Signal8

 ⟨191 unchanged words⟩  cap, industry, listing exchange (NASDAQ/NYSE/AMEX), listing market (market="otc" / "exchange" — the reliable OTC test; the exchange label is inconsistent on OTC rows), float data source and, TTM dividend yield.yield and snapshot cash (minCash / maxCash). CASH RUNWAY: a null cash_runway_months carries cashRunwayReason saying why. "estimate_would_be_negative" means the last cash figure rolled forward by the burn rate is below zero today — ESTIMATED exhausted, not a reported zero balance (cashRunwayAsOf is the cash figure's date). Such rows have no months, so maxCashRunway drops them unless includeRunwayExhausted=true. Any other reason, or a null reason, is NOT a runway claim. CASH: estimatedCurrentCash (with estimatedCashAsOf) is the dilution snapshot's cash PROJECTION in USD, present only for snapshot-covered companies — null means NOT MEASURED, never "no cash", and minCash / maxCash can only return covered companies. Each row also carries listingMarket ("exchange" / "otc", null = unknown). Each row carries dividendYield (trailing-twelve-month cash dividends  ⟨62 unchanged words⟩ 

 ⟨39 unchanged words⟩  (exact match): NASDAQ, NYSE, or AMEXAMEX. For OTC use market=\"otc\" instead","type":"string"},"excludeHalted":{" ⟨39 unchanged words⟩ ,"external"],"type":"string"},"includeRunwayExhausted":{"description":"When true, maxCashRunway also returns companies whose estimated cash is already exhausted (cash_runway_months null, cashRunwayReason \"estimate_would_be_negative\"); they sort first on an ascending runway sort. Default false","type":"boolean"},"industry":{"description":"Filter by company industry ⟨21 unchanged words⟩ "minimum":1,"type":"number"},"market":{"description":"Listing market: \"otc\" (OTC Markets) or \"exchange\" (NASDAQ / NYSE / NYSE American / Arca / Cboe). Companies with no listing record match neither","enum":["exchange","otc"],"type":"string"},"maxCash":{"description":"Maximum dilution-snapshot cash projection in USD (estimatedCurrentCash). Snapshot-covered companies only — an uncovered company is not measured, not excluded for having cash","minimum":0,"type":"number"},"maxCashRunway":{"description":"Maximum estimated months of ⟨68 unchanged words⟩  trading volume","type":"number"},"minCash":{"description":"Minimum dilution-snapshot cash projection in USD (estimatedCurrentCash). Snapshot-covered companies only","minimum":0,"type":"number"},"minCashRunway":{"description":"Minimum estimated months of ⟨105 unchanged words⟩ ","computed_float","market_cap_computed","updated_at","dividend_yield","estimated_current_cash"],"type":"string"},"sortOrder": ⟨12 unchanged words⟩ 

 ⟨41 unchanged words⟩  catalysts. Runway is estimated from current burn rate. Companies whose ESTIMATED cash is already exhausted are included by default and listed first: they carry cash_runway_months null with cashRunwayReason "estimate_would_be_negative" (the last cash figure, dated cashRunwayAsOf, rolled forward by the burn rate is below zero today). That is an estimate from a dated balance, never a reported zero balance — say "estimated exhausted as of today, last cash figure <cashRunwayAsOf>", not "has no cash". Rows also carry estimatedCurrentCash / estimatedCashAsOf (dilution-snapshot cash projection; null = not measured) and listingMarket.

{"$schema":"http://json-schema.org/draft-07/schema#","additionalProperties":false,"properties":{"includeExhausted":{"description":"Include companies whose estimated cash is already exhausted (default: true)","type":"boolean"},"industry":{"description":"Filter by company industry ⟨20 unchanged words⟩ ":100,"minimum":1,"type":"number"},"market":{"description":"Listing market: \"otc\" (OTC Markets) or \"exchange\" (NASDAQ / NYSE / NYSE American / Arca / Cboe)","enum":["exchange","otc"],"type":"string"},"maxMonths":{"description":"Maximum months ⟨27 unchanged words⟩ 

 ⟨377 unchanged words⟩  snapshot carries ANY figure — "floatEstimate" (~bound "point" / <= "ceiling" / >= "floor"), "tradeableFloatCeiling", or an "affiliateBasis" range — "floatShares" stays null and THAT figure is the answeranswer. 🔴 RENDERING RULE (2026-10-18): print NO "~", with"<=" itsor label;">=" glyph. A "point" (or a "floatEstimate" carrying "statedAsOf" {asOf, source, events[]}, the issuer's own dated count) is printed as the PLAIN number followed by "as of <asOf>" and the caveat (the unsized later issuances in "statedAsOf.events"); a "ceiling" / "floor" / range is NOT a figure — write "float not stated (the filings put it at most / at least N; <reason>)" and never the bare number. A lower tier that also answered and passed the  ⟨704 unchanged words⟩  | "shares_outstanding_ceiling" | "rolled_forward_lower_bound" | "recital_rolled_forward_estimate" | "successor_exchange_estimate" | "affiliate_sizing_estimate" | "restricted_block_ceiling" — at most O/S less affiliates, less the restricted blocks the filings place outside them and the part of the unplaced blocks too large to fit inside them), "bound" ("point" = about X, render "~X"; "ceiling" = AT MOST X, render "<= X"; "floor" = AT LEAST X, render ">= X") and "asOf"; a floor or ceiling is a BOUND, never a point value. null means norecital_rolled_forward_estimat

 ⟨40 unchanged words⟩  "floatEstimate" {shares, bound: "point" ~ | "ceiling" <= | "floor" >=, basis}, the. snapshot'sRENDERING own(2026-10-18): labelleda figure:"point" presentis itprinted as the plain number with its markeras-of and caveat; a "ceiling" / "floor" is a BOUND, not a figure — say "float not stated (at most / at least N)" and never asthe abare pointnumber, never ina "~" / "<=" / ">=" glyph, never arithmetic; neither present = not measured, never 0. The low-float board admits on that figure's VALUE: a <=an 8M ceiling is in, a >= 26M floor or a ~27M27M estimate is out), short interest, dilution, ⟨268 unchanged words⟩ 

 ⟨36 unchanged words⟩  (shares) + "floatEstimateBound" ("point" ~ / "ceiling" <= / "floor" >=) + "floatEstimateBasis" (the working). —RENDERING present(2026-10-18): ita as"point" is the plain number with its markerbasis as the caveat; a "ceiling" / "floor" is a BOUND, e.g.not a figure — write "<=float not stated (at most / at least 7,953,957)"., never the bare number and never a "~" / "<=" / ">=" glyph. minFloat / maxFloat and the computed_float sort compare ⟨165 unchanged words⟩ 

