← All companies

NLYAnnaly Capital Management, Inc.

Industry Mortgage REIT Updated 25 Sep 2026

The idea

A 14% yield at 1.06x book, and the yield is the whole base case. From 2021 to 2025 the company reported $12.71/share of EAD while a holder's economic result was about zero — the dividend was paid out of carry and the book absorbed the cycle — and the share count doubled. Since 2023 it has earned double-digit economic returns; the open question is whether that is the new resi-credit/MSR mix or the MBS basis tightening under the GSEs' $200bn purchase mandate.

Events to watch

  • Q3 2026 results ~2026-10-21 (DERIVED from cadence) — book value per share after a 7.9% fall in the stock since 2026-08-20
  • EAD against the raised $0.75 dividend
  • funding of the $442.5m Series I redemption on 2026-10-01
  • the current-coupon MBS basis (94bp at 2026-06-30) and the GSE mandate's remaining capacity (~$45bn of $200bn settled through May)
How we got the target

Dossier model table (NLY/dossier-2026-09-23.md §5), book value × price-to-book, because an mREIT is valued on book: base 1.05x a flat $20.15 BVPS; bear 0.90x on $17.90 (basis +32bp at the company's own −8.7%-NAV-per-25bp sensitivity, linearly extrapolated); bull 1.15x on $21.50. Price-only: the ~$6.00 of dividends over two years (14.0% yield) is NOT in these figures, so the upside the console computes understates total return by roughly 28 points in the base case. Price is market data's 2026-09-22 close, fetched 2026-09-23. Time frame: mid-2028 (Q2 2028 book).

Target: from the research notes. Prices: market data get_latest_stock_price, close of 2026-09-22 (fetched 2026-09-23)

What would prove it wrong

4 warning signs
  1. BVPS below $18.45 at any quarter-end (the Q2 2025 figure, the lowest quarter-end BVPS in the filings read).

  2. EAD below the dividend in two consecutive quarters (streak is nine at Q2 2026).

  3. Economic leverage above 7.0x without a stated reason (5.5-5.8x since 2023).

  4. ATM issuance while P/B is below 1.0.

    Details

    — management's own first criterion for raising.

How far along

10 of 10 steps done

Next step

on the Q3 2026 release (~2026-10-21) read book value per share first, then EAD against the $0.75 dividend. The stock fell 7.9% from $23.29 to $21.44 between 2026-08-20 and 2026-09-22 with no company filing to explain it beyond the Series I redemption notice; if the book fell with it, the falsifier ($18.45) is one more basis shock away. Added 2026-09-25: read footnote 2 of the GAAP→EAD reconciliation for the swaps' net interest component — $716.5m in 2025 against ~$1.2bn in 2024, from 77% of EAD to 35% — because that is the line the dividend-coverage promises rest on. Then check how the $442.5m Series I redemption on 2026-10-01 was funded.

Needs your decision

S6 product test / expert call, S7 conviction.

Failed checks

none. S8 is closed conservative and S2's capital-allocation verdict is neutral (both added 2026-09-25), so this backfill adds no kill criterion (§25 requires one only for an over-promiser or a capital destroyer) — but dossier-2026-09-25.md proposes one for the owner: EAD below the dividend in any quarter in which swap carry is under $500m annualised. ⚠️ S3's new earnings-quality check found the swap carry inside EAD down ~40% in a year, from ~77% of EAD to 35%, with the replacement futures carrying no EAD interest component — and three of S8's six rows are scored on EAD. S0 scored 1/4, which is a finding, not a verdict (SKILL.md §4): the two structural 0s are that the dividend was paid out of carry while the book absorbed the cycle, and that the book was refilled with new shares.

Who runs it

The Front Office

Step by step

  1. S0How it does as AI gets cheapDone

    lens applied (cluster: Additions 2026-09-23 — NLY (Annaly Capital Management), industry stages, 2026-09-23)

    SKILL.md §4.3 risk-carrying variant, spread branches, with tests 1 and 3 TRANSLATED for an mREIT (stated in the file)

    Full notes

    lens 1/4: earnings real 0 (2022-25 EAD $12.71/sh against an economic result of about −$0.15/sh — BVPS $31.88 → $20.21 with $11.52 of dividends paid), edge survives the cycle 0 (economic spread positive 5/5 but 1.89% → 1.22% → 1.40%, dividend cut 26% in 2023, GAAP spread negative 2023-24, swap carry now fading), capital theirs 1 (economic leverage 5.5-5.8x inside a stated 10:1 ceiling, 37 repo counterparties, +25bp MBS spread = −8.7% NAV), no dilution 0 (diluted shares 353.7m → 740.3m since 2020, 2.09x). Cycle measure: the mortgage basis, 137 → 126 → 87 → 94bp (Dec-23 → Jun-26), held tight by the GSEs' $200bn purchase mandate — the case needs it not to widen. Falsifier: BVPS < $18.45. Liquidity: ADV $183.1m/day (21 sessions to 2026-09-22, market data NYSE:NLY). P/B 1.06; yield 14.0%.

  2. S1Where the idea came fromDone

    watchlist (owner, 2026-09-23 addition, WATCHLIST_PLAN.md §24). No Tier 0 promotion. One board claim row (Whalen, 2026-09-19, bullish).

  3. S2Quick deal-breaker checkDone

    survives. Economic leverage 5.6x. $9.6bn available for financing (~57% of capital). EAD $0.79 vs $0.75 dividend, above it nine quarters running. EY unqualified ICFR opinion.

    ⚠️ What survives is the company, not the share: book per share roughly halved since 2021 ($31.88 → $20.15) while the share count doubled ($7.78bn of common raised 2022-25, $955.7m in H1 2026). GAAP net interest income was −$111m in 2023. No buybacks since 2020. · capital allocation (added 2026-09-25): NEUTRAL. §25 scores an mREIT on ROE and BVPS compounded with dividends. Five years: $1.93bn of common dividends paid and $8.33bn of common stock sold, with no buyback since 2020 ($955.7m more raised in H1 2026). 2022-2025 economic result ≈ ZERO — BVPS −$11.67 against $11.52 of dividends on a $31.88 start; 2023-2025 alone compounds ≈ +42% (+6.0%, +11.9%, +20.2%). GAAP ROE 15.1% / −18.3% / 8.2% / 14.7%, four-year average 4.9%, against the company's own EAD ROE of 13.28% and 14.47%. The discipline was kept, though: implied average issue price ~$21.21 (2023), ~$19.91 (2024), ~$22.63 (2025) against year-end book of $19.44 / $19.15 / $20.21 — above book in each of the last three years, so the doubling of the share count was book-accretive per share, which the S0 lens's dilution 0 understates. On 2021-2025 this reads destroys; on 2023-2025 compounds; neutral is the honest label across the window, and it turns on whether the post-2022 recovery is durable — S7's open question. Issue prices are implied from share-count change and absorb stock compensation, so they are slight over-estimates.

