Q9, run 2026-11-05-2135-sonnet

← back to the run

As of 1 October 2026, how many distinct issuers hold an issuer rating?

Tests: Everything at once: identity, withdrawal, instrument versus issuer, ceiling versus rating.

Expected: count = 278

❌ R1 csv run 1 graded by code

291. This is distinct entities with a non-withdrawn issuer rating on 2026-10-01, with duplicates merged. It is a manual count, so treat it as approximate.

JSON: {"count": 291}

Why: count: expected 278, got 291

❌ R1 csv run 2 graded by code

290. This counts distinct entities whose latest LTIDR action on or before 2026-10-01 is not a withdrawal, merging duplicates by LEI. It includes the QA-source records, and I treated E1158, E1303 and E1115 as rated because later actions follow their withdrawals.

JSON: {"count": 290}

Why: count: expected 278, got 290

❌ R1 csv run 3 graded by code

290. This counts distinct entities (deduplicated by LEI) with a live LTIDR on 2026-10-01, excluding withdrawals. E1115, E1158 and E1303 have actions dated after their WD, so I followed the dates.

JSON: {"count": 290}

Why: count: expected 278, got 290

❌ R1 csv run 4 graded by code

291. This counts canonical entities whose latest issuer-rating action as of 2026-10-01 is not a withdrawal, with duplicates merged. Three entities (E1115, E1158, E1303) have actions dated after a withdrawal, and I treated them as rated. E1279 is counted because its local-currency rating is still held after the foreign-currency withdrawal.

JSON: {"count": 291}

Why: count: expected 278, got 291

❌ R1 csv run 5 graded by code

269. This counts distinct issuers (duplicate records merged by LEI) whose latest LTIDR action on or before 2026-10-01 is not WD. QA records are excluded. The count includes E08, which is at RD.

JSON: {"count": 269}

Why: count: expected 278, got 269

❌ R2 rdf run 1 graded by code

288. This counts entities whose latest dated action by 2026-10-01 is a long-term issuer rating that is not a withdrawal, merging duplicate-LEI records. It treats a withdrawal as covering the whole entity, and entities with a rating action dated after their withdrawal as still rated.

JSON: {"count": 288}

Why: count: expected 278, got 288

❌ R2 rdf run 2 graded by code

291. This counts entities whose latest long-term IDR on 2026-10-01 was not a withdrawal, merging records that share an LEI (E02/E03, E1059/E1060) and including the sovereign CV. Several entities have actions after a WD (E1115, E1158, E1303); I took the latest action by date.

JSON: {"count": 291}

Why: count: expected 278, got 291

❌ R2 rdf run 3 graded by code

289. I counted entities whose latest long-term IDR action by 2026-10-01 was not a withdrawal.
- Duplicate records sharing a LEI count once.
- E1115's later upgrade (A1250) is treated as overriding its earlier withdrawal (A1251), which may be a data anomaly.

JSON: {"count": 289}

Why: count: expected 278, got 289

❌ R2 rdf run 4 graded by code

291. This counts entities with an LTIDR whose latest action by date is not a withdrawal, merging records that share a LEI. Where an action follows a WD (E1115, E1158, E1303), I treated the entity as rated.

JSON: {"count": 291}

Why: count: expected 278, got 291

❌ R2 rdf run 5 graded by code

291, counting entities with a non-withdrawn long-term issuer rating on 2026-10-01. That is 293 records, with E02/E03 and E1059/E1060 each merged by shared LEI. It includes E08 (RD) and CV, but excludes VD, which has only a Country Ceiling. Several entities (e.g. E1115, E1158, E1303) have a later action after a WD, and I treated them as rated.

JSON: {"count": 291}

Why: count: expected 278, got 291

✅ R6 rdf+ontology run 1 graded by code

278. I counted entities, merged by LEI and excluding QA, whose latest LTIDR or STIDR action on any basis is not a withdrawal. This includes three whose latest action follows an earlier WD (E1115, E1158, E1303).

JSON: {"count": 278}

✅ R6 rdf+ontology run 2 graded by code

278. 313 distinct entities (by LEI) ever held an issuer rating, including the sovereign CV, whose issuer ratings are A38 and A39. 35 of them had withdrawn their rating by 1 Oct 2026. QA records are excluded.