 ⟨320 unchanged words⟩  names that tier (computed / sec-10k / polygona /market-data fmpvendor tier), "floatShares" is ITS figure under ITS ⟨174 unchanged words⟩  "source" with "snapshotFloatWithheldReason" as context (GELS:e.g. Polygona vendor figure of 5.12M, not "<= 11.18M" on 11.18M shares ⟨668 unchanged words⟩ BOUND, never a point value. null means no figure could be deriv

 ⟨85 unchanged words⟩ delisted company. Each row carries "source" ("polygon" | "computed" | "sec_10k" |or "fmp"a market-data vendor tier) because float quality varies by provider —  ⟨15 unchanged words⟩ 

 ⟨566 unchanged words⟩  within the last day (the source is ana FMPvendor screener that lags a rename by ~a day ⟨403 unchanged words⟩ 

 ⟨129 unchanged words⟩  `sharesOutstandingSource` — quote the date with the count. When no MEASURED count exists, `sharesOutstanding` and `marketCap` are a BEST ESTIMATE (the newest dated filed or reference count, moved onto today's split basis): `sharesOutstandingEstimated: true`, `sharesOutstandingEstimateKind` ("at_least" = the filings record later issuances of unstated size, so the true count is likely higher) and `marketCapEstimated: true`. Say it is an estimate and give its date.

Cheap pre-check (1 credit) for whether Signal8 holds a dilution analysis for  ⟨333 unchanged words⟩ 

Get the historical shares-outstanding and cash series (10 credits) from SEC XBRL filings, with offering and material-cash-event ⟨321 unchanged words⟩ 

Get ONE instrument family's rows (10 credits) with per-field SEC citations — each figure carries  ⟨688 unchanged words⟩ 

Get post-offering price performance (10 credits): for each recorded offering, how the stock ⟨284 unchanged words⟩ 

Get the dilution pressure score (5 credits): a 0-100 score with its seven-component breakdown  ⟨507 unchanged words⟩ 

Get the dilution snapshot header (25 credits — the most expensiveheaviest tooldilution herecall): summary totals (potential new shares and  ⟨37 unchanged words⟩  those, and call get_dilution_coverage first to avoid payinga 25wasted creditscall foron an uncovered ticker. 🔴 FLOAT: a declined  ⟨329 unchanged words⟩ 

 ⟨98 unchanged words⟩ treat a low numerator as a low-risk finding. When `scoreDenominatorThin` is true the score rests on fewer than 70 measured points — report it as a PARTIAL reading ("N of M measured"), never as a full one. A non-null `scoreWithheldReason` (e.g. `unmeasured_score_components`, `unquantified_live_instruments`, `no_dilution_snapshot`) means there is NO score — not ⟨117 unchanged words⟩  that as the company being unable to raise. `overheadSupply` bands in-the-money supply PLUS `details.registeredResaleOverhangShares` (/`…Pct`): already-issued affiliate-held shares outside the tradeable float on an EFFECTIVE resale registration — existing shares that can be sold today, never new shares, so never add them to `totalDilutiveShares`; null means not measured. ALWAYS RETURNS 200 — read `available` and `reason` ⟨201 unchanged words⟩ 

 ⟨987 unchanged words⟩ so state the ends that exist as inequalities. They bound the NON-AFFILIATE float, which still INCLUDES restricted shares held by non-affiliates — NOT the tradeable floatShares. This object also appears on PUBLISHED floatsfloats; withthere bothit endsis equalcontext, tonever a bound on floatShares —(a "convention_conservative" tradeable figure can sit BELOW floatLowerBound because it isalso adeducts the restricted shares the bound onlyincludes when— floatSharesthat is null.not a contradiction). (3) "floatEstimate" is a BADGED ESTIMATE  ⟨60 unchanged words⟩  | "rolled_forward_lower_bound" | "recital_rolled_forward_estimate" | "successor_exchange_estimate" | "affiliate_sizing_estimate" | "restricted_block_ceiling" — at most O/S ⟨61 unchanged words⟩ point value. null means no figure could be derived from the snapshot's structured fields. An estimate never becomes floatShares — present shares WITH basis and direction, never alone. Unrelated to the boolean "floatEstimated", which flags a published floatShares back-derived from a USD public-float disclosure. (4) "vendorFallback" is a third-party frderiv

 ⟨619 unchanged words⟩  SCHEDULE 13D/G are dropped and listed in `companyFilter.droppedForms` — unless `companyFilter.subjectMatch` is true, in which case they ARE searched, matched by the company they are about — and with no formTypes the default is 10-K ⟨130 unchanged words⟩ 

 ⟨353 unchanged words⟩ },"industries":{"description":"COMPANY FILTER. Comma-separated industryIndustry names separated by \"|\", exact and case-sensitive (e.g. \"Biotechnology\Biotechnology|Airlines, Airports & Air Services\" — some names contain \", \", so prefer \"|\"; a bare comma with no space after it also separates). An unknown name returns a 400 listing ⟨530 unchanged words⟩ },"sectors":{"description":"COMPANY FILTER. Comma-separated sectorSector names separated by \"|\" (or a bare comma), exact and case-sensitive (e.g. \"Healthcare,Technology\Healthcare|Technology\"). An unknown name returns a 400  ⟨69 unchanged words⟩ 

 ⟨149 unchanged words⟩  Coverage: Signal8's archive of SEC EDGAR filings in the archive's held range (see `archiveFrom` in each response; currently from Octoberlate 2021, expanding back to today2016 (aboutas 3.9Mthe filingshistory backfill fills in), updated through the day;day. The archive only holds filings olderfrom thanits 2021-10`archiveFrom` maydate; bebefore absentthat it holds only a scatter (well under 1% of most years), so anABSENCE emptyBEFORE resultarchiveFrom isIS notNOT proofEVIDENCE somethinga filing was never filed.filed — check SEC EDGAR for earlier periods. Some rows are record-only (bodyStored: false — ⟨22 unchanged words⟩ 

 ⟨243 unchanged words⟩  Coverage: Signal8's archive of SEC EDGAR filings in the archive's held range (see `archiveFrom` in each response; currently from Octoberlate 2021, expanding back to today2016 (aboutas 3.9Mthe filingshistory backfill fills in), updated through the day;day. The archive only holds filings olderfrom thanits 2021-10`archiveFrom` maydate; bebefore absentthat it holds only a scatter (well under 1% of most years), so anABSENCE emptyBEFORE resultarchiveFrom isIS notNOT proofEVIDENCE somethinga filing was never filed.filed — check SEC EDGAR for earlier periods. Some rows are record-only (bodyStored: false — ⟨22 unchanged words⟩ 

