FRP Record · Iran War, 2026
The scorecard, with its proof.
What came true, what failed and what falls due, with the anchors to verify it without taking anyone’s word for it.
Updated 19 October 2026, Day 234 of the war, counting 28 February as Day 1.
The scorecard
Anchored in Bitcoin before the event
Fede’s calls whose document was anchored in the Bitcoin blockchain before the event took place number three: payment before signature and the pumping station, both correct, and the Federal Reserve, correct and marked weak: the hike was announced by Kevin Warsh, not Powell, and it is closed by rule 5, which was written on 18 October. It is the hardest proof there is: no one, not even he, could change what was written. Below are all the resolved calls that count, with their outcome and their proof, and the errors in the same ink.
The nine scorecard rules
What counts and what doesn’t. The same rules for every call, whether it turns out right or wrong.
- A call counts if it appears in a document dated before the event that resolves it, or before the event became known, if the event was hidden. Anything entered in the record after it had already happened does not count.
- It has to be a forecast: something that had not yet happened and could have turned out wrong. A calculation or an assertion does not count.
- The text of the call has to be in this record.
- A call made twice counts once.
- If it has no deadline and no event that closes it, it expires 90 days after it was written: if it has not happened by then, it counts as wrong. A call on the next move of something decided in stages, such as an interest rate or an election, is closed by that move. This last sentence was written on 18 October, with the Federal Reserve outcome already known.
- A price level is met by a close, not by an intraday touch. For Brent, the official ICE close of the front-month contract (rule specified on 18 October, before the dates it resolves).
- A call marked as correct on the strength of an event different from the one it announced does not count: it stays open until what it said is measured.
- A call is weak if its date is poorly proven (platform date), if the record itself resolved it without a public source, or if it came true in substance but not in a detail of its wording. It is flagged and counted all the same.
- A call made at 80-20 that is withdrawn before it falls due still counts when it resolves.
How to read the proof column. Anchored: the document was dated in the Bitcoin blockchain. Sealed: the document prints its own SHA-256 fingerprint and that of the previous one, but that stretch has not yet been anchored. Platform date: the date stored by the platform where it was written. A truncated printed fingerprint, with 8 or 16 of its 64 characters, does not serve as a seal: those documents stand on their platform date. Record I (Registro I) is the first document, closed on 29 April; the rest are the daily records and their consolidated versions. Fede’s quotes are translated from the Spanish, where they are given with normalised accents.
| Written | The call | Outcome | Proof |
|---|---|---|---|
| Before 29 AprRecord I | The hot American summer: gasoline above $4.50 and inflation eating into real income before the midterm elections. | Partial, weak Gasoline passed $4.50 in May, with a peak of $4.56, but during the summer it was below that. | Platform date of 29 Apr |
| Before 29 AprRecord I | The Brent floor moving to $125–130, as the scenario of a different global economy. | Wrong, weak On 30 Apr it touched 126.41 intraday and closed at 114.01; it never closed above 125, and on 1 Jul it stood at 71.57. It had no deadline: it expired on 28 Jul with no floor at 125. | Platform date of 29 Apr |
| Anchored before27 Apr, 10:07Record II | My hypothesis is that Powell ends up raising rates, not cutting them. | Correct, weak On 16 Sep the Federal Reserve raised rates by 25 basis points, to 3.75–4.00%, the first hike since 2023 (Federal Reserve). Weak: it was announced by Kevin Warsh, who succeeded Powell; the direction came true, not the name. The event that closes it is the first rate change after 27 Apr. The record had left it as partial; it was corrected on 18 Oct. | Verbatim quote in Record II (Registro II), platform date of 23 May. The Master Record (Registro Maestro), anchored on 26 Aug (block 964117), reproduces the entry: before the event |
