# gap-series-mstr.csv: method Strategy (MSTR), one row per weekly 8-K, reported dates 2025-01-06 to 2026-11-09. 85 rows. Built 2026-11-09. Companion files: `mstr-convert-principal.csv` (convert principal by date, per series, from the filings) and `mstr-convert-moneyness.csv` (the dates MSTR's daily close crossed each conversion price). All three sit beside this note. ## The construction ``` claims_usd(t) = debt + preferred - cash - strc_holdings, from the latest MSTR Snapshots row dated on or before t. A step function, never interpolated. claims_btc(t) = claims_usd(t) / btc_price(t) + btc_claim_static BPS_sats(t) = cumulative_btc(t) / basic_shares(t) * 1e8 CEBE_sats(t) = (cumulative_btc(t) - claims_btc(t)) / basic_shares(t) * 1e8 gap_sats(t) = BPS_sats - CEBE_sats, which reduces to claims_btc / basic_shares * 1e8 gap_usd(t) = claims_usd(t) / basic_shares(t), which needs no Bitcoin price gap_expansion = the change between consecutive rows ``` BPS here is basic Bitcoin per share. No floor is applied: a per-share quantity is not clamped, because clamping would hide a net cash issuer. MSTR is never net cash in this window. ## The split Every row splits its gap into two parts that sum to it exactly, in sats and in dollars. - **Convert treatment** (`gap_convert_*`): the convertible notes, which CEBE holds as claims at face and fully diluted Bitcoin per share counts as shares, plus STRK. STRK is a convertible preferred (0.1 class A share per share, a $1,000 conversion price). Treating it with the notes is a judgement: fully diluted Bitcoin per share counts it as shares, so it is grouped where its treatment is. - **Claims fully diluted Bitcoin per share cannot see** (`gap_unseen_*`): the non-convertible preferred (STRF, STRC, STRD, STRE) and every other senior claim (the other long-term secured debt), plus any Bitcoin-denominated claim (`btc_claim_static`, zero for MSTR). **Cash netting: pro rata.** Cash and STRC holdings net against both parts in proportion to their gross face, so `gap_convert = gap x (notes + STRK) / (debt + preferred)` and `gap_unseen` is the rest. The convert principal on a row is the principal outstanding at the date of the snapshot that row reads, so the split moves on the same step function as the claims and never exceeds the debt it sits inside. Event dates (issue, conversion, repurchase) are in `mstr-convert-principal.csv`. The convert part is measured as claims at face. It is not the dilution fully diluted Bitcoin per share applies, which depends on the conversion share counts rather than the face. ## The price rule Every row is priced at the site's Bitcoin close for its own reported date: the close for date D is the CoinGecko print at 00:00 UTC on D+1, served at `/api/btc-close?date=D`. No price is ever carried forward. 76 rows read the site's archive of captured closes, 8 read the live route (6 of them younger than the 45-day settle window, which can still restate by a few dollars), and 1, the 2026-11-09 row, has no close yet: it strikes at 00:00 UTC on 2026-11-10, so that row's sats figures are blank and its dollar figures are complete. The Yahoo BTC-USD daily close for the same date is carried beside it in `btc_price_alt` and is not used for any figure. The site's archive now reaches back past the window's first row, so no row needs the earlier back-period fallback. ## Shares The transactions tab states `basic_shares_derived` in thousands. `basic_shares_derived_thousands` carries it as stated and `basic_shares` carries the count used, so the factor of 1,000 is visible. Snapshots rows state shares in units and leave the thousands column blank. ## Where each row comes from | row_source | rows | dates | holdings and shares from | |---|---|---|---| | sheet transactions tab | 70 | 2025-01-06 to 2026-07-20 | the ledger row (counter 50 to 119) | | sheet snapshots weekly row | 14 | 2026-08-03 to 2026-11-02 | that week's Snapshots row, dated at the 8-K's as-of date or filing date | | held filing row | 1 | 2026-11-09 | the 8-K filed 2026-11-09 (acc 0001193125-26-452693), not yet on the Sheet; `held_row` TRUE | `reported_date` is the 8-K filing date on the Snapshots and filing rows, and the ledger's own date on ledger rows. Four ledger rows sit on a date other than their 8-K's and say so in `note`. Weekly 8-Ks no row covers, listed and not filled: 2025-02-03, 2025-02-18, 2025-03-03, 2025-03-10, 2025-04-07, 2025-07-07, 2025-07-28, 2025-10-06, 2025-12-22, 2026-03-30, 2026-05-04, 2026-05-26 and 2026-07-27. Every one reports no Bitcoin purchase. Three report sales of MSTR stock, so the share count moved in a week with no row: 2025-12-22 (4,535,000 shares), 2026-05-04 (492,210) and 2026-07-27 (5,429,160). ## Inputs, each banked in the repository's MSTR provenance folder `gap-converts-2026-11-09` with its URL and fetch time | input | sha256 | |---|---| | Snapshots tab, live public read 2026-11-09T13:54:20Z | `5cb1db80d9bfa1a2d5f52f1e52242eaefb46cbe686b2c04b43469cb8312a67bf` | | Instruments tab, live public read 2026-11-09T13:55:29Z | `3c65957518b6dbe5d0e4c13ec8da829618d012afe96f7a0e7721f9a318be925a` | | Transactions tab, the repository's tracked copy (70 MSTR rows; the live tab has no public