  4. S3Reading the company's reportsDone

    (gate-minimum)

    FY2025 10-K Item 1A and Item 7. Q2 2026 10-Q MD&A, Item 3 sensitivity and counterparty table; 2026 DEF 14A; 8-Ks 2025-12-22 (ATM $2.5bn), 2025-12-29 (Chief Legal Officer retirement), 2026-08-19 (Series I redemption, $442.5m, 2026-10-01)

    Full notes

    Q1 and Q2 2026 transcripts (Equibles), read Q&A-first. Three disclosed risks with locators: leverage, spread compression, the GSE relationship. Two dodges logged below. Red flags: none — no NT filings, no comment letters since 2021, EY since 2012. · EQ (added 2026-09-25): 3 flags. §25 test 6 is the sector test and it is the spine here. (1) EAD exceeded GAAP by $6.12/share — 54% — over five years (EAD $17.36 vs GAAP $11.24), essentially all of it 2023 (+$2.86 vs −$3.61); over 2022-25 the dividend was $11.52/share against $4.85 of GAAP earnings, 2.4x. The difference went through book value, which is why S2's four-year economic result is zero. 2025 is clean: EAD $2.92 = GAAP $2.92. (2)

    ⚠️ The biggest ingredient of EAD is collapsing. Footnote 2 of the 2026-01-28 reconciliation puts the swaps' net interest component — which sits INSIDE EAD — at $716.5m (2025) against ~$1.2bn (2024), i.e. 35% of EAD, down from ~77%, and the FY2025-Q1 call says the unwound swaps were replaced by futures that carry no EAD interest component. EAD rose 29% while its largest component more than halved: genuine improvement, or a measure whose year-on-year comparability has broken — the disclosure does not separate them, and three of S8's six rows are scored on EAD. (3) "Transaction expenses and non-recurring items" is added back every period and is growing — $20.3m (2024) → $27.8m (2025), +37% — and footnote 4 says it is securitisation costs on the resi whole-loan shelf management is deliberately expanding. Clean: the reconciliation is complete and footnoted (Q4 2024 added back $2,010.7m of investment losses against $1,958.8m of derivative gains — the hedge working, disclosed as such). Cash-conversion, accruals and working-capital tests are recorded as not meaningful for a levered MBS book, not skipped.

  5. S4Rivals and suppliersDone

    (cluster: Additions 2026-09-23 — NLY (Annaly Capital Management), industry stages, 2026-09-23)

    price- taker in Agency (the 2026 price-setter is the GSE mandate); some pricing power on the OBX shelf and in MSR via the subservicer model; the concentration is funding (37 counterparties, $88.0bn).

  6. S5Who owns the stockDone

    13F at 2026-06-30: BlackRock 87.2m (+5.5m), State Street 29.4m, Invesco 7.7m, Citadel 7.3m, T. Rowe Price 7.2m (−11.3m, −61%). Insiders 0.14% (11 people, 1.07m sh)

    Full notes

    Insider pattern: CEO sold 3 × 50,000 in 2026 at $22.88-23.13, COO and CFO also sold; no open-market purchase in the filings read. Short interest 20.7m (2.7%), 3.8 days. Shelf: the $2.5bn ATM IS the overhang — drawn at $447-509m a quarter in H1 2026.

    ⚠️ Q2 13F ingestion incomplete (537 vs 765 filers; Vanguard not yet in — not an exit).

    ⚠️

    ⚠️ the Sixth Street row is NOT a holding (duplicate of Schwab IM).

  7. S6What workers and customers sayDone

    (cluster: Additions 2026-09-23 — NLY (Annaly Capital Management), industry stages, 2026-09-23)

    Whalen's claim row (owns it, ~0.9x book basis, survivor-of-consolidation thesis). BTIG upgrade to Buy, Jan 2026 [search-summary]. No short report found. Product test / expert call open OPEN (user)

  8. S7The case for it, and pricesDone

    (cluster: Additions 2026-09-23 — NLY (Annaly Capital Management), industry stages, 2026-09-23)

    four kill criteria written; variant: judge it on economic return per share over a cycle, not on yield or book.

    Full notes

    Conviction open OPEN (user) · valuation (bear/base/bull; moved from S8 2026-09-23): P/B model to mid-2028: bear ~$16.1 (basis +32bp, book −11%, dividend cut), base ~$21.2 (1.05x a flat book), bull ~$24.7 (1.15x on $21.50). Base-case total return is the dividend (~+27% over two years), price flat. A basis trade, not a compounder.

  9. S8Do the bosses keep their word?Done

    conservative

    6 rows, written 2026-09-25 from all eight archived calls and the 2026-01-28 release.

    Full notes

    3 beat, 3 met, none missed, none withdrawn, and no departed rows (Finkelstein CEO and Wolfe CFO throughout; the only Item 5.02 since 2024 is the Chief Legal Officer's 2025-12-29 retirement). Rows: the $0.70 dividend guided by the CFO on 2025-05-01 to hold for the year held (and was raised to $0.75 in Q2 2026); the CEO's 2025-07-24 "cover and potentially out earn the dividend for the remainder of the year" beat (EAD $0.73, $0.74 vs $0.70); his 2026-01-29 "expect outearned certainly in Q1" beat (EAD $0.76); the standing "economic return approximating or above the dividend" beat (FY2025 economic return 20%+ vs a 14.6% dividend on opening book); equity issued above book in each of the last three years, the stated criterion, met; economic leverage held at 5.5-5.8x against a 10:1 ceiling, met.

    ⚠️ Read the verdict with what is being promised. This company gives no revenue, EPS, margin, book-value or full-year guidance of any kind; the longest promise runs one quarter; and three of the six rows are about covering a dividend management itself sets. The variable that decided the holder's return — book value per share, $31.88 (2021) → $20.21 (2025) — is never promised. Context just outside the scored window: the dividend was cut 26% in 2023 ($0.88 → $0.65), which is why these rows read as a record being rebuilt rather than extended. Unscoreable and recorded rather than guessed: the $125m bulk MSR purchase the CEO said on 2024-10-24 would close before year-end is never confirmed closed on the next call.

  10. S9What to watch nextDone

    trigger written; next earnings expected ~2026-10-21 (DERIVED from cadence: 2025-10-22, 2026-04-21, 2026-07-21 releases). Tier 0 EDGAR sweep covers the feed (cik 1043219)

    Full notes

    Leading indicator: the current- coupon MBS basis and the GSE mandate's remaining capacity.