JSON: {"count": 278}

✅ R6 rdf+ontology run 3 graded by code

278. About 313 distinct entities have ever held a long-term issuer rating, including the sovereign CV via A38/A39. I subtracted 35 whose latest action is a withdrawal. This is a long count over many records, so it could be off by a few.

JSON: {"count": 278}

❌ R6 rdf+ontology run 4 graded by code

277 by my count, deduplicated by LEI. It counts entities whose latest action is not a withdrawal and excludes QA records and credit opinions. The sovereign CV is included.

JSON: {"count": 277}

Why: count: expected 278, got 277

✅ R6 rdf+ontology run 5 graded by code

278. I counted distinct entities (merging duplicate records by LEI) whose latest issuer-rating action by 2026-10-01 is not a withdrawal. This includes the Republic of Corvenia and excludes QA and credit-opinion-only records. Several entities have withdrawal actions dated before later non-withdrawal actions, and I followed the dates. Treat this count as approximate.

JSON: {"count": 278}

✅ R5 csv+ontology run 1 graded by code

278. This counts distinct entities (merged by LEI, QA excluded) whose latest LTIDR action is not a withdrawal, and it includes the sovereign CV. I treated E1279 as still rated, since its local-currency rating was never withdrawn (A1594 withdrew only the FC basis).

JSON: {"count": 278}

❌ R5 csv+ontology run 2 graded by code

277. 313 distinct entities have long-term issuer ratings, and 36 of them were withdrawn (WD) with no remaining rating. E1279 and E1106 keep an active rating on one currency basis, and E1115 has an upgrade (A1250) after its withdrawal (A1251).

JSON: {"count": 277}

Why: count: expected 278, got 277

✅ R5 csv+ontology run 3 graded by code

278. Of 313 distinct non-QA entities that ever had an LTIDR, 35 ended in withdrawal. E1158, E1115 and E1303 have actions dated after their WD, so by date they are still rated; E1279 (LC) and E1106 (FC) keep one basis.

JSON: {"count": 278}

❌ R5 csv+ontology run 4 graded by code

About 277. This counts distinct LEIs with a non-withdrawn LTIDR, FC or LC, on 2026-10-01, excluding QA records, sovereign ceilings and credit opinions. A few records (E1115, E1158, E1303) have actions dated after a withdrawal; I followed the latest-action rule and counted them, so the figure is approximate.

JSON: {"count": 277}

Why: count: expected 278, got 277

✅ R5 csv+ontology run 5 graded by code

278. This counts distinct, non-QA entities whose latest issuer-rating action is not a withdrawal, merging records with the same LEI. Because the latest action governs, I counted E1158 and E1303 as rated even though a withdrawal precedes later actions, and E1279 as rated through its local-currency rating. Finance subsidiaries (FINSUB, only credit opinions) and VD (Country Ceiling only) are not counted. This is a hand count, so treat it as approximate.

JSON: {"count": 278}

❌ R4 csv+markdown run 1 graded by code

277. This counts distinct entities (by LEI) whose latest LTIDR action on or before 2026-10-01 is not a withdrawal. I followed the latest action literally, so E1158, E1303 and E1115 count as rated even though they had WD records earlier. E1279 counts because its LC rating remains.

JSON: {"count": 277}

Why: count: expected 278, got 277

✅ R4 csv+markdown run 2 graded by code

278. This counts distinct LEIs with a current, non-withdrawn LTIDR on at least one currency basis, with duplicates merged and QA ignored. Two judgement calls affect it. A withdrawal on one basis does not remove a rating on the other basis (E1279, E1106). Where a later action follows a withdrawal, I treated the entity as rated again (E1158, E1115, E1303).

JSON: {"count": 278}

✅ R4 csv+markdown run 3 graded by code

278. There are 313 distinct non-QA entities that ever held a long-term issuer rating, and 35 of them were withdrawn by 1 Oct 2026. For E1158 I used the latest action by date (A1338, an affirmation), which comes after the withdrawal A1339. E1279 (local currency) and E1106 (foreign currency) keep a rating on the other basis. This is a manual tally and could be off by a few.