read) | `ece4db77f8427c538aff47ac1a4a3da1e3f30a806875d8e58ea35bffb0bde097` | | The 2026-11-09 filing row (snapshots shape) | `b0952c8f8d4ba34f5b076e43460f8abc9951e89b9c57cf79070cf743acf508e4` | | The 2026-11-09 filing rows (deals shape) | `19d20df45999c63cca21af4097302a41cba46c8b84d5e6da0aad462915f4ef9b` | | Site Bitcoin closes for every row date | `07fcf776bec845901b8a6df140e3f4b97a3e4427e492d323c413558e251def01` | | Yahoo BTC-USD daily closes, 2024-12-31 to 2026-11-08 | `53823b17a795fe7996cf3e9e96353879274826571d0b2e433e6dae5b6de8370a` | | MSTR daily closes, site stock price rail | `1774e0a11dc1c67cd7285708319809096b45007ebc27af203589e151323d3df5` | | EDGAR filing index for CIK 1050446 | `99beed1fc6464bd62c4002a8a583d8ea1d76aaec75015ae9fa54907a46a0d24e` | The convert principal is read from the debt note of each period filing (FY2024 10-K 0000950170-25-021814, the 2025 10-Qs 0000950170-25-063536, 0000950170-25-102209 and 0001193125-25-262568, FY2025 10-K 0001050446-26-000020, the 2026 10-Qs 0001050446-26-000031, 0001050446-26-000044 and 0001050446-26-000060) and the 8-Ks for each event. Each page and its sha256 is in that folder's manifest; each row of `mstr-convert-principal.csv` names its accession. ## Columns of gap-series-mstr.csv Money in US dollars, Bitcoin in BTC, per-share figures per basic share. A blank cell means the figure does not exist for that row, never zero. | column | meaning | |---|---| | reported_date | the row's date (see above) | | transaction_id, count | the ledger row's id and counter; blank on Snapshots and filing rows | | cumulative_btc | Bitcoin held | | basic_shares_derived_thousands | the ledger's share column as stated, in thousands; blank off the ledger | | basic_shares | basic shares used | | btc_acquired, btc_avg_price | the week's purchase as the ledger or filing row states it; blank where the row carries none | | snapshot_used, snapshot_date | the Snapshots row the claims read | | debt_usd, preferred_usd, cash_usd, strc_holdings_usd, btc_claim_static | that row's claims inputs | | claims_usd | net senior claims in dollars | | btc_price_a45, a45_print_date, a45_lag_days | the site close for the row date, that date, and the lag (always 0) | | btc_price_alt, alt_print_date | the Yahoo close for the same date, carried only | | btc_price_used, price_source | the price every sats figure uses, and where it came from | | claims_btc, bps_sats, cebe_sats, gap_sats | as in the construction | | gap_usd | gap in dollars per share | | gap_expansion_sats, gap_expansion_usd | change from the previous row; blank where either side has no figure | | note | plain-words notes on the row | | convert_principal_usd | convertible note principal outstanding at snapshot_date | | convert_series_count | convertible note series outstanding at snapshot_date | | convert_pref_face_usd | STRK face at snapshot_date | | gap_convert_sats, gap_convert_usd | the convert treatment part | | gap_unseen_sats, gap_unseen_usd | the part fully diluted Bitcoin per share cannot see | | row_source | where the row's holdings and shares come from | | held_row | TRUE on the row from the 2026-11-09 filing, not yet on the Sheet | | filing_accession | the 8-K the row rests on (the one the ledger row cites, where it cites one) | | est, est_detail | TRUE where the row's source marks an input estimated, and which | `mstr-convert-principal.csv`: date, series, class, event, principal_usd (outstanding after the event), conversion_price (dollars per class A share), source_accession, source_detail. Period end rows restate each filing's balance; event rows carry the issue, redemption, conversion and repurchase dates. A conversion the filing dates only by quarter carries the quarter end, and says so. `mstr-convert-moneyness.csv`: series, conversion_price, date, close, event, price_source. For each series, its state at the first close inside the window (or at issue), every close on the other side of the conversion price from the close before it, and its state at the last close. ## Checks run on the file - gap_convert + gap_unseen = gap on all 85 rows, in dollars and on the 84 priced rows in sats. - gap_usd = claims_usd / basic_shares, and claims_usd = debt + preferred - cash - strc_holdings, on all 85 rows. - gap_sats = BPS - CEBE = claims_btc / basic_shares x 1e8 on all 84 priced rows. - The site's rules engine (rules version 2026-11-05.2) gives the same CEBE as this file on the latest priced row, 2026-11-02, at the same price: 154,442.5 sats. On the quarter-end Snapshots rows the engine adds accrued interest from the filing to debt, which this construction does not, so it reads 19 to 93 sats lower there. ## Not the same series as the /evidence/ page The /evidence/ page's gap series is struck against fully diluted Bitcoin per share (holdings over assumed diluted shares) and prices each row at the 8-K's stated average purchase price. This file measures against basic Bitcoin per share and prices every row at the site close. Its claims also differ: it adds preferred issuance week by week from the 8-Ks, where this file steps the claims at each Snapshots row. The two figures measure different quantities and are not interchangeable.