Sources and notes

Sources 11 documents
Short nameDocumentWhere it came fromRead on
10K-2025Annaly 10-K for FY2025, filed 2026-02-12https://www.sec.gov/Archives/edgar/data/1043219/000104321926000013/nly-20251231.htm2026-09-23
10K-2023Annaly 10-K for FY2023 (2021-2023 columns of the results summary)https://www.sec.gov/Archives/edgar/data/1043219/000162828024005115/nly-20231231.htm2026-09-23
10Q-Q2-26Annaly 10-Q for the quarter ended 2026-06-30, filed 2026-07-29https://www.sec.gov/Archives/edgar/data/1043219/000104321926000058/nly-20260630.htm2026-09-23
DEF14A-26Annaly DEF 14A filed 2026-04-30https://www.sec.gov/Archives/edgar/data/1043219/000110465926052954/tm261328-1_def14a.htm2026-09-23
8K-ATM-25Annaly 8-K filed 2025-12-22 — Distribution Agency Agreements for a $2.5bn at-the-market programhttps://www.sec.gov/Archives/edgar/data/1043219/000119312525328728/d93819d8k.htm2026-09-23
424B5-ATM-25Annaly 424B5 prospectus supplement filed 2025-12-22 (ATM program)https://www.sec.gov/Archives/edgar/data/1043219/000119312525328718/d83910d424b5.htm2026-09-23
8K-CLO-25Annaly 8-K filed 2025-12-29 — retirement of the Chief Legal Officerhttps://www.sec.gov/Archives/edgar/data/1043219/000119312525335363/d15017d8k.htm2026-09-23
8K-SERI-26Annaly 8-K filed 2026-08-19 — redemption of the Series I preferredhttps://www.sec.gov/Archives/edgar/data/1043219/000104321926000061/nly-20260819.htm2026-09-23
FORM4-2026Form 4 filings, 2026 (CEO 0001688537-26-000005/-000009/-000011; COO 0001631803-26-000004/-000006; CFO 0001796495-26-000005/-000010)python ledger/research/tools/threespread.py insiders NLY --days 365, then each filing's XML on www.sec.gov2026-09-23
13F-Q2-26Institutional holders by manager, 13F report date 2026-06-30 (ingestion incomplete — see the edge notes)python ledger/research/tools/threespread.py holders NLY2026-09-23
CLUSTER-NLYAdditions 2026-09-23 cluster pass, NLY memberledger/research/_clusters/cluster-additions-NLY-2026-09-23.md2026-09-23
Links to other companies 23 links
holds 6context 4insider transaction 3issuer of 2regulator of 2officer of 2supplier dependency 1governance structure 1event 1auditor of 1
FromLinkToDateWhere it says so
NLYissuer ofAnnaly $2.5bn at-the-market common stock program (agreements dated 2025-12-22)2026-06-3010Q-Q2-26 — Item 2, MD&A, capital paragraph following the results table; and Note, Capital Stock
NLYcontextAnnaly $2.5bn at-the-market common stock program (agreements dated 2025-12-22)2025-12-22424B5-ATM-25 — Cover page, program description
NLYcontextNLY2025-12-3110K-2025 — Item 7, MD&A, overview
NLYcontextNLY2026-06-3010Q-Q2-26 — Item 2, MD&A, Business Environment
NLYissuer ofAnnaly 6.750% Series I Fixed-to-Floating Rate Cumulative Redeemable Preferred (NLY.I)2026-08-198K-SERI-26 — Item 8.01
Secured-financing counterparties (37: 23 North America, 10 Europe, 4 Japan)supplier dependencyNLY2026-06-3010Q-Q2-26 — Item 2, MD&A, Counterparty Risk Management table
Fannie Maeregulator ofNLY2025-12-3110K-2025 — Item 1A, Compliance, Regulatory & Legal Risks
Freddie Macregulator ofNLY2025-12-3110K-2025 — Item 1A, Compliance, Regulatory & Legal Risks
NLYholdsAgency mortgage-backed securities2026-06-3010Q-Q2-26 — Item 2, MD&A, Business Environment
Onslow Bay Financial LLCgovernance structureNLY2026-06-3010Q-Q2-26 — Item 2, MD&A, Business Environment (residential credit paragraphs)
David L. Finkelsteininsider transactionNLY2026-08-04FORM4-2026 — Form 4 accessions 0001688537-26-000005, -000009, -000011, Table I
Steven F. Campbellinsider transactionNLY2026-02-25FORM4-2026 — Form 4 accessions 0001631803-26-000004, -000006, Table I
Serena Wolfeinsider transactionNLY2026-05-05FORM4-2026 — Form 4 accessions 0001796495-26-000005, -000010, Table I
David L. Finkelsteinofficer ofNLY2026-08-04FORM4-2026 — Form 4 accession 0001688537-26-000011, reporting-owner relationship
Serena Wolfeofficer ofNLY2025-12-3110K-2025 — Item 1, Information about our executive officers
Anthony C. GreeneventNLY2025-12-298K-CLO-25 — Item 5.02
Ernst & Young LLPauditor ofNLY2025-12-3110K-2025 — Report of Independent Registered Public Accounting Firm
BlackRock, Inc.holdsNLY2026-06-3013F-Q2-26 — BlackRock, Inc. 13F-HR, report date 2026-06-30
The Vanguard Group, Inc.holdsNLY2023-12-29DEF14A-26 — Stock Ownership Information, 5% Owners table
State Street CorpholdsNLY2026-06-3013F-Q2-26 — State Street Corp 13F-HR, report date 2026-06-30
T. Rowe Price Associates, Inc.holdsNLY2026-06-3013F-Q2-26 — T. Rowe Price Associates 13F-HR, report date 2026-06-30
Charles Schwab Investment Management, Inc.holdsNLY2026-06-3013F-Q2-26 — Charles Schwab Investment Management 13F-HR, report date 2026-06-30
Sixth Street Partners Management Company, L.P.contextNLY2026-06-3013F-Q2-26 — 3spread duplicate-share-count warning, NLY holders table
History 3 entries
  • 2026-09-23 — first pass, run by hand in a session on the owner's request ("run annaly capital via pipeline skill and add it to research watchlist"). NLY was already on watchlist.yaml (added earlier the same day, WATCHLIST_PLAN.md §24) and scheduled in cluster-additions-2026-09-23 and names-additions-2026-09-23-1; both slots stay pending for HHH and 1913.HK, with NLY removed from their member lists and a note pointing here. Stages: S0/S4/S6/S7 written to Additions 2026-09-23 — NLY (Annaly Capital Management), industry stages, 2026-09-23; S1, S2, S3 (gate-minimum, two transcripts), S5 and S9 closed; S8 modelled and left open. Run finding: from the end of 2021 to the end of 2025 the company reported $12.71/share of EAD and a holder's economic result was about zero — book value per share fell $11.67 against $11.52 of dividends — and the share count doubled over the same span. Since 2023 the picture has turned (economic return +6.0%, +11.9%, +20.2%, +6.9% in H1 2026; issuance above book), and the question S7 leaves for the owner is how much of that recovery is the new resi-credit/MSR mix and how much is the basis tightening under the GSE bid. Evidence and access in dossier-2026-09-23.md. Nothing here is human-verified.
  • 2026-09-23 — stage redefinition, owner decision (WATCHLIST_PLAN.md §25): S8 is now Management credibility and is not yet run for this name; the old S8 valuation figures moved to the valuation: line under S7, and any proposed buy price or position size was removed (the pipeline no longer drafts either). No research was redone. S2 capital allocation and S3 earnings quality are also pending, in the same backfill slot.
  • 2026-09-25 — backfill slot backfill-2026-09-23-8, the three checks added on 2026-09-23 (WATCHLIST_PLAN.md §25). No other stage touched. S8 written: conservative — 6 rows, 3 beat, 3 met, none missed, none withdrawn, no departed rows — with the qualification that this company promises almost nothing checkable: no guidance of any kind, the longest promise one quarter, and three of the six rows about covering a dividend it sets itself. Run finding (S3): the largest single component of EAD is fading fast. Footnote 2 of the 2026-01-28 reconciliation puts the swaps' net interest component, which sits inside EAD, at $716.5m for 2025 against ~$1.2bn for 2024 — from about 77% of EAD to 35% in one year — while the FY2025-Q1 call says the unwound swaps were replaced by futures that carry no EAD interest component. EAD rose 29% over the same span. Either the portfolio genuinely improved or the measure's comparability broke, and the disclosure does not separate them; three S8 rows and the dividend narrative rest on it. Second finding: EAD exceeded GAAP by $6.12 a share (54%) over five years and the 2022-25 dividend was 2.4x GAAP earnings — the gap went through book value. S2 gained its capital-allocation table: verdict neutral — $1.93bn of dividends paid against $8.33bn of stock sold and no buyback since 2020, a four-year economic result of about zero and a four-year average GAAP ROE of 4.9%, against three straight years of issuance above book (~$21.21 / ~$19.91 / ~$22.63 vs book of $19.44 / $19.15 / $20.21), which is the stated discipline kept and makes the doubled share count book-accretive per share. A seventh kill criterion is proposed to the owner and not added, the slot's scope allowing new ones only for an over-promiser or a capital destroyer. Thin evidence: one promise (the $125m bulk MSR purchase of Q3 2024) could not be confirmed closed and is left unscored; stock-based compensation is not separately tagged in this registrant's XBRL. No Equibles and no market data call was spent. Locators in dossier-2026-09-25.md. Nothing here is human-verified.
Full notes, 25 Sep 2026 15 min read

NLY — dossier 2026-09-25 (backfill slot backfill-2026-09-23-8)

Model-drafted, unattended, by the cloud research routine. Nothing here is human-verified. Scope: only the three checks added on 2026-09-23 (WATCHLIST_PLAN.md §25) — S8 management credibility, S2 capital allocation, S3 earnings quality. Every other stage stands as written in dossier-2026-09-23.md and _clusters/cluster-additions-NLY-2026-09-23.md.

