Methodology

Scoring rules (GNT-BDS v1.1.1)

What earns each score on all 34 questions, how they are weighted, the special cases, the evidence required, and every change to the rules.

Every change to these rules gets a version number. All 128 companies are scored against GNT-BDS v1.1.1. Version 1.1 was fixed as the public baseline on 20 October 2026; version 1.1.1, on 21 October 2026, clarified several rules after an independent audit, with the same questions and weights. Any change to the rules gets a new version number and a dated entry in the change history below. Scores that moved because of a rule change are labelled “Rule change” in the change log, with the previous score.

Weights

The 34 questions fall into five areas, and questions within an area weigh the same. An area’s score is the average of its question scores (0 to 5) scaled to 100; the total is the weighted sum of the area scores. The last column shows how much one question at 5 adds to the total.

Area	Weight	Questions	One question at 5
Financial disclosure	30%	8	3.75
Strategic disclosure	25%	7	3.57
Governance disclosure	20%	7	2.86
English IR website	15%	7	2.14
Japanese–English parity	10%	5	2

Where questions are not yet reviewed, two figures are shown: a floor counting them as zero, and a ceiling from the reviewed questions only. Every company listed now is reviewed on all 34, so each has one figure.

The scale

Score	Meaning
5	Complete and current. An English reader is not behind a Japanese one.
4	Whole, but late, or without the supporting material around it.
3	Cut down: part of the substance exists only in Japanese.
2	Token: the fact exists in English, the substance does not.
1	Present but undercut: the company publishes it and then disclaims it.
0	Searched and not found within the stated scope. The note says where we looked.
—	Not reviewed yet. No score, and left out of the ceiling.

One test separates a 4 from a 3: does an English reader have to open the Japanese document to get the substance? If not, and the English is whole but late or missing its supporting material, it is a 4. If so, it is a 3. A 4 is a delivery problem; a 3 is a content problem.

Where the evidence does not separate two adjacent levels, the lower one is recorded. A 4 means at least 4.

Special cases

A company with no English IR material at all is recorded as not found on every question, with one shared note.
English documents on the company’s own site count wherever they sit, including when they are linked only from Japanese pages. A company with English documents but no English pages is scored on every question; the missing pages count only in O1 and O4.
A question nobody has reviewed yet is shown as not reviewed. It is never entered as a zero.
Some questions have special cases, set out under each question below: a single-segment company (F5), a plan that has ended (S1), no Q&A in either language (S5), no plan in either language (S7), no cross-shareholdings at all (G7).
O7 is capped at 4, and R5 has no 4.

Evidence required

The search covers the English IR site in full, the latest two quarters of results materials, the latest annual or integrated report, the latest Corporate Governance Report, the latest AGM notice, a site search, and one outside search for the company name with the document type.
Everything is read in a browser, document by document. A title is not evidence; the document is opened.
The English is read against the Japanese: pages, sections and dates.
Dates are the dates the company’s site lists. Where the English site reuses the Japanese dates but the English files were created weeks later, period after period, the file dates are used, and the note says so.
Every answer records the document read (a link to the document itself, not a listing page), the date it was checked, and a note saying what was found.
A zero records where we looked and when.
Samples and windows are fixed per question: at least ten links for O2 and O3, the latest two results for R1, five named documents for R2, twelve months for R3.
Evidence sent by a company is accepted only as a link to material on its public website, and is checked the same way.

The 34 questions

Open a question to see what earns each score and the rulings that apply to it.

Financial disclosure

F1Quarterly and full-year results summaries (tanshin) published in Englishup to 3.75 pts

Score	What earns it
5	A results summary (tanshin) for every period, on the Japanese timetable, complete with its attachments.
4	Every period, but days to weeks behind the Japanese; or the full year plus only some quarters.
3	Only the full year or half year; or the statements without the statutory notes (most often the segment note).
2	One period only, or headline figures only.
0	No English results summary.