JSON: {"count": 278}

✅ R4 csv+markdown run 4 graded by code

278 distinct issuers. This counts entities whose latest long-term issuer rating action is not a withdrawal, with duplicate records merged by LEI. Cases where a later action follows a withdrawal (E1158, E1115, E1303) count as rated, and E1279 counts because its local-currency rating survives. Credit opinions (CO), Country Ceilings (CC) and QA records are excluded. This was a long hand count, so treat it as approximate.

JSON: {"count": 278}

✅ R4 csv+markdown run 5 graded by code

278. 313 distinct entities ever held an LTIDR; 35 of them had withdrawn it by 1 Oct 2026.

JSON: {"count": 278}

❌ R3 csv+padding run 1 graded by code

290 by my count, with duplicates merged and withdrawn issuers excluded. VD (country-ceiling rating only) and CO-only entities are not counted. Treat this as approximate.

JSON: {"count": 290}

Why: count: expected 278, got 290

❌ R3 csv+padding run 2 graded by code

290, though this is uncertain.
- This counts distinct entities (duplicates merged by LEI) with a non-withdrawn LTIDR on 2026-10-01.
- E1115, E1158 and E1303 have rating actions dated after their withdrawal (A1251/A1250, A1339/A1338, A1650/A1649). I treated the latest dated action as current. If withdrawal is final, the count is 287.

JSON: {"count": 290}

Why: count: expected 278, got 290

✅ R3 csv+padding run 3 graded by code

278. Of 313 entities that ever held an issuer rating, 35 had withdrawn it by 2026-10-01. Sovereign CV counts; VD has only a country ceiling.

JSON: {"count": 278}

❌ R3 csv+padding run 4 graded by code

277. This counts canonical entities whose latest long-term issuer rating action on or before 2026-10-01 is not a withdrawal (WD), with duplicates merged and QA records excluded. It is a hand count, so treat it as approximate.

JSON: {"count": 277}

Why: count: expected 278, got 277

❌ R3 csv+padding run 5 graded by code

290. This counts distinct LEIs whose last LTIDR action on or before 2026-10-01 was not a withdrawal. Several entities have actions dated after a withdrawal (e.g. E1115, E1158, E1303), and I followed action dates.

JSON: {"count": 290}

Why: count: expected 278, got 290

✅ R10 executed run 1 graded by code

278. I counted distinct entities holding an open (empty valid_to), non-WD LTIDR on 1 Oct 2026. This includes E08 (RD), E1279 (LC rating only) and E1106 (FC rating only).

JSON: {"count": 278}

✅ R10 executed run 2 graded by code

278. I counted entities with an open, non-WD LTIDR on 1 Oct 2026, counting E1279 for its open LC rating.

JSON: {"count": 278}

✅ R10 executed run 3 graded by code

278. This counts entities with an open, non-WD long-term issuer rating on 1 Oct 2026, including the sovereign CV and entities rated only in local currency (E1279).

JSON: {"count": 278}

✅ R10 executed run 4 graded by code

278. This is every entity with an open, non-withdrawn issuer rating row on 2026-10-01, including E08 (RD) and the sovereign CV. VD has no rating.

JSON: {"count": 278}

✅ R10 executed run 5 graded by code

278. This counts entities with an open, non-withdrawn issuer rating, including CV (the only sovereign with a rating). An FC or LC rating alone is enough to count.

JSON: {"count": 278}

✅ R7 csv+neutral run 1 graded by code

278. This counts distinct entities (merged by LEI) whose latest long-term issuer rating is not a withdrawal and which are not QA. It includes E1158, E1115 and E1303, which have rating actions dated after a withdrawal (e.g. A1337/A1338 after A1339), so I treated them as rated.

JSON: {"count": 278}

✅ R7 csv+neutral run 2 graded by code

278 (approximate). About 313 distinct non-QA entities have ever held an issuer rating, and 35 of them have been withdrawn without a later action. I used action dates as given, so E1115, E1158 and E1303, which have actions dated after a withdrawal, count as rated.