Sources read this run, all free: SEC EDGAR XBRL companyfacts for cik 1043219; the 8-K Ex. 99.1 of 2026-01-28 (0001043219-26-000007, the FY2025 release, which carries the quarterly and annual GAAP→EAD reconciliations and their footnotes); the Item 5.02 8-K of 2025-12-29; and all eight earnings calls in ledger/research/_transcripts/NLY/ (FY2024-Q3 through FY2026-Q2, market data). No Equibles call and no market data call were made for this name — the 2026-09-23 pass had used Equibles for two transcripts; the archive covers eight.

Transcript guard: each of the eight files names its own fiscal quarter in its opening turn. None is mislabelled. market data renders the CEO as both "David Finkelstein" and "David L. Finkelstein", and the FY2026-Q2 file's operator turn is labelled Operator with a trailing space.

A unit caution that runs through every table below. NLY did a 1-for-4 reverse split in September 2022. XBRL CommonStockSharesOutstanding shows 1,398,240,618 at 2020-12-31 (pre-split as filed) and 364,934,065 at 2021-12-31 (post-split as restated), so the two are not comparable. Every per-share figure here is post-split, taken either from the diluted weighted-average series or from reported book value per share.


S8 — management credibility

Who was in the chair, from EDGAR

PersonRoleNote
David L. FinkelsteinCEO (and CIO)throughout the window; current
Serena WolfeCFOthroughout the window; current

The only Item 5.02 filing since 2024 is 2025-12-29 (0001193125-25-335363): the Chief Legal Officer's transition to Senior Advisor effective 2026-01-01 and subsequent retirement. No CEO or CFO change in the window, so there are no departed rows — every promise below belongs to the people still making them.

What this company actually promises, which is the first finding

Annaly gives no revenue, EPS, margin, book-value or full-year guidance of any kind. Eight consecutive calls were read in the §3.1 order and the forward statements management makes are of three narrow types: (a) whether the dividend it has itself just set will be covered next quarter, (b) leverage staying inside a stated range, and (c) the criterion it applies to raising equity. §25 asks for four checkable promises; there are six, so this stage does not close as partial — but the owner should read the verdict knowing that the longest horizon any of these promises runs to is one quarter, and the variable that has decided this holder's return over five years, book value per share, is not something management has ever promised.

The say-vs-do table

#Promise (metric, period, figure)Made by / whereResult / whereVerdict
1The dividend, raised from $0.65 to $0.70, will be maintained for the rest of 2025 — "Given our outlook for 2025, we expect to maintain this level for the remainder of the year all else equal"Serena Wolfe, CFO — FY2025-Q1 call, 2025-05-01$0.70 paid in Q2, Q3 and Q4 2025 (each quarter's call states it), then raised to $0.75 in Q2 2026met
2Dividend covered and potentially out-earned for the remainder of 2025 — "we've outearned the dividend and we expect to certainly cover and potentially out earn the dividend for the remainder of the year, all else equal"David L. Finkelstein, CEO — FY2025-Q2 call, 2025-07-24EAD $0.73 (Q3 2025) and $0.74 (Q4 2025) against a $0.70 dividend — 8-K Ex. 99.1 of 2026-01-28beat (out-earned, not merely covered)
3Q1 2026 will out-earn the dividend — "We expect outearned certainly in Q1, and we feel like the dividend is safe here"David Finkelstein, CEO — FY2025-Q4 call, 2026-01-29EAD $0.76 in Q1 2026 against a $0.70 dividend — FY2026-Q1 call, 2026-04-22beat
4The portfolio should generate an economic return approximating or above the dividend — a standing commitment stated in the same turn as row 2David L. Finkelstein, CEO — FY2025-Q2 call, 2025-07-24Full-year 2025 economic return 20%+ (FY2025-Q4 call) against a dividend of $2.80 on an opening book of $19.15 = 14.6%beat
5Equity is raised only when it is accretive — management's stated first criterion is not to issue below book (the criterion the 2026-09-23 pass wrote into the kill criteria)management, standing policyEstimated average issue price against year-end book: 2023 ~$21.21 vs $19.44; 2024 ~$19.91 vs $19.15; 2025 ~$22.63 vs $20.21 — above book in each of the last three years (see the arithmetic and its caveat in S2 below)met
6Economic leverage kept conservative and inside the stated ceilingmanagement, standing policy (10:1 ceiling per the S0 lens)5.7x (2023), 5.5x (2024), 5.6x (2025), 5.7x (Q1 2026), 5.6x (Q2 2026) — FY2025 10-K and Q2 2026 10-Q tables, carried in dossier-2026-09-23.mdmet

Scored: 6 rows — 3 beat, 3 met, 0 missed, 0 withdrawn. Every row belongs to executives still in the chair.

One promise could not be scored and is recorded rather than guessed. On the FY2024-Q3 call (2024-10-24) Finkelstein said the company had committed to one bulk MSR transaction of $125m in market value "which we expect to close before year-end". The FY2024-Q4 call reports the MSR portfolio ending the quarter higher and announces a new $385m of MSR anticipated to settle in Q1 — it never confirms the $125m closed. The portfolio roll is consistent with it having closed; that is an inference, not evidence, so the row is left unscored.

Two things the table does not show, which belong beside it

The dividend was cut 26% in 2023 — just outside the scored window. $0.88 to $0.65 a quarter (post-split), recorded in the 2026-09-23 pass. §25 scores the last 4–8 quarters and the eight archived calls begin in Q3 2024, so the cut is not a row; it is the one capital-return commitment this company has broken in recent memory and the reason rows 1–3 read as management rebuilding a record rather than extending one.

The promises are about a number management sets. Rows 1, 2 and 3 are all versions of will we cover our own dividend next quarter. A company that sets its dividend below its own distributable earnings and then reports that it covered the dividend has met a bar it chose. It is still worth scoring — a nine-quarter coverage streak is a fact and the streak is real — but it is not the same kind of promise as a revenue or margin guide, and the verdict below says so.

Dodge log — carried forward, one line added

The 2026-09-23 pass logged two dodges (Q2 2026 KBW on dividend coverage versus the required ROE; Q1 2026 Jones Trading on the 190bp book-value attribution) and noted that neither topic had repeated. This run re-read both calls in the §3.1 order and confirms both entries as written.

S8's roll-up of the log, and it matters more after this run's S3 work: both dodged questions ask management to decompose a reported number — the portfolio's own return against the return it needs, and what moved book value. The earnings-quality section below finds that the largest single component of EAD, swap carry, fell about 40% in one year while EAD rose. A management that declines to decompose its return is harder to take at face value on a measure whose composition is changing that fast. That is a reason to keep the two dodges on watch, not a third dodge.