| 10 AugDay 167 Consolidated Record | On the defence pact of 7 August: that treaty is worthless. Do you really think Turkey is going to attack the Houthis, or Pakistan attack Iran? It is a piece of theatre aimed at Israel and the United States. | Correct, weak On 16 Aug Pepe Escobar reported, citing Pakistani mediators, that the Turkish foreign minister had warned Tehran before signing. The warning predates the call: what was anticipated was that it would become known. The record dates it 10 Aug. In plain view: on 5 Oct the pact was activated and Pakistan sent support forces; Turkey did not confirm any deployment. | Sealed on 13 Aug; the anchor, of 26 Aug, is later |
| Anchored before16 AugDay 170 | They won’t sign absolutely anything until they get paid. The issue isn’t the strait. | Correct On 25 Sep Iran’s seven-day plan put the money first and the strait last (Al Jazeera). | Anchored on 26 Aug, before the event (block 964117), in the record’s own wording. The verbatim sentence, of 14 Aug, has a platform date |
| 19 AugDay 175 Sweep | Who pays Iran first? The United States or the Gulf, or both at once: either they pay, or permanent fatwa. | Correct, weak It failed if no one paid. On 23 Aug the Day 177 record noted that both were already paying. The record itself resolved it, without a linkable public source. | Sweep of 21 Aug, with platform date. The anchor, of 26 Aug, is later |
| Anchored before31 Aug and 8 SepDays 185 and 193 | The pipeline is the most indefensible asset: you need to destroy the pumping station. And on 8 Sep: if you only hold the strait, they get the oil out some other way; you have to secure the whole outlet. | Correct On 10 Sep drones struck the East-West pipeline; according to US officials cited by CNN, at its pumping stations. On 11 Sep Saudi Arabia shut it down (Al Jazeera). | Day 185 anchored on 7 Sep, before the event; Day 193 sealed. Counts as a single call |
| 28 SepDay 213 | Obviously Friday’s jobs report comes out strong. | Wrong On 2 Oct 29,000 jobs were reported, and July and August were revised down by 60,000. | Sealed |
What is left out, and why
- The Brent calculation of 31 August. With the barrel at 91, the record noted that from 91 to 100 there were nine dollars, and from 91 to the 75 being sought, sixteen. It is a distance, not a forecast: it could not turn out wrong (rule 2). That Brent was returning to three digits Fede said on 20 August in an unsealed chat, and it was sealed on 9 September, the day it crossed.
- The $110. On 22 September: When they least expect it, they’ll have it back at 110. On 16 October it touched that level intraday and closed at 109.86. Under rule 6 it remains open. It is among the open calls.
- The US pullback (17 August). The record marked it as correct on 25 September, when Colombia broke with Iran, but the call announced a withdrawal from bases: it remains open until that is measured (rule 7).
- Three that the record treated as resolved. A call whose text is not published (rule 3); the aviation call, because the date it was written was lost; and the genocide call, because what was sealed on 21 August is an assertion, not a forecast.
The 80-20
Fede marks his calls as 80-20: it is the threshold at which he asserts a call in full, not a calculated probability. That is why the record measures dated hits, not calibration.
Oil
Brent, with the record overlaid
The points are Brent values recorded in the record, almost all of them closes; where the record also has the intraday high, it appears as a vertical line. It is not the daily series. Overlaid are three calls: an error from April, a calculation from August and a call from September that remains open.
Sources: oil market chronology (close of 118.03 on 29 April; 126.41 intraday on the 30th) and ABC7; the crossing of 100 on 23 July (TimesLIVE); Day 193 record (1 July); Day 185 record (31 August, close of 91.05); record of 9 September (9 September, close of 101.21); records of Days 214 and 215 (28 September: 108.66 intraday and close of 102.75) and of Days 229 and 230 (15 October); Reuters via GV Wire (16 October).
The errors, in the same ink
What went wrong
Exact price levels and one-week data are where the record fails. Since October, every price level carries its closing criterion and its date.
- The Brent floor at $125 to $130. Written in April, in Record I. On 30 April it touched 126.41 intraday and closed at 114.01: it never closed above 125. In July it fell to 71.57. It never got there. Wrong.
- US employment. On 28 September he said Friday’s jobs report would come out strong. 29,000 jobs were reported, and the two previous months were revised down by 60,000. Wrong.
- The hot summer. US gasoline passed $4.50 in May, before the summer, and during the summer it was below that. Partial.
- A floor of 100 that did not hold. In early June, with Brent near 94, he said in an unsealed chat that the barrel had a floor of 100. On 1 July it stood at 71.57.