Two tests, periods and completeness; the lower result counts.
A results deck or a headline sheet is not a tanshin.
A missing cash-flow note is not a cut if the Japanese omits it too. A missing segment note is, for any company with more than one business.

F2English results summary is a full translation, not an excerptup to 3.75 pts

Score	What earns it
5	The whole tanshin read page for page against the Japanese: summary, qualitative section, statements and notes.
4	Whole, except attachments the Japanese side publishes separately (supplementary data).
3	Part of the document is cut: the qualitative review or the notes, the segment note above all.
2	The summary cover pages only.
0	No English tanshin.

An omitted voluntary auditor’s review report does not count against it.

F3Financial statements including notes available in Englishup to 3.75 pts

Score	What earns it
5	Full statements and notes for the latest period, from any current English source.
4	Full statements and notes, but only at the year end; the latest quarter has no notes.
3	Statements without notes.
2	Headline figures only.
0	None in English.

F4Annual securities report available in English in any formup to 3.75 pts

Score	What earns it
5	The full annual securities report (yuho) in English for the current year, with the independent auditor’s report.
4	The full report one year behind; or under another name with the same content; or complete except for an untranslated auditor’s report.
3	Part of it: the financial statements or extracts only.
0	No English version in any form.

The document is opened and checked for the statutory parts (business overview, risks, management’s analysis, governance, statements, auditor’s report). A title alone does not count.

F5Segment revenue and profit disclosed in English, consistent definitionsup to 3.75 pts

Score	What earns it
5	Sales and profit by segment every period, with segment definitions stated and held constant.
4	Every period, but the definitions are not stated, or the figures appear only in the deck.
3	Sales without profit, or only at the year end.
2	Segments named, with no numbers.
0	None in English.

A single-segment company that says so in English and gives sales by business line is scored on that, and the note says why.

F6Multi-year financial data file (Excel or CSV) downloadable in Englishup to 3.75 pts

Score	What earns it
5	An Excel or CSV file with five years or more of the income statement, balance sheet, cash flows and per-share data (segments count in its favour).
4	Excel or CSV, but narrower or shorter than that.
3	Excel or CSV a year out of date, or a single statement.
0	No downloadable data file.

An HTML table or a PDF fact book does not count; the note says what was there instead.
The file is opened and read (years covered, which statements) before a 4 or 5 is given.

F7Guidance disclosed in English with the basis explainedup to 3.75 pts

Score	What earns it
5	The full-year forecast with its assumptions (exchange rates, market view, segment forecasts), and revisions explained.
4	The forecast with part of the basis, such as an exchange-rate assumption or segment numbers.
3	The numbers only.
2	A qualitative outlook only, or an English statement of why no forecast is given.
0	None in English.

A 5 needs at least two of the three kinds of assumption plus explained revisions.
A company that forecasts only the next quarter, as a stated policy, can score 4.

F8English financial disclosure current to the latest Japanese filing periodup to 3.75 pts

Score	What earns it
5	English financial disclosure covers the same period as the latest Japanese filing.
4	One reporting period behind.
3	Two periods behind.
2	More than a year behind.
0	No English financial disclosure.

Judged by period, not by day: English a fortnight after the Japanese, for the same quarter, is still a 5.

Strategic disclosure

S1Medium-term management plan in English with numerical targetsup to 3.57 pts

Score	What earns it
5	The current medium-term plan in English with its numerical targets, as complete as the Japanese.
4	Complete, but published later than the Japanese, or a revision not yet translated.
3	An English summary with only some of the targets.
2	The plan named, with no numbers.
0	None in English.

Any lag behind the Japanese, even one day, makes it 4.
If a plan has ended with no successor, what is current is scored and the note says so.

S2IR presentations published in Englishup to 3.57 pts

Score	What earns it
5	A deck for every results briefing, current, with whatever narrative the Japanese side adds (script, transcript, Q&A, summary).
4	A deck for every period, but no English counterpart to that narrative.
3	Decks only occasionally, or only for the full year.
0	No English presentations.