Verdict

conservative — 6 of 6 met or beaten, 3 of them beaten, with no departed rows. Under §25's wording ("beats or meets nearly every first guide") that is the correct label, and it is recorded with two qualifications that a reader should not skip:

  • The promise set is unusually thin and short-horizon. No annual guidance of any kind exists on this name; the longest promise runs one quarter; three of the six rows are about covering a dividend management itself sets.
  • The one thing a holder needs promised is never promised. Book value per share went $31.88 (2021) → $20.21 (2025) while the dividend was paid on time every quarter. S2 below puts a number on what that combination was worth.

Because the verdict is conservative and not over-promises, the S7 valuation table's base case is not haircut for management's guide (§25).


S2 — capital allocation and returns

§25 directs mortgage REITs to return on equity and book value per share compounded with dividends rather than to ROIC. Both are below. Figures from XBRL companyfacts (cik 1043219) unless marked; book value per share from the quarterly releases as carried in dossier-2026-09-23.md.

Where the cash went, and where it came from

$m202120222023202420255y total
Common dividends paid321.1412.1325.1375.9494.91,929.1
Common stock issued552.22,636.0673.71,558.32,910.68,330.8
Buybacks——————
Operating cash flow3,076.85,372.42,367.23,310.7692.914,820.0
Diluted weighted shares (m)357.1411.6494.5522.7641.0+79.5%

Plus $955.7m of common equity raised in H1 2026 (2026-09-23 dossier) and 740.3m diluted weighted shares at Q2 2026 — 2.09x the 353.7m of 2020.

The single most compact statement of this company's capital allocation: over five years it paid $1.93bn of common dividends and sold $8.33bn of common stock, and it has not repurchased a share since 2020. Operating cash flow is shown for completeness and is not the constraint — for a levered MBS book it is dominated by portfolio flows, not by earnings.

Book value per share, compounded with dividends

20212022202320242025Q2 2026
BVPS31.8820.7919.4419.1520.2120.15
Economic return (BVPS change + dividends)—≈ −24%+6.0%+11.9%+20.2%+6.9% (H1)

Over 2022–2025: book value per share fell $11.67 against $11.52 of dividends paid — a four-year economic result of about −$0.15 a share, i.e. zero, on a starting book of $31.88 (the arithmetic the 2026-09-23 pass wrote into S0; this run re-read the inputs and did not change it). Over 2023–2025 the same measure is +6.0%, +11.9%, +20.2% and compounds to roughly +42%.

Which window you choose decides the verdict, so both are stated. 2022 — the year the Agency basis blew out and book fell 35% — is the whole of the difference.

Return on equity

Common equity taken as reported BVPS × shares outstanding; net income attributable to common from XBRL.

2022202320242025
Net income to common ($m)1,614.7(1,784.8)847.41,869.3
Average common equity ($m)10,6859,72910,39812,682
GAAP ROE15.1%(18.3)%8.2%14.7%
EAD return on average equity (company's own, 8-K of 2026-01-28)——13.28%14.47%

Four-year average GAAP ROE: 4.9%. The company's own EAD return on equity printed 13–14% in both years it is disclosed for. That gap is the subject of the earnings-quality section below.

Issuance: at what price, and was the stated discipline kept

Shares outstanding and issuance proceeds give an implied average issue price. This is an estimate, not a disclosure — the share-count change also absorbs stock compensation and any dividend reinvestment, both of which would pull the implied price slightly down — so it is stated as an approximation and the direction of its error is stated with it.

202320242025
Common stock issued ($m)673.71,558.32,910.6
Increase in shares outstanding (m)31.7778.28128.61
Implied average issue price~$21.21~$19.91~$22.63
Book value per share, year-end19.4419.1520.21
Implied price ÷ book~1.09x~1.04x~1.12x

Management's stated criterion was kept in each of the last three years, and that is the part of this record that deserves saying plainly: equity sold above book raises book value per share for the people already holding it. It is also why the S0 lens's dilution test — scored 0 on a share count that doubled — understates the picture: the count doubled, and it doubled at prices above book.

The counter-fact, equally plain: no share was repurchased while the stock traded below book for most of 2022–2024, and the company has no repurchase programme mentioned in its FY2025 release. The only lever used has been the one that grows the balance sheet.

Verdict: neutral

  • Not compounds: a four-year economic result of approximately zero (2022–2025), a book value per share still 37% below its 2021 level, a share count at 2.09x, a 26% dividend cut in 2023, and a four-year average GAAP return on common equity of 4.9%.
  • Not destroys: the last three years have compounded at roughly +42% on the same measure; the dividend has been covered by EAD for nine consecutive quarters; leverage has been held at 5.5–5.8x against a 10:1 ceiling; and every equity raise in the last three years was struck above book, which adds to book value per share rather than diluting it.

On the 2021–2025 window this reads destroys; on 2023–2025 it reads compounds. The honest label across the period §25 asks for is neutral, and the verdict turns entirely on whether the post-2022 recovery is durable — which is precisely the question S7 left open for the owner. Per §25 a neutral verdict adds no kill criterion and does not change S2's glyph.


Earnings quality

§25's sixth test names the check for this sector directly: the company's distributable ("available for distribution") earnings against GAAP. That is the spine of this section. The bridge read is the one in the 8-K Ex. 99.1 of 2026-01-28, which carries both the quarterly (Q4 2025 / Q3 2025 / Q4 2024) and the annual (FY2025 / FY2024) reconciliations with footnotes.

Three flags.

  1. EAD has exceeded GAAP by half over five years, and the dividend has been paid at more than twice GAAP earnings.
per diluted share202120222023202420255y
EAD4.654.232.862.702.9217.36
GAAP net income to common6.393.92(3.61)1.622.9211.24

EAD exceeds GAAP by $6.12 a share, 54%, over the five years — and essentially all of the gap is 2023, where EAD was +$2.86 against GAAP of −$3.61, a $6.47 difference in one year. Over 2022–2025 the dividend was $11.52 a share against GAAP earnings of $4.85 a share — 2.4x. The difference did not vanish; it went through book value, which is why S2's four-year economic result is zero. 2025 is a genuinely clean year and the file should say so: EAD $2.92 against GAAP $2.92, identical.

  1. The largest single component of EAD is collapsing, and its replacement does not appear in EAD. Footnote 2 to the reconciliation discloses the net interest component of interest-rate swaps — which is inside EAD even though the swaps' mark is excluded — as $716.5m for 2025 against $1.2bn for 2024, roughly −40% in one year. Against EAD of $2,024.9m (2025) and $1,564.6m (2024), swap carry was 35% of EAD in 2025 and about 77% in 2024. On the FY2025-Q1 call (2025-05-01) the CFO explained where it went: swap income fell because swaps were unwound and replaced with futures positions that do not have an interest component in EAD. So EAD rose 29% while its biggest ingredient more than halved and part of the hedge book moved to an instrument whose economics EAD does not capture. Two readings are available — genuine improvement in the underlying portfolio, or a measure whose year-on-year comparability has broken — and this run cannot separate them from the disclosure. Either way, EAD 2024 and EAD 2025 are not the same number measured the same way, and three of S8's six rows are scored on it.
  1. A "non-recurring" add-back that is the cost of the strategy, and it is growing. Transaction expenses and non-recurring items is added back in every period shown — $6.3m (Q4 2024), $8.1m (Q3 2025), $7.2m (Q4 2025); $20.3m (FY2024) → $27.8m (FY2025), +37%. Footnote 4 says what it is: costs incurred in connection with securitizations of residential whole loans — the running cost of the OBX shelf, which management is deliberately growing ("we expect to grow the OBX platform", FY2025-Q1 call). A cost that scales with a business you are expanding is recurring by definition.