 ⟨590 unchanged words⟩  figure, and never present it as floatShares. A "shares_outstanding_ceiling" whose "reason" states a declared affiliate floor ("the insider holdings the filings size total at least N shares") may likewise be served at the share count, but ONLY as the last resort when no lower tier answered or its answer was refused (2026-10-11). Since 2026-10-01 that fall-through is GUARDED: a lower-tier ⟨289 unchanged words⟩  "rolled_forward_lower_bound" | "recital_rolled_forward_estimate" | "affiliate_sizing_estimate" | "restricted_block_ceiling" — at most O/S less affiliates, less the restricted blocks the filings place outside them and the part of the unplaced blocks too large to fit inside them), "bound" ("point" = about  ⟨200 unchanged words⟩ 

 ⟨103 unchanged words⟩ , short interest, dilution, and news/catalyst flags. Listing-compliance data (row "compliance", the compliance/split badges, "reverseSplitSignal", "deficiencySignal") is Signal8 Pro: for any other caller "compliance" is null with "complianceWithheld": "pro_required" — WITHHELD, which is not "not measured" and never "no deficiency". Off-hours it falls back to the last session.  ⟨223 unchanged words⟩ 

 ⟨167 unchanged words⟩  Coverage: Signal8's archive of SEC EDGAR filings in the archive's held range (see `archiveFrom` in each response; currently from Octoberlate 2021, expanding back to today2016 (aboutas 3.9Mthe filingshistory backfill fills in), updated through the day;day. The archive only holds filings olderfrom thanits 2021-10`archiveFrom` maydate; bebefore absentthat it holds only a scatter (well under 1% of most years), so anABSENCE emptyBEFORE resultarchiveFrom isIS notNOT proofEVIDENCE somethinga filing was never filed.filed — check SEC EDGAR for earlier periods. Some rows are record-only (bodyStored: false — ⟨22 unchanged words⟩ 

 ⟨231 unchanged words⟩  Coverage: Signal8's archive of SEC EDGAR filings in the archive's held range (see `archiveFrom` in each response; currently from Octoberlate 2021, expanding back to today2016 (aboutas 3.9Mthe filingshistory backfill fills in), updated through the day;day. The archive only holds filings olderfrom thanits 2021-10`archiveFrom` maydate; bebefore absentthat it holds only a scatter (well under 1% of most years), so anABSENCE emptyBEFORE resultarchiveFrom isIS notNOT proofEVIDENCE somethinga filing was never filed.filed — check SEC EDGAR for earlier periods. Some rows are record-only (bodyStored: false — ⟨22 unchanged words⟩ 

 ⟨105 unchanged words⟩  Coverage: Signal8's archive of SEC EDGAR filings in the archive's held range (see `archiveFrom` in each response; currently from Octoberlate 2021, expanding back to today2016 (aboutas 3.9Mthe filingshistory backfill fills in), updated through the day;day. The archive only holds filings olderfrom thanits 2021-10`archiveFrom` maydate; bebefore absentthat it holds only a scatter (well under 1% of most years), so anABSENCE emptyBEFORE resultarchiveFrom isIS notNOT proofEVIDENCE somethinga filing was never filed.filed — check SEC EDGAR for earlier periods. Some rows are record-only (bodyStored: false —  ⟨294 unchanged words⟩  has not confirmed full coverage of the range. Every response carries `archiveFrom` and `rangeStartsBeforeArchive`: a range with no dateFrom, or starting before archiveFrom, is NEVER definitive (`searchIndex.complete: false`, `reason: "outside_held_range"`, `heldFrom`) and `notes` says so — report "the archive only holds filings from <archiveFrom>; absence before that is not evidence", never "no such filing exists". Pass a dateFrom on or after archiveFrom for a definitive empty answer. The `ciks` filter matches the filing's own CIK ⟨206 unchanged words⟩ 

 ⟨85 unchanged words⟩ "Start filing date (YYYY-MM-DD). The archive startsonly 2021-10.holds filings from its `archiveFrom` date (stated in every response; currently late 2021, expanding back to 2016): omitting dateFrom, or starting earlier, makes an empty result non-definitive.","type":"string"},"dateTo":{ ⟨871 unchanged words⟩ 

 ⟨107 unchanged words⟩  because the counted filings exceed the shares outstanding. 🔴 A bucket (`institutional` / `insider` / `beneficial`) whose shares exceed the shares outstanding by more than a 2% timing tolerance has its `percentOfOutstanding` WITHHELD — null, with `percentWithheldReason: "ownership_exceeds_outstanding"` — and its `totalShares` still published: the percent would not be a measurement (filings on another share basis such as ordinary shares against an ADS count, another class reported under this one, shares lent and reported by both lender and buyer, or filings dated before a buyback or redemption). Report the share figure and the reason; NEVER present such a bucket as 100%, as 0%, or as any percent. When any bucket is withheld, `retail.reason` is `ownership_exceeds_outstanding` and `filedSumPercent` is absent. A null percent with NO `percentWithheldReason` means no share count was available — not measured for a different reason. `sharesOutstanding` is the denominator of every percentage and is DATED: `sharesOutstandingAsOf` + `sharesOutstandingAsOfKind` ("stated" = the issuer's statement date; "checked" = when the figure was last confirmed against a vendor; "none" = date not recorded) and `sharesOutstandingSource` (a vendor-free label). Never present a count without its date. `beneficial.staleShares` / `stalePercentOfOutstanding` / `staleHolderCount` is a MEASURED SUBSET of the counted 5%+ total resting on 13D/G schedules >= 24 months old with no newer reaffirming filing by the same holder family — not a deduction; the total still includes it. A row's `reaffirmedBy` names the newer family filing; `staleFiling.ageMonths` marks a counted row with none. `stalePercentOfOutstanding: null` is not measurable (no share count), never zero.

 ⟨88 unchanged words⟩  never "no market cap" and never $0. The count itself rides as `sharesOutstanding` with `sharesOutstandingAsOf` + `sharesOutstandingAsOfKind` ("stated" = the issuer's statement date; "checked" = when it was last confirmed against a vendor; "none" = date not recorded) and a vendor-free `sharesOutstandingSource` — quote the date with the count.