A briefing video with no English version does not cost a point; the slides and transcript carry the substance.
A deck published late is scored under R1, not here.

S3English equity-story or investment-highlights pageup to 3.57 pts

Score	What earns it
5	A dedicated English page or section for investors (“why invest”, “investment highlights”, “at a glance”) with quantified strengths.
4	Dedicated and specific, but mostly qualitative.
3	A company-overview page doing the job: strengths described, not framed for an investor.
2	A generic corporate introduction.
0	None in English.

S4Growth drivers identified and quantified in Englishup to 3.57 pts

Score	What earns it
5	Growth drivers named, with numbers attached: market size, share, contribution to the target, capacity.
4	Named, some quantified.
3	Named, qualitative only.
2	Vague (“expand overseas”).
0	None in English.

A judgement question: the note explains the reading.

S5Results-briefing transcripts or Q&A published in Englishup to 3.57 pts

Score	What earns it
5	An English transcript or Q&A for every briefing the Japanese side publishes one for.
4	Every period, but later than the Japanese; or a Q&A summary where the Japanese has a full transcript.
3	Occasional, or the full year only.
2	A single old one.
0	None in English while the Japanese side publishes one; the note names the Japanese document.

An English Q&A that arrives any number of days after the Japanese is a 4.
If neither language publishes a Q&A, the answer is recorded as not found and the note says the Japanese side has none either.

S6Integrated or annual report available in Englishup to 3.57 pts

Score	What earns it
5	The current edition of the integrated or annual report, complete.
4	The complete report, one edition behind.
3	An abridged English cut, highlights or extracts.
0	None in English.

Currency is judged from the period the report states it covers, never from the year in its title.

S7Progress against the previous plan, including misses, explained in Englishup to 3.57 pts

Score	What earns it
5	The previous plan’s targets set against actual results, with the differences explained.
4	The comparison, but for some targets only, or without the explanation.
3	Progress described with some numbers, but not target against actual.
2	A mention only, such as progress on the current plan with no figures.
0	None in English (including a comparison published only in Japanese).

No penalty for having beaten every target and no credit for candour: what is scored is whether a dated comparison exists and is explained.
If there is no plan in either language, the answer is recorded as not found.

Governance disclosure

G1Corporate governance report available in Englishup to 2.86 pts

Score	What earns it
5	The Corporate Governance Report filed with the TSE, in English and current.
4	The same filed report in English, more than a year out of date.
3	A governance narrative written for the website in place of the filing, however current or full.
0	None in English.

The filed report answers a fixed set of questions in a comparable form; a company’s own prose does not, so it scores lower.

G2Director biographies in English with tenure, expertise and independenceup to 2.86 pts

Score	What earns it
5	Every director, with years on the board, career or expertise, and independence stated.
4	All three elements, but one is thin, or given for candidates only.
3	Two of the three.
2	Names and titles only.
0	None in English.

An AGM candidate list counts as covering the board when every sitting director is a candidate or is described elsewhere in English (checked against the governance or securities report). If some directors are covered nowhere in English, it is a 4.

G3Board skills matrix published in Englishup to 2.86 pts

Score	What earns it
5	The current board skills matrix in English.
4	A year old.
3	Partial: directors or skills missing.
0	None in English.

G4Executive remuneration structure and its performance linkage explained in Englishup to 2.86 pts

Score	What earns it
5	The pay structure (fixed, bonus, shares), the performance measures and their weights, and the amounts by type of pay or by director.
4	Structure and measures, but the weights or a breakdown of amounts are missing (a total only).
3	The structure only, or amounts without the performance measures.
2	A mention.
0	None in English.

Amounts by type of pay for the previous year still count for a 5 when a newer figure exists only in the Japanese securities report.
The weights are those of the performance measures (for example operating profit 60%, ROE 40%), for every performance-linked element: bonus and stock alike. An element with a single measure counts as weighted, and so does a weight on a stated group of measures. The split between fixed pay, bonus and shares alone, or weights for one element but not another, is a 4.