Clean, with the number:

  • The reconciliation itself is complete and unusually legible. Every adjustment is named, footnoted and quantified in the same exhibit, including the two offsetting mark-to-market lines that dominate it — in Q4 2024, $2,010.7m of investment losses added back against $1,958.8m of derivative gains removed, which is the hedge working and is disclosed as such.
  • Auditor and filing record: Ernst & Young since 2012, unqualified ICFR opinion, no NT filings, no comment letters since 2021 (2026-09-23 pass).
  • Leverage and funding are disclosed at the level the test needs: economic leverage 5.5–5.8x since 2023, 37 repo counterparties, and the company's own sensitivity table (+25bp MBS spread = −8.7% of NAV).

Tests recorded as not meaningful rather than skipped. §25's cash-conversion, accruals, working capital and capitalised-cost tests are written for an operating company. For a levered agency-MBS book, operating cash flow is dominated by portfolio purchase and paydown flows — $14.8bn over five years against $9.9bn of cumulative GAAP net income — and an accruals ratio over total assets of roughly $110bn would be arithmetic without meaning. Stock-based compensation is not separately tagged in this registrant's XBRL. The sector-specific test in §25 test 6 is the one that applies, and it is done above.

EQ summary line for the scorecard: EQ: 3 flags — EAD exceeded GAAP by $6.12/sh (54%) over five years and the 2022-25 dividend was 2.4x GAAP earnings; swap carry inside EAD fell from ~$1.2bn (77% of EAD) to $716.5m (35%) in one year while the replacement futures carry no EAD interest component; securitisation costs added back as "non-recurring" in every period and up 37% to $27.8m. 2025 is clean — EAD $2.92 = GAAP $2.92. Cash-conversion and working-capital tests not meaningful for a levered MBS book.


One item proposed to the owner and deliberately NOT written into the scorecard

This run would add a seventh kill criterion — EAD falling below the dividend in any quarter in which swap carry is under $500m annualised, i.e. the point at which the fading swap component stops being covered by asset yield. It is not added, because the backfill slot's own note limits new kill criteria to an over-promiser or a capital destroyer and NLY is neither (conservative / neutral). It is raised here and in the run report so the owner can add it if they want it.


What this run changed on the scorecard

  • S8 ⬜ → ✅ conservative — 6 rows: 3 beat, 3 met, none missed, none withdrawn; no departed rows. ⚠️ the promise set is thin and one quarter long, and book value is never promised.
  • S2 keeps its ✅ glyph and gains the capital-allocation table and the verdict neutral.
  • S3 keeps its ✅ (gate-minimum) glyph and gains the EQ: summary above.
  • Dodge log unchanged — re-read and confirmed, with an S8 roll-up note.
  • S7 valuation base case is not haircut: S8's verdict is conservative.
  • ⚠️ failed-gates line and ➡️ Next action updated with the swap-carry finding.
Full notes, 23 Sep 2026 14 min read

NLY — Annaly Capital Management, Inc. — dossier, 2026-09-23

Run by hand in a session on the owner's request ("run annaly capital via pipeline skill and add it to research watchlist"), following .claude/skills/stock-sourcing-pipeline/SKILL.md. It covers the name-pass stages of slot names-additions-2026-09-23-1, for NLY only. Model-drafted; nothing here is human-verified. Sourced facts carry the document they were read from; [background] marks background knowledge and [search-summary] marks search-result evidence. Private: this directory never reaches dist/.

Industry stages (S0, S4, S6, S7) are in _clusters/cluster-additions-NLY-2026-09-23.md.

Access this run

  • threespread.py check: OK — key present, HTTP 200 from api.3spread.com.
  • SEC EDGAR: reachable with the User-Agent in tools/edgar.py.
  • Equibles: reachable. 3 calls spent (Q2 2026 transcript, Q1 2026 transcript, short interest).
  • market data: prices (1 call). Next-earnings calendar not called (paid plan, per earlier runs).
  • WebSearch: 2 queries.

§1 — S1 Source

watchlist (owner, 2026-09-23 addition, WATCHLIST_PLAN.md §24). The row was added to watchlist.yaml earlier on 2026-09-23 with CIK 1043219 and cluster mortgage_reit. No Tier 0 promotion. Board cross-reference: one claim row, Chris Whalen, 2026-09-19, bullish (S6 in the cluster file).

§2 — S2 Kill test

An mREIT has no revenue line in the usual sense, so the kill test is read on the four series that decide one: earnings power (EAD and net interest spread), book value, leverage and funding, and the share count.

20212022202320242025Q2 2026
Interest income $bn (XBRL)1.982.783.734.845.961.81 (qtr)
GAAP net interest income $bn1.731.47−0.110.251.140.49 (qtr)
Economic NII ex-PAA $bn——1.481.491.940.59 (qtr)
Net income to common $bn2.281.61−1.780.851.870.78 (qtr)
EAD / share4.654.232.862.702.920.79 (qtr)
Net interest spread ex-PAA1.89%1.70%1.32%1.22%1.40%1.50%
BVPS31.8820.7919.4419.1520.2120.15
Economic leverage——5.7x5.5x5.6x5.6x
Diluted wtd shares m357.1411.6494.5522.7641.0740.3

Sources: XBRL companyfacts (interest income, net interest, net income to common, diluted shares); FY2023 10-K and FY2025 10-K Net Income (Loss) Summary (EAD, spread, BVPS, leverage); Q2 2026 10-Q same table. 2020 diluted shares 353.7m (XBRL, split-adjusted). Dashes: not extracted this run.

Share count over five years: 353.7m (2020) → 740.3m (Q2 2026 diluted weighted), 2.09x; 753.7m outstanding at 2026-07-24 (10-Q cover). The only repurchases in the XBRL record are 2019–2020 ($223.6m, $209.4m); none since. Common issued for cash $2.64bn / $0.67bn / $1.56bn / $2.91bn (2022–2025, ProceedsFromIssuanceOfCommonStock) and $955.7m in H1 2026 through the ATM.

Funding and maturity wall. No term-debt wall in the corporate sense: the balance sheet is funded by repo (33-day weighted average maturity at Q2 2026, per the CFO on the Q2 call; 35 days at 2025 year-end, FY2025 10-K) across 37 counterparties, plus non-recourse securitisation debt and MSR/whole-loan warehouse lines ($8.3bn capacity, $2.8bn committed, Q2 call). The maturity wall is every month. Liquidity: $8.0bn unencumbered assets including $5.5bn cash and unencumbered Agency MBS; $9.6bn available for financing (Q2 call, CFO).

FCF is not the measure for a REIT that must distribute; the relevant coverage is EAD over the dividend, above it for nine consecutive quarters (Q2 2026 10-Q).

Analyst coverage count: analysts from ten firms asked questions on the two calls (KBW, Piper Sandler, UBS, BTIG, Green Street, Compass Point, J.P. Morgan, Citizens JMP, RBC, Jones Trading); a formal count was not pulled.

Survives or dies: survives. Leverage is low for the model and stable, liquidity is 57% of capital, EAD covers the dividend, and the auditor (Ernst & Young, since 2012) gave an unqualified ICFR opinion (FY2025 10-K). ⚠️ What survives is a company, not a per-share compounder: the five-year record is a halved book per share refilled by a doubled share count (cluster file §1). That is S0's finding and S8's problem; it is not a kill.