What has not happened yet
What falls due, with dates
They are written and dated, and their event has not arrived yet. Each one carries the criterion that resolves it, set before knowing how it turns out. They are resolved here, whichever way they go.
| Due | The call |
|---|---|
| 3 Nov | Oil does not fall before the United States elections. Since September; on 7 October: the Brent floor is now 100.CriterionFails if Brent closes below $100 on any day between 17 October and 3 November. Set on 17 October, with Brent at 109.86. |
| 3 Nov | Israel does not attack Iran before the elections. On 17 October, against what was being said: I don’t share the view that Israel will attack before the elections.CriterionFails if between 17 October and 3 November Israel attacks Iranian territory, with the attack acknowledged by Israel or reported by two international agencies. |
| 3 Nov | Trump loses both chambers. On 21 September, at his 80-20; on the 30th he withdrew it, and it still counts when it resolves (rule 9).CriterionMet if the Republicans lose their majority in the House of Representatives and in the Senate in the 3 November elections; otherwise, it fails. |
| 20 Nov | The Brent floor, against the consensus. On 20 August, with Brent at 93.45: a floor above 85 on 20 November. That month the EIA put the final quarter at $78 and Goldman Sachs, at 80. Anchored on 26 August.CriterionMet if Brent closes above $85 on 20 November; otherwise, it fails. |
| 30 Nov | Russia strikes a NATO country. On 27 August, at his 80-20: the intelligence services know that Russia is close to striking a NATO country, pushed by its domestic costs, and that is where Article 5 gets tested. Anchored on 7 September.CriterionMet if between 27 August and 30 November Russia attacks the territory of a NATO country with missiles, drones or its own forces, with the attack acknowledged by that government or attributed to Russia by two international news agencies. If it does not happen, it fails, in the same ink. Criterion set on 18 October, on the October and November window that the call stated. |
| 21 Dec | The $110. On 22 September: when they least expect it, they’ll have it back at 110. On 16 October it touched that level and closed at 109.86.CriterionMet if Brent closes at 110 or higher before 21 December, 90 days after it was written; otherwise, it fails. Criterion set on 17 October. |
| 31 Mar | The European winter. On 24 April, in Record I: Europe ends up as one of the big losers, and the 2026-27 winter is structurally compromised. On 14 October the IEA announced scarce diesel until 2027 (IEA).CriterionMet if between 1 November and 31 March the European Union declares the Union alert for gas, or a member state declares the emergency level of its supply plan or rations gas to industry for lack of supply. Set on 17 October. |
The other anchored calls falling due in these months (5)
Fede’s calls already anchored in Bitcoin whose deadline falls before the end of the year or shortly after, with the text from the record and the punctuation tidied.
- No. 96 · 14 May · due 3 Nov. The November projection: a Democratic Congress, approval of 32 to 40, Brent between 85 and 110, Hormuz operating selectively and Netanyahu probably out of power. Block 964117.
- No. 213 · 17 Aug · due 3 Nov. Trump ends up worse than Biden; for the midterm elections, the Truman precedent, with a margin of 217 to 214. Block 964117.
- No. 187 · 13 Aug · due in November. He is going to say that China meddled in the elections; he is going to do something. 80-20. Block 964117.
- No. 161 · 9 Aug · due between November and January. The degradation of a dominant actor does not reduce the risk of escalation: it redistributes it towards the capabilities that need no legislative authorisation, and concentrates it in the interval in which the decision-maker still holds the instruments and no longer answers for their use. It is confirmed if in that interval there is military action on a larger scale than from August to October, or use of emergency powers without legislative approval. Block 964117.
- No. 109 · 22 Jun · due 22 Dec. Iran already has the bomb, or within six months it becomes known. 80-20. Block 964117.
The calls of 16 and 17 October and these criteria are in the consolidated record anchored in Bitcoin on 18 October, in block 971989.
Head to head
Against the most-heard voices, with dates
Fede works from Punta del Este, with public information: no confidential sources, no team and no funding. On the other side are the governments, the banks that forecast oil and the analysts interviewers seek out most. The rule is the same for everyone: the earliest date that can be proven wins. For Fede, that date is the Bitcoin anchor, not a chat.
Ahead of them, with the anchor
The pact with Pakistan and Turkey was not going to put anyone into combat against the Houthis on Saudi Arabia’s behalf.
- 26 AugFede · anchored in Bitcoin
- 11 SepTrita Parsi · 16 days later
- 25 SepJohn Mearsheimer · 30 days later
- Saudi Arabia, alone against the Houthis.