 ⟨122 unchanged words⟩ , volume, cash runway, float, shares outstanding (each row's `shares_outstanding` is DATED by `sharesOutstandingAsOf` + `sharesOutstandingAsOfKind` (stated / checked / none = date not recorded) and a vendor-free `sharesOutstandingSource`; never present the count without its date), market cap, industry, listing exchange (NASDAQ/NYSE/AMEX ⟨76 unchanged words⟩ 

 ⟨23 unchanged words⟩ `exhibit_id`, `exhibit_type`) and the snippet with context. WebsearchQuery syntax: unquoted words must ALL appear (any order, any distance); "quoted phrase", = exact wording; OR between alternatives (merger OR acquisition); -word or -"phrase" to exclude; w1 NEAR/10 w2 (also NEAR(10), -excludeor bare NEAR = within 10) = words close together; word* = prefix of 4+ letters (dilut* matches dilution, dilutive); unquoted$AAPL or a capitalised ticker scopes to that company; a form code (8-K, 10-Q, S-1), "item 5.02" / "8-K 1.05" and a bare year (2024) scope to that form / 8-K item / year. Unquoted words match anywhere in the filing, so quote multi-word phrases. When the archive states how it read the query it is in meta.understanding. Best for locating specific clauses ("change of  ⟨46 unchanged words⟩ CIK), filing, accession number, or form type.type, restrict matching to sections (section=risk_factors, mdna, ex99 …; confirmed by meta.sections, otherwise meta.unsupported says the whole filing was searched) and filter by exhibit type (exhibitTypes=EX-10). Coverage: Signal8's archive of SEC EDGAR filings  ⟨59 unchanged words⟩ 

 ⟨27 unchanged words⟩  CIK number","type":"string"},"exhibitTypes":{"description":"Keep only filings carrying ANY of these exhibit types (comma-separated): a family such as EX-10 (material contracts) matches every member (EX-10.1, EX-10.2…); an exact type such as EX-99.1 (usually the press release) only itself. Others: EX-97 clawback policy, EX-3 charter/bylaws, EX-4 instruments/warrants, EX-21 subsidiaries, EX-23 consents. \"10.1\" is accepted for EX-10.1. Listed in appliedFilters when the rows prove it; otherwise in ignoredFilters / meta.unsupported — treat the rows as unfiltered then.","maxLength":400,"type":"string"},"facets":{"description":"true = also return `facets` { forms, years, items } — [{ value, count }] over the whole match set. Absent when not measured (never zero). Default false.","type":"boolean"},"filingId":{"description":"Filter to a specific ⟨79 unchanged words⟩ ,"stemmed"],"type":"string"},"matchedSections":{"description":"Per-row `matchedSections` (default true): the sections the text matched in. ABSENT on a row means not measured (index without sections), never \"matched nowhere\".","type":"boolean"},"pattern":{"description":"Search query (minimum 2 characters), websearch syntax: \"exact phrase\", OR, -exclude.-exclude, w1 NEAR/n w2, word*, $TICKER. A CIK/filing-scoped search whose text is not indexed  ⟨14 unchanged words⟩ "minLength":2,"type":"string"},"section":{"description":"Restrict MATCHING to sections (comma-separated): business, risk_factors, cybersecurity, legal_proceedings, mdna, market_risk, controls (10-K/10-Q/20-F narrative items, by meaning — mdna is 10-K Item 7, 10-Q Part I Item 2 and 20-F Item 5), ex99 (press-release / EX-99.x exhibit text) and item_X_YY (the text under an 8-K item, e.g. item_5_02). Aliases: mda, risk, legal, cyber, press_release, 5.02. Unknown names are a 400. Needs q. Applied ONLY when the response's meta.sections names it — otherwise meta.unsupported says why (e.g. the archive index does not have sections yet) and matches may come from ANYWHERE in the filing.","maxLength":400,"type":"string"},"

 ⟨165 unchanged words⟩  open them at their sec.gov URL. Full-text `q` uses: websearchQuery syntax: unquoted words must ALL appear (any order, any distance); "quoted phrase" for an= exact phrase;wording; OR between alternatives;alternatives (merger OR acquisition); -word or -"phrase" to exclude.exclude; w1 NEAR/10 w2 (also NEAR(10), or bare NEAR = within 10) = words close together; word* = prefix of 4+ letters (dilut* matches dilution, dilutive); $AAPL or a capitalised ticker scopes to that company; a form code (8-K, 10-Q, S-1), "item 5.02" / "8-K 1.05" and a bare year (2024) scope to that form / 8-K item / year. The archive's reading of the query (tickers/companies, forms, items, years, synonym expansions such as buyback → share repurchase) is returned in `meta.understanding` when stated; pass understand=false to search the words literally. Pass facets=true for `facets` — counts of matching filings per form type, per year and per 8-K item over the WHOLE result set (absent when not measured, e.g. on the Postgres engine — never read absence as zero). With sort=relevance, page 1 shows at most a few filings per company; meta.diversity { maxPerCompany, hidden: [{cik, count}] } names who has more — page on with nextCursor (not page) or filter by that CIK. SECTION SEARCH: section=risk_factors (or mdna, business, ex99 for press releases, item_5_02, …) restricts matching to those sections; rows then carry matchedSections and meta.sections confirms it — if meta.unsupported names section instead, the whole filing was searched. exhibitTypes=EX-10 keeps filings with a material contract (EX-99.1 press release, EX-97 clawback, …); facets=true adds facets.sections / facets.exhibitTypes when measured. An empty text result is only definitive when ⟨233 unchanged words⟩ 

 ⟨132 unchanged words⟩  Exact match.","type":"string"},"exhibitTypes":{"description":"Keep only filings carrying ANY of these exhibit types (comma-separated): a family such as EX-10 (material contracts) matches every member (EX-10.1, EX-10.2…); an exact type such as EX-99.1 (usually the press release) only itself. Others: EX-97 clawback policy, EX-3 charter/bylaws, EX-4 instruments/warrants, EX-21 subsidiaries, EX-23 consents. \"10.1\" is accepted for EX-10.1. Listed in appliedFilters when the rows prove it; otherwise in ignoredFilters / meta.unsupported — treat the rows as unfiltered then.","maxLength":400,"type":"string"},"facets":{"description":"true = also return `facets`: { forms, years, items }, each a list of { value, count } over the WHOLE result set (not just this page). Present only when the archive measured them (its OpenSearch engine); absent means NOT MEASURED, never zero. Default false.","type":"boolean"},"foreignIssuer":{"description":"COMPANY FILTER. true = ⟨190 unchanged words⟩ ,"stemmed"],"type":"string"},"matchedSections":{"description":"Per-row `matchedSections` (default true): the sections the text matched in. ABSENT on a row means not measured (index without sections), never \"matched nowhere\". Ignored without q.","type":"boolean"},"maxFloat":{"description":"COMPANY FILTER. Maximum public ⟨125 unchanged words⟩ ":{"description":"Full-text query over filing bodies,bodies. websearchQuery syntax: all unquoted words must appear;ALL appear (any order, any distance); \"quoted phrase\" for= exact wording; OR 