§3 — S3 Filings (gate-minimum)

Read: FY2025 10-K Item 1A (summary and GSE factor) and Item 7 (overview, net income summary, rates table); Q2 2026 10-Q Item 2 MD&A (business environment, results table, ATM paragraph) and Item 3 (rate and MBS spread sensitivity, counterparty table); 2026 DEF 14A ownership table and compensation summary; 8-Ks of 2025-12-22, 2025-12-29, 2026-07-21 (results cover) and 2026-08-19; the 424B5 of 2025-12-22; the EDGAR filing index for red flags; Q1 and Q2 2026 call transcripts.

The three biggest risks management discloses, in its own words

  1. Leverage. > "Our strategy involves the use of leverage, which increases the risk that we may incur substantial losses." > — FY2025 10-K, Item 1A, Summary of Risk Factors, first bullet under Risks Related to Our Liquidity and Funding

Sized by the company's own table: a +25bp MBS spread shock is −8.7% of NAV, a +75bp rate shock −3.9% of NAV and −5.1% of economic net interest income (Q2 2026 10-Q, Item 3).

  1. Spread compression. > "An increase in the interest payments on our borrowings relative to the interest we earn on our interest earning assets adversely affects our profitability." > — FY2025 10-K, Item 1A, Summary of Risk Factors, same section

This is the risk that was realised in 2023, when GAAP net interest income was −$111.4m and the dividend was cut from $3.52 to $2.60 a year.

  1. The GSE relationship. > "Any new laws, regulations or administrative actions modifying the relationship between Fannie Mae, Freddie Mac and the federal government could affect our business model or business operations." > — FY2025 10-K, Item 1A, Compliance, Regulatory & Legal Risks

Live in 2026 in the helpful direction (the GSEs' $200bn purchase mandate tightened the basis); the same factor cuts the other way if the mandate ends or the conservatorship changes form.

Fourth, recorded because the case rests on the dividend:

"We have not established a minimum dividend payment level and cannot assure stockholders of our ability to pay dividends in the future." — FY2025 10-K, Item 1A, Risks of Ownership of Our Common Stock

Latest-quarter delta (Q2 2026 10-Q MD&A)

  • EAD $0.79 (from $0.76 in Q1), dividend raised to $0.75 from $0.70. > "This marked the ninth consecutive quarter in which EAD surpassed the dividend." > — Q2 2026 10-Q, Item 2, Business Environment
  • BVPS $20.15 (+1.7% q/q); economic return 5.5% in the quarter, 6.9% for the half.
  • Portfolio $139.2bn at period-end against $107.5bn a year earlier (+29.5%); average equity +25.4%.
  • GAAP leverage 7.4x, economic leverage 5.6x.
  • Capital mix: Agency 57%, residential credit 22%, MSR 21%.
  • ATM: 20.3m shares / $447.2m in the quarter, 42.6m / $955.7m in the half.
  • The mortgage basis tightened 106bp → 94bp over the quarter (MD&A rates table).

Capital actions on file

  • ATM, $2.5bn, 13 distribution agency agreements dated 2025-12-22, replacing the May 2025 program (8-K 2025-12-22; 424B5 2025-12-22). $955.7m used in H1 2026, so roughly $1.5bn of capacity was left at 2026-06-30 if the new program carried all of it (the 10-Q does not split usage between the old and new programs — not established).
  • Series I preferred redemption: all 17,700,000 shares of the 6.750% Series I at $25.00, payable 2026-10-01 (8-K 2026-08-19) — $442.5m. The funding source is not stated in the 8-K.
  • Series J 8.875% fixed-rate preferred issued in 2025, which the 10-K calls "the first sizeable non-rated preferred stock issuance in several years" (quoted in full below). > "Given strong investor demand for mortgage REITs, Annaly was able to raise $2.6 billion in common equity capital through its at-the-market sales program at accretive levels and issued a Series J preferred stock in what marked the first sizeable non-rated preferred stock issuance in several years." > — FY2025 10-K, Item 7, overview
  • No buyback authorisation in use; no repurchases since 2020 in XBRL.

Proxy skim (2026 DEF 14A)

  • Insider ownership: all executive officers and directors as a group (11 people) 1,070,909 shares, marked "*" (under 1%) — about 0.14% of 753.7m. 5% owners on the proxy's (older 13G) basis: BlackRock 56,856,178 (7.8%), Vanguard 49,300,019 (6.7%).
  • What the bonus is paid on: a corporate scorecard (75%) plus individual objectives (25%) for the annual cash incentive; the scorecard includes Tangible Economic Return and EAD return on equity, with relative metrics requiring above-median (55th percentile) performance for target; PSUs vest on multiple company performance metrics over three years (proxy compensation summary). Paying on economic return, not on EAD alone, is the right incentive for the S0 test 1 problem.

Insider pattern (3spread insiders --days 365, Form 4 XML read directly)

46 Form 4 filings since 2025-09-23; the open-market transactions in the ones read:

DateInsiderCodeSharesPrice
2026-02-18David Finkelstein, CEO and Co-CIOS50,000$23.13
2026-02-18Steven Campbell, President and COOS28,225$23.12
2026-02-18Serena Wolfe, CFOS16,536$23.14
2026-02-25Steven CampbellS26,491$22.83
2026-04-30David FinkelsteinS50,000$22.88
2026-05-05Serena WolfeS16,537$22.48
2026-08-04David FinkelsteinS50,000$22.89

Plus annual award grants (code A) and tax withholding (F) on 2026-02-01 and 2026-02-25, and director grants after the 2026-06-10 annual meeting. No open-market purchase in any filing read. The CEO held 761,012 shares after the 2026-08-04 sale; his three 2026 sales total 150,000. The evenly sized, spaced 50,000-share sales look like a plan (a Form 144 was filed 2026-08-03), but no 10b5-1 footnote was checked — not established.

Red-flag sweep

  • No NT 10-K / NT 10-Q in the filing index; one 10-Q/A in 2018; four 8-K/As.
  • SEC comment-letter correspondence (CORRESP/UPLOAD): last exchange 2021-07-06; none since.
  • Auditor Ernst & Young LLP, "since 2012"; unqualified ICFR opinion (FY2025 10-K).
  • Executive change: Anthony Green, Chief Legal Officer, Chief Corporate Officer and Secretary, retired — Senior Advisor from 2026-01-01 to 2026-03-31 (8-K 2025-12-29). An orderly, disclosed succession event; no other departures found in the 8-K index for 2025–2026.
  • Going concern: not applicable on these numbers; not searched on Equibles to save budget.

Transcripts — Q1 2026 (2026-04-22) and Q2 2026 (2026-07-22), Equibles, read Q&A-first

Guidance and commentary changes (CFO and CEO prepared remarks):

  • Q1: EAD $0.76, BV −1.9% to $19.82, economic return 1.5%; ~$510m raised on the ATM (the 8-K says $509m, per Equibles' correction note); the offset to EAD was lower swap income.
  • Q2: EAD $0.79 on a higher Agency coupon (5.11%, +11bp) and lower repo (3.84% average); the offset again lower swap income. OPEX-to-equity up 11bp to 1.40%, which the CFO expects to moderate. Dividend raised to $0.75; the CEO says the board stress-tested it and expects to earn it over time, with some quarters possibly slightly below it (paraphrased).
  • Long-term capital allocation target reaffirmed in Q1: 50% Agency / 30% residential credit / 20% MSR.

