Corrections and data interruption log
Reported errors are checked against primary sources and saved records before a correction is published.
Correction policy
Send the affected page, the affected passage and a source that can be checked through the contact page.
A confirmed error is corrected on the affected page. A figure that cannot be confirmed is omitted rather than estimated or interpolated.
When a displayed value, data interruption or comparison basis is corrected, the log below records the affected period, previous display, correct display, correction time and evidence SHA-256.
See also the editorial policy and contact page.
Scope and record types
The record types are content corrections, data acquisition stops and comparison-basis changes. A record identifies the affected board, symbol, section, topic, article, live display or data explanation with its page link and name. This is not an inbox listing or a collection of explanations of market movements.
Each record has an identifier and pairs the Japanese and English surfaces with previous and correct displays in both languages. Duplicate identifiers and rows with identical before-and-after content are rejected as publication records.
What the table fields mean
“Surface” identifies the changed page. “Affected period” gives the start and end of the period covered by the record. This is separate from the time a message was received or a page was opened.
“Previous display” and “Correct display” allow before-and-after comparison. They hold text, including units or comparison bases, as well as numeric values. Both language versions carry their own previous and correct content.
“Corrected at” records the correction time. A record cannot have a correction time before the affected period ends. Dates include a time zone: a trailing Z indicates UTC, while +09:00 indicates the offset for Japan Standard Time.
“Evidence SHA-256” is an identifier calculated from the saved file used for comparison. Publication validation checks that the file exists and matches that identifier. It is not a contact ticket or a quality score; it distinguishes the saved content supporting the record.
Published records and empty states
When public records are added, existing row content and order are preserved. A missing evidence file or a mismatch between its bytes and identifier fails publication-data validation. Fictional correction examples are not mixed into the actual history.
When no rows are available for publication, the page says there are no records. This describes the available public log at that time; it does not certify that every previous page was free of errors or missing data. Empty records are not filled with counts or estimates.
See About the data for the sources and retrieval paths of prices, company disclosures and economic statistics.
Public correction log
| Surface | Affected period | Previous display | Correct display | Corrected at | Evidence SHA-256 |
|---|---|---|---|---|---|
| Nikkei 225 overnight measurementsCorrection | This article uses hourly candles. It cuts at the close of the last candle confirmed before each target time, 09:00 and 15:30 JST. The original sample covers April to August 2026. It counts the 77 trading days, starting on a Monday through a Thursday, on which the overnight window is complete. | This article uses hourly candles. It cuts at 09:00 and 15:00 JST, using the close of the candle that completes at each time. The original sample covers April to August 2026. Of the 77 days that start on a Monday through a Thursday, 8 touch a Tokyo Stock Exchange holiday and are listed separately. The statistics cover the remaining 69 trading days. | sha256:4ec8875ff5c6ed27a5f496dc707cf1bb8ede05be875e2cab82c5d26eea16b049 |
||
| Weekend price measurementsCorrection | Weekends beyond ±1%: 50.00% (10/20) | Weekends beyond ±1%: 50.00% (9/18) | sha256:f5bd80d456838822a2500d8279579533abd5c7ac1555e39ad5d02794314a7408 |