G5AGM notice in English with sufficient lead time to voteup to 2.86 pts

Score	What earns it
5	The reference documents for every proposal in English, three weeks or more before the meeting, with an English route to vote.
4	All of that, but the voting website is Japanese-only, or the lead time is about two weeks.
3	Under two weeks, or the English stops at the agenda.
0	No English notice.

A “translated from part of the Japanese original” line on the first page is not scored down; the contents are checked instead.

G6Shareholder vote results disclosed in Englishup to 2.86 pts

Score	What earns it
5	Vote counts and approval ratios for each resolution.
4	Counts without the full breakdown.
3	Counts for some resolutions only.
2	The outcome only: the reader learns it passed, not how.
0	None in English.

G7Cross-shareholdings in English with reduction schedule and rationaleup to 2.86 pts

Score	What earns it
5	Holdings disclosed (number and amount), a reduction target or schedule, progress against it, and the board’s reasoning.
4	Policy, a stated reduction target and progress, without a schedule.
3	Policy and numbers, with no target.
2	Only the standard policy wording from the governance report.
0	None in English.

A 4 needs a stated target, a number or “zero”.
A company that states in English that it holds no cross-shareholdings scores 5.
Deemed holdings (shares in a retirement-benefit trust whose votes the company directs) are in scope: they are scored with the other holdings, against the same policy, target and progress.

English IR website

O1English IR section reachable within two clicks of the English homepageup to 2.14 pts

Score	What earns it
5	English IR one click from the English homepage.
4	Two clicks.
3	Three clicks, or only through a hover menu.
2	English IR on a separate site the English homepage does not link to.
0	No English IR section.

No English pages, but English documents linked from the Japanese IR pages: 2.

O2English document links resolve to English documentsup to 2.14 pts

Score	What earns it
5	Every sampled English link opens an English document.
4	One Japanese document among them, labelled as such.
3	Several, or unlabelled.
2	Frequent.
0	No English IR section.

At least ten English document links are sampled across document types, and the note names them.

O3English IR archive free of dead linksup to 2.14 pts

Score	What earns it
5	Every sampled link is live.
4	One dead link.
3	A few.
2	Many.
0	No English IR section.

The same sample as O2.

O4English IR pages render without blank JavaScript failuresup to 2.14 pts

Score	What earns it
5	Every English IR page renders its content in a browser.
4	A widget fails (a stock chart, an embedded frame), but the content is there.
3	A list or section renders empty.
2	Main pages are blank.
0	No English IR section.

A third-party page (such as a share-price site) that blocks automated browsers is not counted as an IR failure.
No English pages, only English documents linked from Japanese pages: 0, recorded as not found.

O5Single page listing all English IR documents by date and typeup to 2.14 pts

Score	What earns it
5	One library page with every document type, dated and typed (filters count).
4	One dated page, but not every type, or not typed.
3	Split across several pages by type, each dated.
2	Scattered and undated.
0	None.

O6English archive genuinely populated, not an unverifiable shellup to 2.14 pts

Score	What earns it
5	Several years of documents across types.
4	At least a year.
3	Current items only.
2	A shell with a few items.
0	None.

O7Named English IR contact publishedup to 2.14 pts

Score	What earns it
4	An IR contact that is dedicated and immediately obvious from the English IR site.
3	A real IR contact route (dedicated address, phone or IR enquiry form) that exists but is not obvious.
2	A contact named only in a document header; the site offers no IR route.
0	No English IR presence to reach anyone through.

O7 has no 5: it is capped at 4.
What is scored is whether an English reader can reach IR, not whether a person is named.
A shared company enquiry form with IR preselected is not a dedicated route (3).

Japanese–English parity

R1Earnings materials published in English simultaneously with Japaneseup to 2 pts

Score	What earns it
5	The tanshin and the deck in English on the same day as the Japanese, every period read.
4	The tanshin on the same day in most periods read, the deck later.
3	Days to weeks late; or the tanshin on time but no English deck while the Japanese has one.
2	More than a month late.
0	None in English.