What the analysts asked about (by topic):

  • Equity issuance — both calls (J.P. Morgan, Rick Shane in Q1; J.P. Morgan again in Q2). The street's recurring question is how hard the ATM will be pushed at a small premium to book.
  • Dividend coverage vs the required return — Q2 (KBW, Boz George; RBC, Kenneth Lee).
  • GSEs and policy — Q1 (Citizens JMP: further government action on mortgage rates), Q2 (Piper Sandler: FHFA/GSE purchase behaviour).
  • Bank capital re-proposal and MSR risk weights — Q1 (Piper, KBW).
  • Residential credit sourcing, returns, delinquencies — both calls (UBS, BTIG, Citizens JMP).
  • Hedging mix, swaps vs Treasuries — Q1 (Compass Point).
  • Book value quarter-to-date — both calls (KBW): +4% economic return QTD as of mid-April; book off a little over 1% QTD as of mid-July.

Dodges:

  • Q2 2026 · dividend coverage vs required ROE · Boz George (KBW). Asked for the portfolio's economic return relative to the ~15% ROE needed to cover the raised dividend. The answer gave the new-money return ranges by strategy from the supplement and said the board is methodical and the dividend earnable — it did not give the portfolio-level return against the required one. Locator: Equibles Q2 FY2026 transcript, second analyst exchange, first Q&A question block.
  • Q1 2026 · book value decline attribution · Jason Weaver (Jones Trading). Asked to split the 190bp book value decline between Agency spread widening and marks on resi credit/MSR, and whether it had reversed in April. The answer ranked the three businesses qualitatively and gave ranges for resi and MSR, but no split of the 190bp and no answer on the April reversal beyond the earlier +4% QTD figure. Locator: Equibles Q1 FY2026 transcript, Jones Trading exchange.

One dodge on each topic; neither repeats yet. Both go into the scorecard's dodge log. The question that did repeat across both calls — how much equity will be issued — was answered both times (price-to-book must be accretive, assets attractive, a light footprint).

§4 — S5 Ownership

13F at 2026-06-30 (threespread.py holders NLY): BlackRock 87,219,586 sh (+5,485,296); State Street 29,426,379 (+1,272,235); Invesco 7,724,098 (+969,593); Citadel 7,346,986 (+2,917,722); T. Rowe Price 7,176,291 (−11,282,820, the largest reduction); Charles Schwab IM 5,427,809; Ameriprise 5,030,245 (−2,094,971); BNY Mellon 4,801,781; KBC 4,582,170; D. E. Shaw 4,320,185 (−1,929,940).

  • ⚠️ Ingestion incomplete: 537 filers for Q2 against 765 for Q1. Vanguard's three entities (64.0m + 30.8m + 4.4m sh at Q1), Geode (20.0m), FMR (14.2m), Allspring (13.5m) and others have no Q2 13F ingested yet — these are not exits.
  • ⚠️⚠️ The Sixth Street row (5,427,809 sh, "NEW") is not a holding. It carries exactly Charles Schwab Investment Management's share count; 3spread flags the duplicate. This is the same filing defect documented in ACGL/dossier-2026-09-22.md §5 and seen again on KNSL. Not carried.

13D/G census (threespread.py owners NLY): passive 13G/13G-As only — the most recent a 13G for the period ending 2026-03-31 (filed 2026-04-29) and a 13G/A for 2026-03-13 (filed 2026-03-26, filer-agent prefix 0000102909, which is Vanguard's CIK [background]). No 13D, no activist. Charter caps ownership at 9.8% (FY2025 10-K, Item 1A).

Short interest: 20.66m shares (2026-08-31, FINRA via Equibles), 2.7% of shares outstanding, 3.8 days to cover. Stable at 16.6–23.5m since mid-June.

Shelf / ATM / buyback: the $2.5bn ATM is the thing hanging over the stock, by design — it is how the company grows — and it was drawn at a rate of $447–509m a quarter in H1 2026. No buyback. Absence of a buyback in the filings read is not proof none exists; the XBRL record shows none since 2020.

Who is on the other side: index and passive money (BlackRock, Vanguard, State Street), income funds, and a sell-side-covered retail base. Insiders own 0.14% and have been net sellers in 2026. The one active manager making a big move, T. Rowe Price, cut 61% of its position in Q2.

§5 — S8 Valuation & sizing (model; buy price and size ⏳ OPEN (user))

An mREIT is valued on book and yield, so the model is book value × a price-to-book multiple, with the dividend as the carry. Two-year horizon (to Q2 2028).

CaseBasis over two yearsBVPS mid-2028P/BValueDividends received (2 yrs)Total vs $21.44
BearGSE bid ends, basis back to ~126bp (+32bp); book −11% at the company's own sensitivity, then a dividend cut to $0.6017.900.90$16.15.10−1%
BaseBasis range-bound; EAD ≈ dividend; book flat20.151.05$21.26.00+27%
BullMix shift works: resi credit and MSR carry the book through; basis stable; modest accretion21.501.15$24.76.00+43%

Inputs: BVPS $20.15 and dividend $0.75/qtr (Q2 2026 10-Q); sensitivity −8.7% NAV per +25bp (10-Q Item 3), extrapolated linearly to +32bp — an extrapolation the table does not support beyond 25bp; bear P/B 0.90 is Chris Whalen's stated cost basis on the board's claim row, used as a reference, not a forecast; bull 1.15x is an assumption. Current: 1.06x book, 14.0% yield, ADV $183.1m/day.

Read the table the way the cluster file does: in the base case the return is the dividend, and the price is flat. The bear case is roughly break-even over two years only because the dividends cover the book loss, which is 2022 again at a smaller size.

Proposed buy price ⏳ OPEN (user): at or below 1.0x the latest reported book (~$20.15 today) with EAD still at or above the dividend — i.e. the stock pays you 14%+ and you are not paying a premium for the ATM machine. Whalen's 0.9x is the level at which a 2022-sized book drawdown is already in the price.

Proposed size ⏳ OPEN (user): liquidity is not the constraint (ADV $183m); the constraint is correlation — this is a levered bet on the MBS basis, and a position here should be sized as a basis trade, not as an equity compounder.

§6 — S9 Monitoring

  • Next earnings: ~2026-10-21 (DERIVED from cadence, not from a calendar): Q3 2025 results were released 2025-10-22, Q2 2026 on 2026-07-21, Q1 2026 on 2026-04-21 — a Tuesday/Wednesday in the third week after quarter-end. The call follows the next morning.
  • Filing feed: Tier 0 EDGAR sweep, cik: 1043219 on the watchlist row.
  • Leading indicator between quarters: the current-coupon MBS basis, which Annaly itself publishes every quarter (94bp at 2026-06-30), and the GSE mandate's remaining capacity (~$45bn of $200bn settled through May per the CEO). The basis moves book value before any filing does; the mandate says how long the buyer holding it stays.
  • Second indicator: the Series I redemption on 2026-10-01 ($442.5m) — how it is funded (ATM, cash, a new preferred) shows up in the Q3 10-Q's capital note.
  • Trigger that converts it to a buy: price at or below the latest reported BVPS while EAD is at or above the dividend and economic leverage is ≤6.0x. At 1.06x today the trigger has not fired.
  • Trigger that kills it: any of the four kill criteria in the cluster file §5 — BVPS < $18.45, EAD < dividend two quarters running, economic leverage > 7.0x, or ATM issuance below book.