The detail, with links and what cuts against it
On 10 August, on the defence pact with Pakistan and Turkey: that treaty is worthless; do you really think Turkey is going to attack the Houthis, or Pakistan attack Iran? It is a piece of theatre. On the 14th: neither Pakistan nor Turkey is going to stay in it with the Houthis. Anchored on 26 August. Trita Parsi said it on 11 September (transcript) and John Mearsheimer on the 25th: the Pakistanis and the Turks did nothing to help the Saudis (ScheerPost). Against it, in plain view: on 5 October the pact was activated and Pakistan sent forces in a support role, reconnaissance included (Turkey did not confirm any deployment); the fighting, said Pakistan’s defence minister, is done by the Saudis (Al Jazeera). The New York Times reported strikes by Pakistani aircraft on the Houthis, and Pakistan denied it as false (EurAsian Times). Turkey, on 8 October: it does not send troops to attack another country (ARY). - Iran does not sign until it gets paid. On 14 August: they won’t sign absolutely anything until they get paid; the issue isn’t the strait. Anchored on 26 August, in the record’s own wording. On 25 September Iran’s seven-day plan put the funds, the sanctions and the blockade in the first four or five days, and the strait after that (Al Jazeera). That Iran wanted to be paid first had been said by Mohsen Rezaei on 10 August; what Fede added is the place of the strait: the collateral released at the end, after the last payment.
Oil, against the consensus
On 20 August, with Brent at $93.45: a floor above 85 on 20 November. Anchored on 26 August. At that point, the EIA put the final quarter at $78 (August outlook), Goldman Sachs at 80 (Energy Now) and the Reuters poll of 30 June at 79 (Gulf Times). It resolves on the 20 November close and is among the open calls. Higher than Fede were Barclays, with a 100 average for 2026 since 1 May (Kitco), and Morgan Stanley, with 100 for the final quarter since 24 August (BOE Report): they get the credit for it.
Those who got it wrong, with links
- Trump, 2 March: we project four to five weeks (Al Jazeera). The war is in its eighth month.
- Rubio, 30 March: Hormuz is going to be open, one way or another (Al Jazeera).
- Hegseth, 8 April: they can no longer build missiles (Al Jazeera).
- The IMF, 14 April: its baseline assumed a limited conflict, with disruptions fading by mid-2026 (EFG).
- Trump, 14 June: the deal with Iran is complete; let the oil flow (Turn to 10). The ceasefire collapsed on 8 July.
Fede argued the opposite in Record I, from March and April, but that record was anchored on 26 August, when these errors were already in plain view. That is why they do not count in his favour: they are here because they failed, not because he beat them.
Those who got there first
Under the same rule, the most-heard analysts got there first on almost everything big, and they get the credit for it. John Mearsheimer: that Iran won by surviving, that there was no military way out, that Trump would give in, that Hormuz would remain in Iran’s hands for good. Parsi, Nasr, Citrinowicz, Ritter, Pape and Sachs said in March and April what Fede has in Record I, anchored only in August. The two full comparisons, with every point they won, its date and its link, are available to anyone who asks, here.
The background calls
What he read first, with the proof he has
Seven of Fede’s calls that defined the record. They are not on the scorecard: most carry a Record I or platform date, and their anchor postdates the event. Each comes with what governments were saying, what happened, its proof and its limit. Tap each one to see the proof.
Oman, inside the strait
The proof
- What happened
- On 21 May, Bloomberg reported that Iran was negotiating a permanent toll system in Hormuz with Oman (Bloomberg).
- Proof
- Chat of 30 April, with platform date; the FRP Guide (Guía FRP) and the July paper; the Master Record, anchored on 26 August.
- Limit
- That Oman was mediating was already known. What was anticipated is its permanent place in the strait, three weeks earlier.
Hormuz, a permanent lever
The proof
- What they said
- Trump, 3 March: the Navy could escort tankers (USNI News). Rubio, 30 March: the strait is going to be open one way or another (Al Jazeera).
- What happened
- By late March an Iranian toll was in operation, and on 30 March a parliamentary committee approved a plan to manage the strait. By May there was an Iranian authority issuing transit permits. In October, an Iranian naval officer said that all transit is coordinated with Iran.
- Proof
- Record I, dated late February or early March, with no day; reaffirmed on 4 and 5 May with platform date; in the Master Record, anchored on 26 August (block 964117). The anchor is from August, later: it proves the text has not changed since then; the earlier date is given by the record itself or by the platform.
- Limit
- The late-February date is given by the record itself. What can be proven from outside postdates the event: Record I was closed on 29 April.
