Blueprint MMA corrections and accountability log: every correction to a forecast, article, or track record entry, with the original claim, the correction, and the date

Correction Standards

• Published picks do not change

  Once an official pick is published, its winner, win probability, method, and date stay in the Track Record. A correction adds context without rewriting the original.

• Public log

  Every correction is posted here with the original claim, the correction, and the date. Nothing is corrected silently.

• Facts and analysis are labeled separately

  We distinguish factual errors, such as a fighter record or result, from changes to explanatory writing. Both types are logged and labeled.

• Explanations can be corrected

  If supporting analysis needs a correction, we update the explanation and record the change here. The original official pick remains visible.

• Questions are welcome

  The Research Desk explains the policy in more detail, and Support can help with a specific entry.

1. Track RecordOctober 22, 2026

   Track Record regrade: picks that were public before the fight

   Original claim: The Track Record graded UFC Fight Night: Ankalaev vs. Guskov (July 25, 2026) as 9 correct winner picks, 3 misses, and 1 void. It graded 84 picks from May 30 to July 19, 2026 from internal prediction logs and described every graded pick as published before the fight.

   Correction: That card is now graded from the picks that were public before its scheduled lock: 8 correct, 4 misses, and 1 void. The July 26 correction had graded four forecasts that were not published before the fights, including a main-event pick of Magomed Ankalaev; the public pick was Bogdan Guskov. Those picks are now graded from the forecast or article that was public before each fight: 81 remain (54 correct, 27 misses) and 3 are removed because no pick was public before the fight.

   Note: The older picks have no forecast proof record, and some of those cards had no pick lock. Where an event lock ran, the pick public at the lock is graded, even if it changed later. 5 internal prediction logs for the Ankalaev vs. Guskov main event had been changed to the actual winner after the fight and were restored to the pick recorded before it. Nothing was deleted; every change is recorded in an append-only ledger.

   2026 benchmark CSV (grading_basis column) Original 7-5 record
2. ArticleSeptember 23, 2026

   Deep Analysis: Rodolfo Bellato vs Christian Edwards

   Original claim: The analysis described Christian Edwards as having a reach edge over Rodolfo Bellato.

   Correction: The frozen evidence lists Edwards’s reach as 78 inches and Bellato’s as 77.6 inches. The 0.4-inch difference falls within our fact checker’s 0.5-inch tolerance for effectively equal reach; the analysis should not present it as a meaningful reach advantage.

   Note: The original model response and published prediction remain preserved. The saved response does not establish how this comparison affected the probability.

3. ArticleSeptember 8, 2026

   Deep Analysis: Sean King III vs Jessie Rosas

   Original claim: The analysis described Jessie Rosas as having a longer reach than Sean King III.

   Correction: The frozen evidence lists Jessie Rosas’s reach as 68.1 inches but contains no verified reach measurement for Sean King III. It therefore does not support the article’s claim that Rosas has the longer reach.

   Note: The original model response and published 52% King prediction remain preserved. The saved response does not establish how this unsupported comparison affected the probability.

4. ArticleAugust 31, 2026

   Deep Analysis: Dan Hooker vs Salahdine Parnasse

   Original claim: The preparation analysis described competing team-affiliation listings as unresolved.

   Correction: Competing Saigon Sports Club and City Kickboxing listings were treated as non-blocking at the source cutoff. The structured conflict ledger did not mark the preparation lens unresolved.

   Note: The stored artifact, official winner, published probability, version, and track-record eligibility are unchanged.

5. Track RecordJuly 26, 2026

   UFC Fight Night: Ankalaev vs. Guskov

   Original claim: The public results summary showed 7 correct winner picks and 5 misses across 12 completed bouts.

   Correction: This correction graded the card as 9 correct winner picks, 3 misses, and 1 void. Four of the graded forecasts, including the main-event pick of Magomed Ankalaev, were not published before the fights, and the event lock was created with this correction, after the fights. It has been replaced: the card is now graded 8-4 with 1 void from the picks that were public before the scheduled lock.

   Note: Method calls: 4/12. Exact-round finish calls: 1/8. Technical verification is available in the linked JSON receipt.

   Correction receipt JSON Original 7-5 record Regrade entry

The proof ledger writes a row automatically every time a published pick is replaced or its result settlement changes, and it rejects edits and deletions. These counts come straight from it.

Changed after the lock13

A new version was published after the event lock. Each one is listed below.

Republished by an operator13

An operator republished the pick. Each one is listed below.

Re-linked by the system1

The pick was re-linked to its corrected fight record. Each one is listed below.

Updated before the lock44

A newer forecast replaced an earlier one before the event lock. The earlier version stays in its proof record.

Result settlement replaced141

A newer verified result replaced an earlier settlement of the same pick.