The methodology articles on this site earn their credibility by being checkable. The thesis articles, the directional ones, spend credibility instead: they make claims about what will happen, and a claim about the future is worth nothing unless it is written down before the fact and graded honestly after.
This page is where that happens. Every directional call gets an entry with the date it was made, a resolution condition specific enough to be settled, and a status. When one resolves, it is marked hit, miss, or partial, and the miss stays exactly as visible as the hit. The point is not to be right. It is to keep a record honest enough that being wrong has a cost, which is the only thing that makes being right mean anything.
This scorecard was seeded in June 2026 from the existing thesis articles. The wording and confidence on each entry are the author's and will be revised in place as evidence arrives; revisions are noted, not hidden.
Open
Made 2026-07-07 · Resolves by 2026-10-31
Unless bitcoin recovers above Strategy's cost basis (about $75,500) or the company restores share issuance as its primary funding source, its Q3 2026 preferred distributions will again be funded at least in part by bitcoin sales.
Resolution condition: Strategy's 8-K filings through 2026-10-31 disclose further bitcoin sales used to fund preferred dividends or to replenish the USD Reserve after such payments. If the quarter's distributions are covered without bitcoin sales (for example, by resumed ATM issuance) while bitcoin stays below the cost basis, this resolves a miss. If bitcoin trades back above ~$75,500 for a sustained period, the entry is voided as untestable (the premise lapsed), not scored.
Confidence: Medium. Source: Then It Sells the Underlying. Note: the June articles that anticipated the first sale were not entered here in advance; this entry is the forward-looking claim, made before the evidence, which is the only kind this scorecard counts.
Interim check (2026-08-10): the two clauses now point opposite ways, and this note resolves neither. Share issuance is primary again (net equity against bitcoin proceeds ran 2.8x in the week to 2 Aug and 6.0x in the week to 9 Aug, 4.4x pooled), which is the entry's own escape clause. Yet weekly bitcoin sales continue: in the week to 9 Aug the $108.6m they raised went entirely to repurchasing STRC preferred, with none reaching the dividends. Distributions were covered without bitcoin sales that week, while coins were still sold to service the preferred stack. The price branch is not live: bitcoin closed below ~$75,500 on all 40 days of Q3 to date, coming no nearer than $8,979 (last close at or above the line: 2026-05-27). Company figures are coverage-reported, not read from the filings. Open; whether the voiding condition has fired, and whether the wording needs tightening, are the author's calls. Full interim mark: The Coins Now Pay a Different Bill.
Interim check (2026-09-19): the company figures in the note above have now been read at source, from the eleven weekly 8-Ks filed between 6 July and 14 September and the 1 September dividend declaration, and every reported figure matches. Two things have happened since, they pull in opposite directions, and this note grades neither. First, the condition's own wording is on the record: the 8-K filed 3 August (SEC accession 0001193125-26-329565) states that "$52.4 million in proceeds from the bitcoin sales were used to fund dividends on Strategy's preferred stock", in a week when bitcoin closed 13% to 17% below the cost basis and share issuance raised 2.8 times what coins did. Second, the selling stopped and the premise lapsed. No bitcoin has been sold since the week to 9 August (3,328 coins across the two selling weeks). Share sales raised $2.94bn in the three weeks to 30 August and paid both August STRC dividends ($52.4m and $50.7m); the week to 30 August bought 4,603 coins at $80,318, roughly $16,000 a coin above the two sales; the two weeks to 13 September saw no share sales and no coin sales at all. The USD Reserve stands at $5.10bn. Bitcoin has closed above ~$75,500 on all 29 days from 22 August to 19 September, never by much at the low (nearest close $75,590, on 16 September); the cost basis is now $75,412. Held open on purpose: the quarter's largest payment (the quarterly dividends on STRF, STRK, STRD and STRE, plus STRC) falls on 30 September, and its funding is disclosed in the filing that follows. Whether a condition met on 3 August outlives a voiding premise that lapsed on 22 August is the author's call, deferred until that filing is in.
Miss
Made 2026-05-15 · Resolved 2026-08-22 · Miss
Jane Street's Q2 2026 13F will show its IBIT and FBTC positions flat-to-lower again, consistent with a continued wind-down of the ETF basis trade rather than a re-entry.
Resolution condition: the Q2 2026 13F (due ~14 August 2026) shows IBIT and FBTC holdings at or below the Q1 levels. A material increase resolves this a miss.
Confidence: Medium. Source: Update on the Dam. This entry was filed with the note "a 13F shows only the long leg", which is false and has been corrected in the source article: both filings mark every row as shares, a call, or a put. The claim above and its resolution condition are unchanged, and are graded exactly as written.