 ⟨129 unchanged words⟩  cap, industry, listing exchange (NASDAQ/NYSE/AMEX), and float data source.source and TTM dividend yield. Each row carries dividendYield (trailing-twelve-month cash dividends with an ex-date in the last 365 days / current price, in percent), nextExDate and lastExDate — null means NOT MEASURED or not a payer, never 0%. Sort results by any sortable column. Returns matching ⟨27 unchanged words⟩ 

 ⟨127 unchanged words⟩  cash remaining","type":"number"},"maxDividendYield":{"description":"Maximum TTM dividend yield in percent. Rows with no measured yield never match.","minimum":0,"type":"number"},"maxFloat":{"description":"Maximum computed public float ⟨49 unchanged words⟩  cash remaining","type":"number"},"minDividendYield":{"description":"Minimum TTM dividend yield in percent (e.g. 3 for 3%). Rows with no measured yield never match.","minimum":0,"type":"number"},"minFloat":{"description":"Minimum computed public float ⟨70 unchanged words⟩ ","shares_outstanding","computed_float","market_cap_computed","updated_at","dividend_yield"],"type":"string"},"sortOrder": ⟨12 unchanged words⟩ 

List allthe readable exhibits (individual documents with stored text) within an SEC filing.filing, by internal filing ID. Returns exhibiteach metadataexhibit's including`id`, exhibit type, description, and content size.size and sec_url. Use thisthe to`id` identifywith whichget_exhibit_content. exhibitsExcludes toXML/XBRL. readFor the COMPLETE document list of a filing by accession number (including files with get_exhibit_content.no Excludesstored XML/XBRLtext), exhibits.use get_filing_documents.

 ⟨234 unchanged words⟩  | "share_count_unavailable" | "split_post_dates_basis" | "split_unconfirmed" | "snapshot_float_unconfirmed" | "refusal_evaluation_failed") means the figure  ⟨735 unchanged words⟩ 

 ⟨68 unchanged words⟩  always excluded (intraday ordering is unknowable). Unlike get_donor_trade_overlap, executive-branchExecutive-branch (exec-) slugs return REAL data here:  ⟨89 unchanged words⟩ 

 ⟨63 unchanged words⟩ market cap, float in SHARES, and price. 🔴 floatShares: null MEANS NOT MEASURED OR WITHHELD — never zero, never "no float". Such rows ARE included (they have a valid market cap); floatWithheldReason says why. Do not substitute a float from another source. floatUnknownCount counts those null-float rows IN the list; it is not a withholding count (withheldCount is). price is on the SAME split basis as the market cap (null when it cannot be stated on one basis), so never pair it with a pre-split price from elsewhere. WHAT THIS IS: rows are splits scheduled AS ⟨243 unchanged words⟩ 

 ⟨25 unchanged words⟩  company name, form type, filing date, exhibit count and exhibit`filingId` count.(for get_filing_document / get_filing_exhibits). If the filing is held in theSignal8's localarchive database,(isInDatabase: returnstrue) fullits metadata;content can be read here; if it is only found on SEC EDGAR, returnsthe response has isInDatabase: false and basic metadata withonly. anFor isInDatabase:the falsefull flag.document list use get_filing_documents.

 ⟨121 unchanged words⟩  / maxMarketCapComputed and returns a real total COUNT. Searches only the 500 largest matching companies by market cap (narrow the filters to reach smaller ones); for float, price, company-type, foreign-issuer or recent-IPO filters, full-text queries, or 8-K item codes, use search_sec_filings with its company filters. Coverage: Signal8's archive of SEC EDGAR filings from October 2021 to today (about 3.9M filings), updated through the day; filings older than 2021-10 may be absent, so an empty result is not proof something was never filed. Some rows are record-only (bodyStored: false — metadata from the EDGAR index with no document body), so their text cannot be read here; open them at their sec.gov URL.

Full-text search across SEC filing content, websearchthat syntaxreturns ("quotedthe phrase"MATCHING PASSAGE: one match per filing, OReach with the exhibit it was found in (`exhibit_id`, -exclude`exhibit_type`), ranked,and onethe matchsnippet perwith filing.context. QuoteWebsearch multi-wordsyntax phrases("quoted —phrase", OR, -exclude); unquoted words match anywhere in the filing. Returns matching snippets withfiling, contextso aroundquote eachmulti-word match.phrases. PowerfulBest for findinglocating specific clauses like ("change of control", "anti-dilution", "right of first refusal",) and for getting the excerpt for a known filing (pass accessionNumber or anyfilingId). specificUse languagesearch_sec_filings acrosswith filings.`q` instead when you need filing-level paging (cursor), totals, date ranges, 8-K item codes, excluded forms, or company filters. Optionally filter by company (CIK), filing, accession number, or form type. Coverage: Signal8's archive of SEC EDGAR filings from October 2021 to today (about 3.9M filings), updated through the day; filings older than 2021-10 may be absent, so an empty result is not proof something was never filed. Some rows are record-only (bodyStored: false — metadata from the EDGAR index with no document body), so their text cannot be read here; open them at their sec.gov URL.