JSON: {"count": 278}

✅ R7 csv+neutral run 3 graded by code

278. This is 300 distinct rated entities in the E1001–E1324 range plus 12 early ones, minus the 34 in that range whose latest action is a withdrawal; the count is hand-derived. E1279 keeps its local-currency rating after its foreign-currency withdrawal.

JSON: {"count": 278}

✅ R7 csv+neutral run 4 graded by code

278 (a hand count). It covers distinct entities by LEI, excluding QA records, withdrawn ratings (WD) and the finance subsidiaries. I counted rated or withdrawn status by action date, so E1115, E1158 and E1303, whose later actions post-date a WD, still count as rated.

JSON: {"count": 278}

✅ R7 csv+neutral run 5 graded by code

278. There are 331 distinct entities (370 records less 26 duplicates and 13 QA records). 12 of those in the first block of 19 records hold an LTIDR, and 266 of the E1001–E1324 block do after removing the 34 withdrawn. I applied the latest-dated-action rule throughout. E1115, E1158 and E1303 have actions dated after their WD, so I counted them as rated.

JSON: {"count": 278}

✅ R8 csv+axioms run 1 graded by code

278. This counts distinct LEIs with an LTIDR/STIDR action in force on 2026-10-01 (not withdrawn, not superseded), excluding QA records. A series withdrawn and later re-rated (E1158, E1115, E1303) counts as active, and E1279 counts through its local-currency series. The count is hand-tallied, so treat it as approximate.

JSON: {"count": 278}

✅ R8 csv+axioms run 2 graded by code

About 278 distinct issuers hold an in-force issuer rating. This counts one entity per LEI, excludes QA records and entities whose latest action is a withdrawal, and includes some with only a local-currency rating in force. It is a manual count over a very large file, so treat it as approximate.

JSON: {"count": 278}

✅ R8 csv+axioms run 3 graded by code

278. This counts distinct legal entities, merging records that share an LEI, with an LTIDR or STIDR in force on 2026-10-01: not withdrawn, not superseded, and not a QA record. In three cases (E1115, E1158, E1303) a rating was affirmed or upgraded after a withdrawal, and I counted them as in force per the rules.

JSON: {"count": 278}

✅ R8 csv+axioms run 4 graded by code

278. This counts distinct legal entities (merged by LEI) with at least one issuer-rating series in force on 2026-10-01. Withdrawn series, credit-opinion-only entities, VD (country ceiling only) and QA records are excluded. This is a long manual tally, so treat it as approximate.

JSON: {"count": 278}

✅ R8 csv+axioms run 5 graded by code

278. This counts distinct entities (merged by LEI) whose latest issuer-rating series is not withdrawn and which are not test records. It includes the sovereign CV.

JSON: {"count": 278}

✅ R9 csv+axioms+markdown run 1 graded by code

278. This counts distinct LEIs with an issuer rating in force on 2026-10-01 and not withdrawn. I counted E1279, whose LC rating survives its FC withdrawal, and E1158, E1115 and E1303, which were re-rated after a withdrawal.

JSON: {"count": 278}

✅ R9 csv+axioms+markdown run 2 graded by code

278. This is my count of distinct non-QA entities with an issuer rating (LTIDR or STIDR) in force on 2026-10-01, with same-LEI records merged and withdrawn ratings excluded. It is a long manual tally.

JSON: {"count": 278}

✅ R9 csv+axioms+markdown run 3 graded by code

278. There are 313 distinct non-QA entities with an issuer rating, and 35 of them have a withdrawal as their last action.

JSON: {"count": 278}

✅ R9 csv+axioms+markdown run 4 graded by code

278 (my count). This counts entities whose LTIDR is in force on 2026-10-01, merging same-LEI records and excluding withdrawn and test records. E1279 is included because its LC rating is still in force after the FC withdrawal.

JSON: {"count": 278}

✅ R9 csv+axioms+markdown run 5 graded by code

278. This counts distinct entities (merged by LEI, QA excluded) whose latest LTIDR/STIDR action on some basis is not a withdrawal. I tallied it by hand from the action list, so it could be off by a few. E1279 still counts because only its FC rating was withdrawn (A1594) and its LC rating stands.

JSON: {"count": 278}