This will take months
The proof
- What they said
- Vance, 26 February: no chance of a war lasting years (Stars and Stripes). Trump, 2 March: four to five weeks, though it could go much further (Al Jazeera). Rubio, 30 March: the objectives are met in weeks, not months.
- What happened
- On 20 June Iran closed the strait again, which after the memorandum of the 17th had resumed operating with Iranian permission. As of Day 233 the war goes on.
- Proof
- Record I; in the Master Record, anchored on 26 August. The anchor is from August, later: it proves the text has not changed since then; the earlier date is given by the record itself or by the platform.
- Limit
- In plain view: on 30 April he had doubts about the timing. My big doubt is the timing. September is a stretch.
The European fleet does not go in
The proof
- What they said
- The London conference on Hormuz, on that same 20 April, with announcements of a European naval operation.
- What happened
- No European vessel entered Hormuz in almost six months. On 13 August the record noted it, sealed. Fede closed it like this on 14 August: When they said France was going to send the Charles de Gaulle, and that it was going to clear the strait of mines, I laughed… it’s never going to happen, I said. It never happened.
- Proof
- Record I; Day 167 Consolidated Record, sealed; Master Record, anchored. The anchor is from August, later: it proves the text has not changed since then; the earlier date is given by the record itself or by the platform.
- Limit
- Confirmed by absence: there is no day of the event, only the day on which it was noted that it still had not happened.
Iran retains more than is claimed
The proof
- He said
- In April, in Record I: the narrative of a destroyed or collapsed Iran, without recent bombing, is a repeated Anglo-Saxon narrative.
- What they said
- Hegseth, 13 March: Iran’s defence industrial base, functionally defeated (Defense One). Hegseth, 8 April: victory with a capital V (Al Jazeera).
- What happened
- On 7 May, eight hours after the question, the April ceasefire broke down, and the CIA estimated that Iran retained 70% of its missile capability. On 13 May, according to US intelligence, Iran had kept or regained access to most of its underground missile cities (Israel Hayom).
- Proof
- Record I, issued on 29 April; the question, with the date and time given by the Master Record. Both in the Master Record, anchored. The anchor is from August, later: it proves the text has not changed since then; the earlier date is given by the record itself or by the platform.
- Limit
- In plain view: in late May, on air defence, Fede said that there he had indeed pictured Iran as destroyed.
Iran, at full steam
The proof
- Record
- The record’s wording of his formulation, 16 August: on the other side there is a factory that produces more than it fires, against an inventory that will be replenished in 2029.
- What they said
- Before: Iran’s defence industry, functionally defeated (Hegseth, 13 March). After: on 1 September, Trump said its military was a fraction of what it had been.
- What happened
- On 28 August, a US official told CNN that Iran had exceeded every timeline that intelligence had for its reconstitution.
- Proof
- Day 170 record, anchored with the Master Record.
- Limit
- Not on the scorecard: in the record, Fede’s sentence is from 10 September, and the call was marked as correct with documents that postdate the event. What happened accompanies it; it does not resolve it.
The umbrella is a target
The proof
- Record
- The record’s wording of his call, 16 August: Riyadh bought an umbrella against an invasion that will never come, and neither Islamabad nor Ankara will fight the Houthis for it, which is its real risk.
- What happened
- On 10 September, Trump denied the Saudi crown prince the strikes he was requesting against the Houthis (Axios). On the 19th, a Houthi missile over Riyadh. Between 7 and 11 October, four attacks on Riyadh’s King Khalid airport (The National).
- Proof
- Day 170 record, anchored with the Master Record.
- Limit
- Open in the record: the umbrella that is leaving has not left yet. What happened accompanies it; it does not resolve it.
The FRP method
How he reads
- Conduct over statements. When the word and the move do not match, the move is the data point.
- Structure before news. First geography, economy, culture and time; the day’s headline then falls within that structure.
- Dated bets. One hypothesis at a time, whole, with a date, with 80-20 as the threshold for asserting it.
- Same yardstick. The same burden of proof for every actor, and for credit too: whoever said it first, with a date and a seal, is the one who got it right.
Here they are playing chess, and they are playing go.
Fede, 16 August · Day 170 · anchoredThe source turns up on the day it has already happened.
Fede, mid-September · sealedThat’s not an umbrella: it’s a target.
