Mk.
Pre-release · research as of 20 October 2026

Falcom article review and reproducibility

Status: pre-release draft, not published. Author: Mk. Research date: 2026-10-20.

Inputs and execution

  • Research repo: llmk51ll/investment-research.
  • Requested source branch: refine/falcom-v9-issuer-calibration.
  • Pinned source SHA: 4c73b5cc7f34be7a2a8e491ee6d35e8e9f7e9143.
  • Read the supplied website brief, source documents and Stage 1 decisions before drafting.
  • Windows python command resolved to a nonworking alias; used the bundled Python interpreter. No dependency installation or model changes required.
  • Before writing: python -m unittest discover -s 3723-Falcom/model -p "test_*.py" — 21 tests passed.
  • Before writing: from 3723-Falcom/model, python cohort_model_v9_issuer_calibrated.py — exit 0.
  • All four §3 scenario rows match the brief at its stated precision. See model-check.json.

To regenerate from an existing checkout at the pinned SHA:

python content/research/falcom-3723/compose.py /absolute/path/to/investment-research

The source checkout must be trusted and unmodified. The generator checks HEAD; a SHA check does not detect uncommitted changes. Only the requested v9 source file is now copied into this website repository; the generator still executes the complete pinned research checkout because the copied file is not standalone.

Differences and precision

  • §3 valuation table: no differences in all 12 reported cells (four means and eight per-share values).
  • Net cash is 78.18 at two decimals versus the brief's rounded 78.2. The model uses its unrounded figure in calculations.
  • Q4 residual computes to 4.86 at two decimals, versus report §5's 4.87 and brief's approximate 4.9. It is the model's implied Q4 other revenue less the mean FY26 Q1–Q3 implied other revenue; underlying precision causes the difference.
  • The separate fixed-Base FY26 forecast gap is 7.88. It does not replace the Q4 run-rate residual and is explained in both language versions.
  • Rounded bridge remains 3,026 +25 +152 −76 = 3,127; exact calculations precede rounding.
  • Quarterly table is generated from the CSV, not copied from the report's rounded table. FY25 H1 total is 6.04, FY25 Q3 total 4.36, FY25 Q4 total 15.73. FY26 Q4 implied revenue is 20.96. FY26 Q2/Q3 categories are labelled Derived because they subtract cumulative totals.

Publication boundary and remaining gates

  • A separate local homepage-copy branch edits homepage copy: removes its investment disclaimer, changes only the requested English sentence, adds the AI note and rewrites Traditional Chinese in formal written language. Discord remains. No navigation, build configuration or publish allowlist changes.
  • Cloudflare's five-file prepare-publish.cjs allowlist excludes this entire directory. A branch build cannot serve this article through that output.
  • PR only. Do not merge or publish as part of this request.
  • Publication remains gated on the FY26 results update, actual cash, FY27 guidance, public model/version-tag readiness, and the roadmap's fresh independent red-team review. This editorial check is not an independent fresh-session review.
  • Website EN/繁中 integration is deferred: one Markdown article contains English first and the equivalent Traditional Chinese version. No live language-toggle integration is claimed.
  • English and Chinese article model links and the source ledger now point to the copied relative file; no private GitHub model link remains. The original public version tag is still pending the November update.

Source-file inspection and requested disclosure update (2026-10-20)

  • Inspected the entire requested source before copying: no personal contact information, owned-position sizes, entry/exit plans, or identifiable individual names found. Corporate/publisher names (Falcom, Konami, NISA, SIE, GungHo, CLE and others) are research counterparties, not individual names.
  • The source includes anonymous investor decision / approval comments and dated model-assumption notes. These are retained unchanged; they are not personal trading plans.
  • It also retains superseded v8 reference code (v8_style_future_value, a flat future-cash hypothesis) and historical valuation discussion. These remain source-history references, not the article's valuation or an endorsed new output. The article continues to use current scenario outputs.
  • Copied ONLY model/cohort_model_v9_issuer_calibrated.py from exact Git blob at 4c73b5cc7f34be7a2a8e491ee6d35e8e9f7e9143, unchanged. SHA-256: ce18c346357a70300e2ae6122dcc063384e7e17c7a55ea607a2750647546e26f. No archive modules, private handoffs, or other files copied.
  • Important limitation: imports require archive.cohort_model_v5_rolling_catalogue, v6, v7, and cohort_model_v8_fy2026_reconciliation; this single source file cannot run independently. The article explicitly states this. Full-checkout tests/model execution remain the reproducibility evidence; this is not a standalone public package.
  • Replaced both pending holdings placeholders with Mk's supplied dated disclosure near the top of each language: no Falcom position as of 20 October 2026; may buy later; will update if it changes. This is the author's statement, not independently verified account information.
  • On that separate homepage branch, the English homepage introduction is unchanged except that work to those; the supplied bilingual AI note is added as its final line. The Falcom Traditional Chinese narrative is rewritten formally with scenario numbers unchanged.
  • Neither PR #1 nor PR #2 is merged. PR #1 remains blocked on the November results update; PR #2 awaits Mk's Cloudflare setup.

Publication authorization and static release — 2026-10-20

Mk subsequently explicitly requested publication of the currently previewed website. This overrides the earlier PR-only release hold for the pre-release article and homepage; FY26 results, the public-version tag and model updates remain pending November. This page records historical review gates above, not a current claim that the article is unpublished.

The release adds a reproducible Markdown-to-static-HTML build and section/article routes, keeping all current numerical tables unchanged. Home uses the approved copy; only research pages show the investment disclaimer. Comments display an unavailable notice only: PR #2, the Pages Function, database and Turnstile setup are not included or activated.

Personal research, not investment advice. — Mk