The latest two results are read, using the dates the sites list.
A same-day English summary with the full English later counts as late.

R2Median lag in days between Japanese and English statutory publicationup to 2 pts

Score	What earns it
5	Median lag of 0 days.
4	1 to 7 days.
3	8 to 30 days.
2	31 to 90 days.
1	Over 90 days.
0	None of the documents in English.

Five documents: the latest two tanshin, the AGM notice, the Corporate Governance Report and the annual securities report. The note shows each date.
A document with no English version counts as over 90 days. An English document with only the financial statements is not a translation of the securities report.

R3Timely disclosures published in Englishup to 2 pts

Score	What earns it
5	Every Japanese timely disclosure in the past twelve months has an English version the same day.
4	One or two missing or late; or the English exists only on TDnet, the same day.
3	About half, or only selected types.
2	A few.
0	None in English.

Twelve months of the Japanese list (forecast and dividend revisions, buybacks, M&A, management changes) are compared with the English list.

R4English disclosure free of machine-translation accuracy disclaimersup to 2 pts

Score	What earns it
5	Nothing on the English site or in its documents is presented as machine-translated.
4	A general website-translation notice on the disclaimer page only.
3	Part of the site is machine-translated and says so (for example a site-wide pop-up), or a few non-statutory documents carry a machine-translation disclaimer.
1	A core statutory document is disclaimed as a machine translation of uncertain accuracy.
0	No English.

“The Japanese original shall prevail” is about which language governs, and is scored under R5, not here.
A translation widget that machine-translates into other languages only (not English) scores 5.
Checked on the rendered page, not in the page source.

R5English version stated as authoritative, not subordinate to Japaneseup to 2 pts

Score	What earns it
5	English stated as authoritative, or filed as an original.
3	Silent: no statement either way.
2	A “for reference only, the Japanese prevails” line on some documents.
1	That line on every document.
0	No English.

These are categories, not degrees, so there is no 4.

See how every company scores on each question →

Change history

Changes to the rules, and the re-checks they caused, newest first. Corrections to individual scores are recorded on each company’s page.

Date	Change
21 October 2026	v1.1.1, after an independent audit. Same questions and weights. English documents count wherever they sit on the company’s site, including when linked only from Japanese pages; a company without English pages is scored on every question, and the missing pages count only in O1 (2) and O4 (0). G4: a 5 needs the weight of each performance measure, not only the pay mix. G2: an AGM candidate list covers the board when every sitting director is covered in English. G7: deemed holdings in a retirement-benefit trust are in scope. Scores affected were re-checked and logged as rule changes.
20 October 2026	v1.1 frozen as the public baseline. Fourteen Japanese question wordings brought into line with the English rubric (above all R5, which now asks whether the English is authoritative, as the English always did). On the capital-policy review, a score out of 100 is shown only from three reviewed actions.
16 October 2026	Rulings on dates (the file creation date governs when the English site reuses the Japanese dates) and on summary-only tanshin for R1 and R2.
15 October 2026	Consistency rulings recorded for S5, R1, R2, R3, R4, F6, F7, O5, O7, S1, S7, G4 and G7, so that every company is scored the same way.
14 October 2026	Bands written down for the remaining 24 questions before they were scored across the cohort.
6 October 2026	O7 scored on whether an English reader can reach IR, not on a named person, and capped at 4.
1 October 2026	Bands written down for the first ten questions, with the 4-versus-3 test. Earlier scores that the written rule contradicted were re-checked and corrected (G1, G5, F1, F4, S2, S6).
25 September 2026	R4 clarified: “the Japanese original shall prevail” concerns which language governs and is scored under R5.
20 September 2026	v1.1 question set rebuilt from the recorded areas, weights and amendments after the original file was lost. The data-correctness deduction is suspended.

For the thinking behind the scores see Methodology; for who runs the Index and how it is funded, Editorial policy and funding.