Resolution note (2026-08-16), entry held open: the Q2 filing landed on 14 August (SEC accession 0001595888-26-000108) and settles one half of the condition, not the other. FBTC: 1,954,174 shares in Q1 down to 886,736 in Q2, a cut of 54.62%, read from both filings. That half is a hit. IBIT: 24,878,191 shares in Q2, and the Q1 figure has never been read at source (the Q1 information table runs to 11,162,326 bytes and the row sits past the fetching tool's five-megabyte ceiling, so it is unread rather than absent). The condition names both funds and requires both, so one failing half settles this a miss with no arithmetic needed. It is not held for being mixed. It is held because the half that would fail has no Q1 figure anyone has read: news write-ups put it near 5.9 million shares, and a write-up is not the filing. This entry resolves the day that row is read. Full working: The Part of the Filing Nobody Read.
Resolution (2026-08-22), MISS: the Q1 IBIT block was read at primary source on 22 August, from the same filing (SEC accession 0001595888-26-000043, CUSIP 46438F101, document q1_2026-13f.xml). The long leg is 5,872,212 shares, $225,610,385, summed from the filing's three manager-split rows, all seven Q1 rows pricing at $38.42. Against Q2's 24,878,191 shares that is an increase of 19,005,979 shares, +323.7%. The condition names both funds and requires both: FBTC is at or below its Q1 level and IBIT is more than four times it, which is the material increase the entry itself named as its miss trigger. This entry resolves a miss. The write-up figure of ~5.9 million shares that the note above declined to grade on is superseded by the filing, not corroborated by it. Nothing here settles the Q4 2025 quarter (accession 0001595888-26-000023), which is still unread, so the source article's conclusion remains unsupported by the document it rested on rather than shown to be wrong.
Open
Made 2026-05-15 · Review 2027-05
OTC desk balances keep draining over the next year, with no sustained recovery back above roughly 150,000 BTC.
Resolution condition: CryptoQuant-tracked OTC desk balance stays below ~150,000 BTC through May 2027 without a multi-month reversal. A durable climb back above that level resolves this a miss.
Confidence: Medium. Source: Update on the Dam. The earlier "depleted by July 2026" framing was a straight-line guess and has been replaced with this horizon; the direction is the claim, not the exact date.
Interim check (2026-07-07): no public CryptoQuant print shows a sustained recovery above ~150,000 BTC; no fresh mid-2026 figure was published in free sources to confirm the level either way. Open, leaning intact.
Interim check (2026-09-19): still nothing to grade against. No dated CryptoQuant print for 2026 turned up in free sources: the newest dated figure in circulation is roughly 146,000 BTC for February 2025, which predates this entry and was read from secondary write-ups (CryptoQuant's own post refused the fetch). Pieces dated 2026 repeat "near 150,000" with no date and no chart. The series sits behind a paywall and its last readable value sits on the entry's own threshold, so neither a drain nor a recovery can be confirmed. Open, and unverified in both directions, which is weaker than the "leaning intact" above. This entry needs a readable source before its 2027-05 review; if the resolution source is restated, the change is noted here.
Open
Made 2026-06 · Multi-year, open-ended
The next large US monetary accommodation arrives framed as an emergency facility (regional-bank stress, a failed long-end Treasury auction, or a foreign-reserve dislocation), not as an announced change of regime.
Resolution condition: the next episode that expands the Fed balance sheet by a comparable magnitude to 2020 or March 2023 is introduced through an emergency facility rather than a planned policy pivot. Open-ended; reviewed when such an episode occurs.
Confidence: Medium. Source: Gradual, Then Big.
Interim check (2026-07-07): the gradual phase is running on script: QT ended December 1, 2025 after repo-market stress, and the Fed has since grown its balance sheet by $200B+ under the label "reserve management purchases" (total ~$6.74T at June 24, 2026), with the Standing Repo Facility absorbing year-end strain. Technical labels, no announced regime change. Magnitude remains far below 2020 or March 2023, so the entry's trigger has not fired. Open. Full status report, with primary-source figures: Gradual, So Far.
Interim check (2026-09-19): the gradual phase has flattened. Total assets were $6.747T on 16 September (H.4.1 via FRED WALCL), up $211bn from the 3 December 2025 low but only $11bn since 24 June; the high so far is $6.760T, on 12 August. The largest four-week rise in the whole run is $105bn, to 31 December 2025, against $392bn in two weeks in March 2023. No emergency plumbing is in use: the Standing Repo Facility has not exceeded $1bn on any day since 1 July, and Fed loans stand at $6.9bn. Trigger not fired. Open.
Nothing here has been graded yet, which is the honest state of a scorecard started in mid-2026. The first real test was the Q2 13F in August. It arrived on 14 August, and it settled half of the entry it was written for while leaving the deciding half resting on a row nobody has read. The dated note on that entry says so in full. A grade withheld for a stated reason is part of the record; a grade guessed at is not.