemaining are a number or null. Indefinite offerings have isIndefinite true. Previously these fields could be the string Indefinite. Rows delivered before 0.2.0 keep the old shape.

A removal tombstone is free. It has the accession number, removed, the detection time, and the SEC attribution line. It has no personal-data fields. Filter it by keeping rows whose eventType is new or backfill.

When a run delivers no filings and no removal tombstones, it pushes one status row after state is saved. eventType is no-new-filings. The row is free. It is not a filing, it is not added to delivered, and it does not move the cursor. Filter it the same way you filter a removed tombstone, on eventType. A run that fails does not push this row.

Names, street, and phone are off by default. Broker-dealer firm name and CRD are included when the filing names a firm, and omitted when it says None. Recipient names and CRD numbers stay off unless includeRelatedPersonNames is on.

6. Incremental delivery and tombstones

State is stored in a named key-value store in your Apify account, private-offering-notice-feed-state-<stateKey>, under the record FEED_STATE. A new run does not get a new copy of that store. The run's own default key-value store is empty every time, so the cursor is not kept there.

• cursorDate is the last business day that was fully scanned.
• delivered is the accession numbers delivered in the current quarter for this stateKey. It is pruned when the quarter changes.
• lastRemovalCheck is the last time a removal check finished.

An accession already in delivered is not fetched or charged again. delivered is saved after every charged filing, and again if the run is blocked, fails, migrates, or aborts. The cursor moves forward only after a day is fully scanned. If the spending limit, the run timeout, or maxResults stops the run in the middle of a day, that day stays unfinished. Filings already charged stay in delivered.

A backfill stores its dateFrom, its dateTo, and the last day of that range that was fully scanned. The next run with those same dates starts the day after that finished day, so the run does not charge day-scanned again for a day it already finished. A different dateFrom or dateTo starts the new range from its first day. Rescanning days that overlap a range you already scanned charges day-scanned again. Filings already delivered from the newest quarter are not charged again; filings from older quarters are delivered and charged again. Clear the dates when you want the cursor to follow new days.

If less than 30 seconds remain before the platform run timeout (ACTOR_TIMEOUT_AT), the run saves its place and stops. Run the same input again to continue. Finished backfill days are not scanned or charged again. The platform run timeout stays 1 hour. A run that stops for that timeout resumes on the next run.

The removal check runs on the first run after Saturday 12:00 America/New_York, and at most once a week. It downloads full-index/{year}/QTRn/form.idx for the current quarter only and streams it. Accessions this stateKey has delivered, dated before that Saturday, and missing from the rebuilt index, become tombstones and are dropped from delivered.

If that download fails, the run still finishes. removal-check is not charged, and lastRemovalCheck stays where it was. The next try is on a later America/New_York calendar day, once that day, until the check succeeds. A second run on the same day does not download the index again.

Removals are detected only for filings delivered in the current calendar quarter. A filing delivered in one quarter and removed after the next quarter begins is not detected. Filings dated on or after that Saturday are left for the next weekly rebuild, because the Saturday index does not contain them yet. The SEC does not support partial downloads of the index, so older quarters are not diffed.

A 404 for a past business day is treated as a holiday with no filings. A 404 for the newest day leaves the cursor where it is, because that index may not be published yet.

On a first run, with no saved cursor and no dateFrom, a missing index is handled differently. The run steps back one business day at a time, up to 5 business days, until it finds a day that has an index. It scans that day, sets the cursor there, and logs the days it skipped. Those skipped days are not charged day-scanned. A cursor run or a date range does not step back. If none of the 5 days has an index, the run pushes the status row with stopReason index-not-ready and leaves the cursor empty.

7. Pricing

• $0.005 to start a run (apify-actor-start) (one start event per GB of memory; this Actor always uses one)
• $0.003 for each new filing pushed (new-filing)
• $0.005 for each business day scanned after the run's first day (day-scanned)
• $0.001 for the weekly removal check (removal-check), charged only when the check downloads the index

Each run needs a maximum cost of at least $0.01, enough for the start and one filing.

Tombstones are free. The no-new-filings status row is free. Saving a filing to the dataset does not add a separate charge. The first day in a run is not a day-scanned event. A three-day backfill therefore charges day-scanned twice. Days skipped while a first run looks for a published index are not charged.

Before each extra day, and before each filing download, the run checks how many of that event it can still charge. On a day that will be charged day-scanned, the filing check also requires that the filing price and the pending day price both fit in the remaining budget. If they do not, it stops before that work. The cursor stays on the last fully scanned day. Nothing that was pushed is left uncharged: the filing is pushed, then charged, then saved. The day charge happens after the cursor is saved. A later run does not fetch or charge an accession that was already saved.

Default memory is 256 MB. Maximum memory is 512 MB.

8. Fair-access design

The Actor calls sec.gov directly. It does not use a proxy and it does not rotate IP addresses.