Search and list SEC filings with filtering by company (CIK), form type, date  ⟨6 unchanged words⟩  forms, optionally with a full-text query over the filing bodies.bodies, Returnsand filingoptionally metadataby COMPANY FACTS (formmarket typecap, filingfloat, dateprice, sector, industry, exchange, company nametype, accessionforeign numberissuer, acceptancerecent timeIPO). Returns rows with `id` (the filingId for get_filing_document / get_filing_exhibits), primarycik, documentcompany_name, 8-Kaccession_number, items)form_type, filing_date, acceptedAt, primaryDocument, bodyStored and itemNumbers, plus `pagination` with `total`, `totalRelation` ("eq" exact; "gte" a lower bound when the count was cut at 10,000 — then `totalEstimate`, when present, is "about N"), `hasMore` and `nextCursor`. With `q`, unfiltered rows also carry a`rank` rankedand a `snippet`. UseCoverage: thisSignal8's archive of SEC EDGAR filings from October 2021 to findtoday (about 3.9M filings), beforeupdated readingthrough theirthe contentday; filings older than 2021-10 may be absent, so an empty result is not proof something was never filed. Some rows are record-only (bodyStored: false — metadata from the EDGAR index with get_filing_documentno ordocument get_filing_exhibits.body), so their text cannot be read here; open them at their sec.gov URL. Full-text `q` uses websearch syntax: unquoted words must ALL appear (any order, any distance); "quoted phrase" for an exact phrase; OR between alternatives; -word to exclude. An empty text result is only definitive when the response's `searchIndex.complete` is true — otherwise `notes` says the index has not confirmed full coverage of the range. The `ciks` filter matches the filing's primaryown CIK: an insider Form 4 filed under the reporting owner's CIK is NOT returned for the issuer's CIK. COMPANY FILTERS (any of minMarketCap … recentIpo) are served from the archive only, using Signal8's company data: (1) only issuer-filed forms are searched (10-K, 10-Q, 8-K, S-1, S-3, 424B*, DEF 14A, 20-F, 6-K, …); third-party forms such as Form 4, 144 and SCHEDULE 13D/G are dropped and listed in `companyFilter.droppedForms`, and with no formTypes the default is 10-K, 10-Q, 8-K, S-1, S-3, DEF 14A, 20-F, 40-F, 6-K; (2) companies with no measured value for a filter are EXCLUDED, never assumed, and `companyFilter.unknown` counts them per filter; (3) unknown values (e.g. a misspelt sector) are a 400 that lists the allowed names; (4) when the archive cannot answer (e.g. a date range it does not fully hold yet) the call FAILS with code COMPANY_FILTER_NOT_APPLIED and a reason — it never returns unfiltered rows; (5) results are cursor-paged (use nextCursor, not page) and cannot be combined with acceptedFrom/acceptedTo/bodyStored; company-filtered text matches carry no snippet (use search_filing_text with the accessionNumber for the excerpt). The response then includes `companyFilter` (matchedCompanies, searchedCompanies, unknown, forms, droppedForms), `appliedFilters` and `notes`. To COUNT or list companies (not filings) by market cap etc., use screen_companies instead.

 ⟨9 unchanged words⟩ CIK numbers to filter by (e.g., \"0000320193320193,0001018724\1018724\"; zero-padded also accepted)","type":"string"},"companyTypes":{"description":"COMPANY FILTER. Comma-separated: operating, fund (registered funds/ETFs; adds N-CSR, N-CEN, N-PX, 485BPOS, 497/497K to the default forms), spac (blank-check companies)","type":"string"},"cursor":{"description":"Opaque keyset cursor from ⟨11 unchanged words⟩  — it is stable while new filings arrive. Required for page 2+ of a company-filtered search.","type":"string"},"dateFrom":{"description":"Start datefiling filterdate (YYYY-MM-DD). The archive starts 2021-10.","type":"string"},"dateTo":{"description":"End datefiling filterdate (YYYY-MM-DD)","type":"string"},"exchanges":{"description":"COMPANY FILTER. Comma-separated listing venues: NASDAQ, NYSE, AMEX (= NYSE American), OTC","type":"string"},"excludeForms":{"description":"Comma-separated form types to ⟨6 unchanged words⟩  Exact match.","type":"string"},"foreignIssuer":{"description":"COMPANY FILTER. true = companies with foreign-issuer evidence (a 20-F, 40-F or 6-K seen, or an ADR listing); false = none seen since the evidence log began (companyFilter.foreignEvidenceSince)","type":"boolean"},"

 ⟨17 unchanged words⟩  each row enriched with rvol, marketCap, floatShares (the measured POINT float, null on a withheld dilution-snapshot row — such a row then carries "floatEstimate" {shares, bound: "point" ~ | "ceiling" <= | "floor" >=, basis}, the snapshot's own labelled figure: present it as the number with its marker, never as a point, never in arithmetic; neither present = not measured, never 0. The low-float board admits on that figure's VALUE: a <= 8M ceiling is in, a >= 26M floor or a ~27M estimate is out), short interest, dilution, and news/catalyst flags.  ⟨231 unchanged words⟩ 

 ⟨22 unchanged words⟩  analyzing bearish sentiment or potential short squeeze setups. 🔴 "shortInterestPercent" null ALWAYS carries "shortInterestPercentReason" (since 2026-10-04): "float_not_point" = the float on record is a snapshot BOUND or ESTIMATE (<= / >= / ~), and a percent-of-float is never struck off a bound — do NOT divide "shortInterest" by a ceiling, floor or estimate from get_float yourself; "float_unavailable" = no float of any kind on record; "short_pct_of_float_unreconciled" = a point float existed but the ratio was implausible or the split basis could not be resolved. A null percent is NOT MEASURED, never 0%.

Since 2026-10-03 the float column ALWAYS carries the dilution snapshot's own figure when the snapshot has one: "float" is the measured POINT float (null on a withheld snapshot row), and such a row then carries "floatEstimate" (shares) + "floatEstimateBound" ("point" ~ / "ceiling" <= / "floor" >=) + "floatEstimateBasis" (the working) — present it as the number with its marker, e.g. "<= 7,953,957". minFloat / maxFloat and the computed_float sort compare the figure's VALUE (COALESCE of the two), so a bounded row is kept; a row with neither is NOT MEASURED, never 0. 🔴 Never feed floatEstimate into a market cap, a percent-of-float or a short-interest ratio — a bound is not a point. Screen companies by price range, volume, cash  ⟨49 unchanged words⟩ 

 ⟨233 unchanged words⟩  "affiliate_unreconciled" | "share_count_unavailable" | "split_post_dates_basis" | "snapshot_float_unconfirmed" | "refusal_evaluation_failed") means the figure was deliberately WITHHELD, notby unavailablethe snapshot AND no lower tier could answer — do not substitute a vendor float of your own. Since 2026-09-29 a snapshot that withheld its float FALLS THROUGH: when a lower tier can answer, "source" names that tier (computed / sec-10k / polygon / fmp), "floatShares" is ITS figure under ITS definition, and "snapshotFloatWithheldReason" carries the snapshot's reason as context ("the filing-derived snapshot could not state a float; showing <source>"). Read that field as provenance, never as a second refusal. A withheld result may carry up to four  ⟨153 unchanged words⟩ confidence" describes the estimate, not a float. TodaySince 2026-09-29 it is populatedthe onlySNAPSHOT FLOAT LADDER figure on affiliate_basis_staleany withheld snapshot ticker and nullmay also carry "rung" ("non_affiliate_estimate" | "shares_outstanding_ceiling" | "rolled_forward_lower_bound"), "bound" ("point" = about X, render "~X"; "ceiling" = AT MOST X, render "<= X"; "floor" = AT LEAST X, render ">= X") and "asOf"; a floor or ceiling is a BOUND, never a point value. null means no figure could be derived from the ordinarysnapshot's outcome.structured fields. An estimate never becomes floatShares — present shares ⟨145 unchanged words⟩ 

 ⟨41 unchanged words⟩ via limit/offset (default 100). Aggregate stats (institutional/insider/beneficial/retailinstitutional/insider/beneficial totals and percentages, plus the `retail` residual) are always included in full. `retail` is the UNATTRIBUTED residual: shares no filing accounts for, which includes recent offering buyers and acquisition holders who have not filed as well as retail investors — an upper bound on retail, never a retail count. When `retail.reason` is `filed_exceeds_outstanding` the residual is withheld (null) because the counted filings exceed the shares outstanding.

