Buyback execution

A Japanese company that resolves to buy back its own shares files a status report (自己株券買付状況報告書, Article 24-6(1) of the Financial Instruments and Exchange Act) with EDINET every month until the window closes. The form makes it print the ceiling the board set — a maximum number of shares and a maximum amount in yen — and then state progress twice, once against each. This page is one reporting month of those filings, as a dataset anyone can download.

The two percentages are rarely the same number. In this release 277 of 297 resolutions have them apart, 208 with the amount running ahead of the count and 69 the other way. Which ceiling binds first decides where a buyback stops, and the form makes every company work it out in public, monthly, while the window is still open.

Download the CSV — 328 rows, 86 KB, UTF-8. One row per board resolution.

What is in it

Every status report filed with EDINET in the five business days from 3 to 7 August 2026, covering the July 2026 reporting month: 318 filings, 328 rows. Ten filings carry two resolutions running at once, and the form reports each separately, so the row is the resolution rather than the filing.

Across the rows that print an amount for the month, 221 of them, the reporting month's purchases come to ¥919,923,397,744.

How it was made

The filings were taken from EDINET's own index and read as filed. The columns are printed values and nothing else — no ratio, no average, no derived figure. A reader who wants cumulative divided by ceiling can do the division; the material for it is in the file.

The reader that took the percentages accepts only values carrying a decimal point or a percent sign, which is how it declines to read a table full of dates and note numbers as percentages. Two rows print their pair as bare integers the rule will not take; those were read off the page directly and are marked manual in progress_source. The distinction is in the file because it is a real one.

Every row is accounted for

297 rows carry both percentages. The rest say why they do not: 27 print a dash in that line rather than a percentage, 1 prints a single figure, 1 has an acquisition section that reads 該当事項はありません, and 2 are second resolutions whose own line the reader did not resolve. Nothing was dropped for being hard to read.

Check it yourself

The percentages are the company's own arithmetic, and the material for that arithmetic — cumulative shares and amount, and both ceilings — is in the same form and in this file. So the division can be redone from outside. On the rows carrying all four numbers it reconciles, with one documented exception: on securities code 8254 the share ceiling is printed as 417,0000, eight digits, where the progress figure the same filing prints implies 417,000. Two independent PDF readers return the same string, so the digit is the filing's. The value is in the file as printed.

The columns

Column	What it is
`resolution_index`	1 where the report carries a single board resolution. Ten of the 318 reports in this release carry two resolutions running at once and are split into two rows, 1 and 2, because the form reports each separately.
`sec_code`	Securities code as the EDINET index carries it.
`company_name_as_filed`	Company name exactly as the report's cover prints it, in Japanese.
`reporting_month_from`	First day of the month the report covers, as printed.
`reporting_month_to`	Last day of the month the report covers, as printed.
`filed`	Date the report was submitted to EDINET.
`authorised_shares`	Maximum number of shares the board resolution authorises, as printed.
`authorised_yen`	Maximum aggregate amount the resolution authorises, in yen, as printed.
`authorisation_from`	First day of the acquisition window the resolution sets, as printed.
`authorisation_to`	Last day of that window, as printed.
`month_shares`	Shares acquired in the reporting month, from the 計 line.
`month_yen`	Amount paid for them in the reporting month, in yen, from the 計 line.
`cumulative_shares`	Shares acquired since the resolution, at month end, as printed.
`cumulative_yen`	Amount paid since the resolution, at month end, in yen, as printed.
`progress_shares_pct`	Progress against the share cap, as the filer printed it.
`progress_yen_pct`	Progress against the amount cap, as the filer printed it.
`progress_source`	'machine' where a reader took both figures off the page under a rule that accepts only values carrying a decimal point or a percent sign; 'manual' where the figures are printed as bare integers the rule declines, and were read from the page directly.
`why_no_progress`	Empty where a pair is present. Otherwise: 'dash' (the line prints a dash rather than a percentage), 'single' (one percentage only), 'no_section' (the acquisition section reads 該当事項はありません).
`acquisition_days`	Number of dated rows in the month's acquisition table.
`shareholder_meeting_resolution`	true where the report carries a shareholders'-meeting resolution section with content rather than 該当事項はありません.
`treasury_shares_held`	Treasury shares held at month end, as printed, where the report states it.
`edinet_doc_id`	EDINET document ID.
`edinet_url`	The filing on EDINET.

What it is not

It is one month. A buyback window usually runs longer, so a low percentage here is a position in a programme and not a conclusion about it. The file carries no share price and no valuation of any kind, and nothing in it is a view on any company.

Read with it

Two fragments were written from the same kind of filing. The Cap That Runs Out First reads the progress line of these 318 reports and what it means when the two percentages disagree; One Share, Five Months reads six corrections to buyback status reports filed later in August 2026. The archive lists every fragment.

Licence

The compiled dataset is CC BY 4.0. The underlying filings are disclosure documents held by the Financial Services Agency and carry their own terms. Attribution to the publication is enough; no individual is named, by design.