• At most one request per second, and one request in flight.
• No input can raise that rate.
• Every request sends User-Agent: Ample Labs support@haveample.com and Accept-Encoding: gzip, deflate.
• On a 403 or 429, the run stops, waits at least 10 minutes, and retries once at 0.5 requests per second. A second 403 or 429 ends the run. The cursor is not advanced for the in-progress day. The filing that failed is not pushed. Filings already charged are saved. A warning is written to the log and the status message.
• On a 5xx response or a timeout, the run waits 2 seconds, then 4 seconds, and stops after 3 tries. Filings already charged are saved before the run fails.
• Each run on the Apify platform waits a random 0 to 30 seconds before the first SEC request, so simultaneous runs do not all start together. Local runs skip that wait. Buyers cannot turn the wait off.
• Before the first SEC request, the run asks https://api.ipify.org for the egress IP and logs it once. If that lookup fails, the run continues.

Do not run more than one copy of this Actor at a time on an account. Apify does not give this Actor a way to see your other runs, so the one-run limit is a rule for how you schedule it, not a lock inside the run. Ten overlapping runs would be enough to reach the SEC's 10 requests per second guidance.

9. Privacy

The source is SEC EDGAR Form D. The filing can include the issuer's address and phone, and the names, roles, and addresses of executive officers, directors, and promoters.

Ample Labs does not store the output. The run executes in your Apify account. The dataset and the key-value store belong to you.

If "Share run data with developers" is on, your runs of public Actors are visible to the Actor's developer. That includes logs, input, and output storages (dataset, key-value store, and request queue). A run linked in an Actor issue is shared with the developer as well. The setting is described in Apify's general resource access page.

You decide whether to turn on names, street, and phone. They are off by default. Only issuer-level fields are on by default, and the issuer street and phone stay off with them. recipientName and recipientCRDNumber may identify a person and stay off unless includeRelatedPersonNames is on. associatedBDName and associatedBDCRDNumber are the broker-dealer firm's record, not personal data, and are included by default. The solicitation states and the count of recipients stay on.

If you process personal data of people in the European Economic Area, you are the controller. GDPR Article 14 requires a controller, in many cases, to tell people when their data did not come from them. Article 14(5)(b) says that duty does not apply where providing the information would be impossible or would involve a disproportionate effort. The official text is at the EU Publications Office copy of Regulation (EU) 2016/679. Whether that exception fits your use is your decision. This section is not legal advice.

10. Lawful use

Do not use this feed for unsolicited mass messaging. The Apify Acceptable Use Policy prohibits that.

If you email anyone using information from this feed, the CAN-SPAM Act, 15 U.S.C. 7704(a)(5), requires a clear identification that the message is an advertisement or solicitation, a working opt-out, and a valid physical postal address of the sender, unless you already have their prior affirmative consent for the identification requirement.

Form D data is taken from federal government records. The California Consumer Privacy Act, Civil Code §1798.140, excludes publicly available information, including information lawfully made available from federal, state, or local government records. That exclusion does not by itself make every later use lawful.

This is not legal advice.

FAQ

Do I pay again for a filing I already received? Not under the same state key, if the filing is from the quarter the delivered list keeps. The list keeps one quarter: it moves to a newer quarter when a newer filing is delivered, or when the weekly removal check runs in a new calendar quarter. Filings from an older quarter are not remembered and are charged again if delivered again. A Form D/A is a separate filing. It is delivered only if formTypes includes D/A, and it is charged as a new filing.

Why did I get a row with eventType no-new-filings? The run found nothing new. It is free. Filter on eventType new or backfill to keep filings only.

Can I get older filings? Yes. Set Date from and Date to. Each extra day in a run costs $0.005.

Are names and addresses included? Not by default. Turn on includeRelatedPersonNames or includeStreetAddress. You are responsible for lawful use; see Privacy and Lawful use.

Can I run several schedules? Yes, one at a time, each with its own stateKey.

11. Support

Questions and takedown requests: support@haveample.com.

12. Changelog

0.2.0

• From 0.2.0, totalOfferingAmount and totalRemaining are a number or null. Indefinite offerings have isIndefinite true. Previously these fields could be the string Indefinite. Rows delivered before 0.2.0 keep the old shape.

0.1.8

• A run that delivers nothing pushes one free no-new-filings row after state is saved. Filter it on eventType, the same way you filter removed.
• A first run with no cursor and no dates steps back up to 5 business days when the latest index is missing. Skipped days are not charged.
• The Console prefill for maxResults is 100. A call that omits the field still uses 1000.

0.1.7

• Add the dataset schema and the output schema. The Output tab lists every field.
• A run needs a maximum cost of at least $0.01, enough for the start and one filing.

0.1.6

• Omit a broker-dealer name or CRD, and a recipient name or CRD, when the filing says None.
• The random start delay is at most 30,000 milliseconds. A draw of 0.999999 stays within that cap.