 ⟨56 unchanged words⟩ :null + halted:true when nothing is recoverable. marketCap is the live price × the SEC-derived share count; when that count is unavailable it is null with marketCapWithheldReason set — NOT MEASURED, never "no market cap" and never $0.

Full-text substring search across all SEC filing exhibitcontent, content.websearch syntax ("quoted phrase", OR, -exclude), ranked, one match per filing. Quote multi-word phrases — unquoted words match anywhere in the filing. Returns matching snippets with context around each match. ⟨31 unchanged words⟩ 

 ⟨69 unchanged words⟩ number"},"pattern":{"description":"Search patternquery (minimum 2 characters)., Substringwebsearch matchsyntax: \"exact phrase\", case-insensitive.OR, -exclude. A CIK/filing-scoped search whose text is not indexed yet falls back to a literal case-insensitive substring match (response engine: \"substring\").","minLength":2,"type":"string"} ⟨20 unchanged words⟩ 

 ⟨6 unchanged words⟩ filtering by company (CIK), form type, and date range.range, Returns8-K paginateditem resultscodes and excluded forms, optionally with a full-text query over the filing metadatabodies. includingReturns filing metadata (form type, filing date, company name, accession number, acceptance time, primary document, 8-K items) plus `pagination` with `total`, `totalRelation` ("gte" when the count was cut at 10,000) and accession`nextCursor`. number.With `q`, rows also carry a ranked `snippet`. Use this to find filings before reading their content with get_filing_document or get_filing_exhibits. The `ciks` filter matches the filing's primary CIK: an insider Form 4 filed under the reporting owner's CIK is NOT returned for the issuer's CIK.

 ⟨17 unchanged words⟩ ,0001018724\")","type":"string"},"cursor":{"description":"Opaque keyset cursor from a previous response's pagination.nextCursor. Prefer it over page for deep paging — it is stable while new filings arrive.","type":"string"},"dateFrom":{"description":"Start date filter ( ⟨8 unchanged words⟩  (YYYY-MM-DD)","type":"string"},"excludeForms":{"description":"Comma-separated form types to EXCLUDE (e.g., \"4,13F-HR\"). Exact match.","type":"string"},"formTypes":{"description":"Comma-separated form types (e.g., \"10-K,10-Q,8-K,S-1,S-3,424B5\"). Exact match — list amendments (e.g. \"10-K/A\") explicitly.","type":"string"},"items":{"description":"Comma-separated 8-K item codes, ALL required (e.g., \"2.02,9.01\" = results + financial statements; \"5.02\" = officer/director changes; \"1.01\" = material agreements).","type":"string"},"page":{"description":"Page number (1-indexed, default: 1). Ignored when cursor is set.","minimum":1,"type":"number"} ⟨10 unchanged words⟩ ":100,"minimum":1,"type":"number"},"q":{"description":"Full-text query over filing bodies, websearch syntax: \"quoted phrase\", OR, -exclude. Quote multi-word phrases — unquoted words match anywhere in the filing.","maxLength":500,"type":"string"},"sort":{"description":"newest (by acceptance time, default) or relevance (requires q; ranks the newest 1,000 matches)","enum":["newest","relevance"],"type":"string"}},"type":"object"}

 ⟨151 unchanged words⟩ not infer them from shelf or ATM capacity. Requires theIncluded Dilutionin SnapshotsSignal8 add-onPro (403 ADDON_REQUIREDFEATURE_NOT_AVAILABLE, requiredTier "pro", without it). That is a tier upgrade, not an add-on purchase — the add-on is no longer sold.

 ⟨235 unchanged words⟩ the company as having none of that instrument. Requires theIncluded Dilutionin SnapshotsSignal8 add-onPro — without it the call returns 403 with code ADDON_REQUIRED,FEATURE_NOT_AVAILABLE whichand `requiredTier: "pro"`. That is a subscription-tier upgrade, not an add-on purchase requirementpurchase; andthere is DISTINCTno fromseparate aadd-on subscription-tierto upgrade.buy.

 ⟨296 unchanged words⟩ counted rather than published on a mixed basis. Requires theIncluded Dilutionin SnapshotsSignal8 add-onPro — without it the call returns 403 with code ADDON_REQUIRED,FEATURE_NOT_AVAILABLE whichand `requiredTier: "pro"`. That is a subscription-tier upgrade, not an add-on purchase requirementpurchase; andthere is DISTINCTno fromseparate aadd-on subscription-tierto upgrade.buy.

 ⟨300 unchanged words⟩ counted rather than published on a mixed basis. Requires theIncluded Dilutionin SnapshotsSignal8 add-onPro — without it the call returns 403 with code ADDON_REQUIRED,FEATURE_NOT_AVAILABLE whichand `requiredTier: "pro"`. That is a subscription-tier upgrade, not an add-on purchase requirementpurchase; andthere is DISTINCTno fromseparate aadd-on subscription-tierto upgrade.buy.

 ⟨255 unchanged words⟩ counted rather than published on a mixed basis. Requires theIncluded Dilutionin SnapshotsSignal8 add-onPro — without it the call returns 403 with code ADDON_REQUIRED,FEATURE_NOT_AVAILABLE whichand `requiredTier: "pro"`. That is a subscription-tier upgrade, not an add-on purchase requirementpurchase; andthere is DISTINCTno fromseparate aadd-on subscription-tierto upgrade.buy.

 ⟨340 unchanged words⟩ counted rather than published on a mixed basis. Requires theIncluded Dilutionin SnapshotsSignal8 add-onPro — without it the call returns 403 with code ADDON_REQUIRED,FEATURE_NOT_AVAILABLE whichand `requiredTier: "pro"`. That is a subscription-tier upgrade, not an add-on purchase requirementpurchase; andthere is DISTINCTno fromseparate aadd-on subscription-tierto upgrade.buy.