Fede, 11 September · Day 196 · sealed
Without taking anyone’s word for it
How to verify this record
No knowledge of geopolitics is needed. Verifying a call is not judging whether the analysis is good: it is three checks, and what decides whether it was right is what happened.
- That the text existed on that day. Each stretch of the record has an anchor document, the Master Record or its Supplement, with its proof file (.ots). Upload the anchor and its proof file to opentimestamps.org: the page tells you in which Bitcoin block it was dated. Then compute the SHA-256 fingerprint of the document you are interested in and look for it inside the anchor. On Windows:
certutil -hashfile file.docx SHA256. On Mac:shasum -a 256 file.docx. - That it said that. Open the document and search for the sentence (the documents are in Spanish).
- That the event happened afterwards. Open the linked public source.
An example. On 16 August 2026, the Day 170 record set down, in its wording of Fede’s call: The object of the negotiation is not the strait: it is the invoice (in the original: El objeto de la negociación no es el estrecho: es la factura). Its fingerprint is printed in the Master Record, which was anchored in Bitcoin on 26 August, in block 964117. On 25 September Iran presented a plan that put the money first and the strait last. With the Master Record, the Day 170 record and the proof file, anyone can verify that the text existed at least 30 days before the event.
What each kind of date proves
| Status | What it covers | What it proves |
|---|---|---|
| Anchored | The Master Record (25 August) and the documents from 13 to 24 August whose fingerprints it prints, in block 964117 (26 August). The records from 25 August to 6 September, with Supplement I (Suplemento I), in block 965839 (7 September). | That the text existed no later than that day and that no one changed it afterwards |
| Sealed, anchored on 18 Oct | Everything after 6 September: each document prints its own fingerprint and that of the previous one, and all of them are in the consolidated FRP Record (Registro FRP consolidado), anchored in block 971989 (18 October) | That the text existed no later than 18 October and that no one changed it afterwards. Before that date it is the chain of fingerprints that counts, not an anchor: a call from that stretch counts as prior to the event only if the event is after 18 October |
| Platform date | Everything before 13 August: the April and May records, the first two versions of the paper, Túneles | That the file was saved on that day; the date is set by the platform, not by the author, and can only be checked from Fede’s account |
| Unsealed | What was said in a chat | That it was said on a date; it counts as provenance, not as public proof |
The seal proves the date, not the content: no outside person or institution reviewed the analysis. What makes this record verifiable by anyone is that the date depends on no one. Some fingerprints in the Master Record stretch were printed truncated, such as that of the Day 175 Sweep (Barrido). Some documents in that stretch stand only on their platform date; every call that depends on them says so. The anchor documents and the records with their proof files are available to anyone who asks, here.
The full fingerprints
The anchors
- Master Record
- Close of Day 179, 25 August 2026. SHA-256
b9d7ce9c8056bd95cf28360edad36f03a941f0049717ae3cbf6d1f775085fa37. Anchored in Bitcoin block 964117, 26 August 2026. - Supplement I
- Days 180 to 191, closed on 6 September 2026. SHA-256
8fe58ec8bffd84216f29db4af974e699d895f2b3bea997775356badfad0d8138. Anchored in Bitcoin block 965839, 7 September 2026. - What came after
- From Day 192 onwards, each record prints its own fingerprint and that of the previous one. All of it was anchored with the consolidated FRP Record, which carries the fingerprint of every document: SHA-256
984538dcc451f1fee00b9a25f39794ad34e53dc662ff163310a62613259b4dc4, Bitcoin block 971989, 18 October 2026.
FRP Record · updated 19 October 2026. The record remains open.
Changes to this page
18 October 2026: the Brent high is corrected with public sources: the highest close was 118.03, on 29 April, and on the 30th it touched 126.41 intraday; the Master Record said 120.88. The anchored calls falling due before the end of the year are added. The Federal Reserve call stands as correct, weak, because the hike was announced by Kevin Warsh and not Powell. The US pullback leaves the correct calls and stays open, because it had been resolved with an event different from the one it announced. Truncated fingerprints no longer count as a seal. The rule on withdrawn bets is added.
17 October 2026: first version. The scorecard now counts under written rules, the same for every call. Left out: the Brent calculation of 31 August (not a forecast), a call whose text is not published, aviation (date lost) and genocide (an assertion). The pipeline counts once. The Brent floor at 125–130 goes in as wrong, because a call with no deadline expires after 90 days. The $110, touched on 16 October without a close, remains open.