 ⟨319 unchanged words⟩ counted rather than published on a mixed basis. Requires theIncluded Dilutionin SnapshotsSignal8 add-onPro — without it the call returns 403 with code ADDON_REQUIRED,FEATURE_NOT_AVAILABLE whichand `requiredTier: "pro"`. That is a subscription-tier upgrade, not an add-on purchase requirementpurchase; andthere is DISTINCTno fromseparate aadd-on subscription-tierto upgrade.buy.

Get float data for a company: floatShares, sourcefloatDefinition, floatDefinitionas-of date, and as-of— date.when the headline float is WITHHELD — the bounds and labelled stand-ins that replace it. Use when analyzing share supply. CARDINAL RULE: null / absent / withheld means NOT MEASURED — never "no float", never "no restricted shares", never zero. A numeric 0 is a MEASURED zero and passes a null check, so decide from the reason/marker field ("floatWithheldReason", "restrictedBlockCount", "provenance", "direction"), never from the value alone. IMPORTANT: check "floatDefinition" to know which float ⟨48 unchanged words⟩ ), or "free_float" (vendor methodology). TODAY "floatDefinition" and "source" are only present on dilution-snapshot results: a vendor-sourced float arrives with neither (just floatShares, asOfDate, floatEstimated) because vendors are not named — read an ABSENT floatDefinition as "not a filing-derived tradeable float" and do not assume it is tradeable. On snapshot tickers "nonAffiliateFloatShares" carries the SEC  ⟨38 unchanged words⟩  "stated_figure_impossible" | "affiliate_basis_stale" | "affiliate_figure_declined" | "affiliate_unreconciled" | "share_count_unavailable" | "split_post_dates_basis") means the figure was deliberately WITHHELD, not unavailable — do not substitute a vendor float.float Onof your own. A withheld result may carry up to four stand-ins, each labelled IN the object; NONE is a float and NONE may be fed to a market cap, a percent-of-float, a short-interest ratio, a screen, a sort or a comparison: (1) "affiliate_basis_staletradeableFloatCeiling" is an UPPER BOUND — the tradeable float is AT MOST this; render it with "<=" and nothing else (null = no bound stateable, never 0). (2) "affiliateBasis.floatLowerBound" / "floatUpperBound" are a RANGE the float lies in (both— ends"between orX neither;and Y"; each end is null =independently when not measurable), so state the ends that exist as inequalities. This object also appears on PUBLISHED floats with both ends equal to floatShares — it is a rangebound toonly statewhen floatShares is null. (3) "floatEstimate" is a BADGED ESTIMATE {shares, basis, direction, confidence}: "basis" is the itemised working (quote it), "direction" says which way it errs ("overstates" = the true float is at most shares; "understates" = at least; "unknown"), "confidence" describes the estimate, not a float. Today it is populated only on affiliate_basis_stale and null is the ordinary outcome. An estimate never becomes floatShares — present shares WITH basis and direction, never alone. Unrelated to the boolean "floatEstimated", which flags a published floatShares back-derived from a USD public-float disclosure. (4) "vendorFallback" is a third-party free-float figure {floatShares, provenance:"vendor", floatDefinition:"free_float", asOfDate, splitAdjusted} offered beside the withhold — NOT our number, toNOT computeon with.the tradeable definition, already refused when our filings refute it. Present it as the vendor free float as of its own date, never in the "Float" slot. "restrictedShares":null means NOT MEASURED (a block  ⟨32 unchanged words⟩  share count — the number itself is unchanged. "tradeableBasis":"convention_conservative" means the filings did not answer and a published convention resolved it — see "tradeableConvention" and do not present the figure as exact.

 ⟨92 unchanged words⟩ = live cumulative premarket volume ÷ the trailing 90-session30-session average cumulative volume AT THE SAME TIME OF ⟨40 unchanged words⟩ = the same live volume ÷ the trailing 90-session30-session average FULL premarket session (answers "what fraction ⟨96 unchanged words⟩ 

 ⟨36 unchanged words⟩  session's volume to a trailing same-session baseline (9030 trading days by default — configurable via "baselineDays" ⟨141 unchanged words⟩ 

 ⟨45 unchanged words⟩ cumulative volume known BY that cutoff ÷ the 90-daytrailing baselineDays (default 30) average of the SAME cutoff (not the full ⟨60 unchanged words⟩  in trading rows (same-session days). Default 90;30; values outside 20-250 are clamped. This is the ⟨42 unchanged words⟩ smoother and slower to react. The warm-up lookback and the minimum-warm-days gate scalescales with it automatically, so a wide window is never under-filled into an inflated ratio.ratio; the minimum-warm-days gate (20 prior sessions) does NOT scale down, so at the 30-row default a ticker needs 20 of its last 30 sessions populated before rvol is non-null. Omit for the standard 90-day30-day baseline.baseline; pass 90 for the pre-2026-09 window.","maximum":250,"minimum":20,"type ⟨56 unchanged words⟩ 

 ⟨27 unchanged words⟩ bid-price compliance (1:250 cumulative reverse-split cap). IMPORTANT: `cumulativeReverseSplitRatio2yr` is null when it CANNOT BE STATED, which is NOT the same as no reverse splits — read `cumulativeWithheldReason` to tell them apart. `unconfirmed_splits_in_window` means at least one in-window reverse split is evidenced by only one source, so no cumulative is defensible: the confirmed splits alone would understate it, and understating a cumulative can put an issuer that breaches the 1:250 cap under it. Each row also carries `confirmed`; a `confirmed: false` split appears only in the announcement calendar, which retains every announcement and retracts none, so it is usually an amended or superseded announcement and occasionally a real split the effected-splits ledger missed. Both are returned. `ratioDisplay` and `type` are likewise null for a ratio whose exact factor cannot be determined.

 ⟨144 unchanged words⟩ :"Sort results by field (default: filing_date). Ties are broken by filing id, so a given page is stable and reproducible for a fixed corpus.","enum":["filing_date","form_type"," ⟨4 unchanged words⟩ {"description":"Sort direction (default: desc). Note that paging is offset-based over a corpus that is continuously ingesting new filings, so when walking many pages of a broad date range, newly-arrived filings can shift rows between pages; page on a closed dateFrom/dateTo window if you need an exhaustive, non-overlapping traversal.","enum":["asc","desc"], ⟨4 unchanged words⟩