Falcom 3723: a launch peak, a catalogue, and the earnings between them
By Mk · Research as of 20 October 2026
Disclosure: As of 20 October 2026, I hold no position in Falcom (3723). I may buy shares in the future. I'll update this note if that changes.
Pre-release — numbers will be updated after FY2026 results (~Nov 2026). This is a review draft. The results date around 11 November is a third-party schedule in the supplied ledger, not a confirmed issuer calendar date. FY2026 revenue of ¥43 oku and EBIT of ¥28 oku are revised forecasts, not actual results.
Units: ¥1 oku = ¥100 million; operating amounts below are in ¥ oku unless stated. Per-share scenario values are in yen. Fiscal years end in September.
Evidence labels: Disclosed = issuer filings; Derived = arithmetic from disclosed figures, including back-solving printed YoY percentages; Inference = interpretation; Assumption = model input. A model estimate is not an issuer disclosure. Sources and their verification limits are in the source ledger.
The short version
What the market may miss — Inference: simultaneous releases can move Japanese shipment revenue into the launch quarter while overseas royalties arrive later. A strong launch year and a quieter following year can belong to the same title economics. The useful question is how much earnings power remains across the cycle.
Three ways this can go wrong:
- Inference / Assumption: “other” revenue may include fading old-title tails and episodic licence payments; treating it as a stable floor could overstate earnings or overlap named-title tails.
- Assumption: hiring and development costs may rise faster than the model, or the assumed major-plus-medium annual release cadence may slip.
- Inference: the unexplained FY26 Q4 amount may be revenue pulled forward through minimum guarantees (MGs), rather than stronger lifetime sales. A lower earnings multiple may also be appropriate after the launch peak.
Model scenarios, not a forecast: across Bear to Bull, the per-share ranges are ¥2,552–3,459 at 10× and ¥2,200–2,926 at 8×. Both multiples matter equally; neither is a selected outcome.
How Falcom makes money
Disclosed: Product covers Japanese packaged sales and related goods; domestic shipment revenue is recognised on shipment. Licensing includes domestic digital downloads, overseas rights, mobile titles and ports. An overseas MG and royalties recouped against it must not be counted twice. These are revenue categories, not separate reporting segments.
Inference (~85% confidence, an analyst judgement rather than a statistical probability): Konami is mainly the Japanese packaged Product channel. The customer amounts track Product much more closely than Licensing; the supplied research finds Konami/Product ratios of about 79–88% across the sampled years. The issuer does not explicitly identify Konami as the distributor, and its customer table does not assign the customer to either revenue category. Domestic digital sales belong to Licensing, with SIE appearing as a customer. The old claim that Konami proves licensing concentration is therefore not retained as an established bear argument.
Why FY26 looks like a launch-quarter peak
The following table is generated from the supplied quarterly CSV. FY25 rows use rounded YoY-derived amounts; FY26 Q2 and Q3 are differences between disclosed cumulative totals, so they are Derived, even where the report uses a broader “official” label. H1 is six months, not a single quarter. Totals are sums of the rounded category inputs and may differ from more precise issuer totals.
| Period | Product | Licensing | Total | Evidence class |
|---|---|---|---|---|
| FY25 H1 | 1.23 | 4.81 | 6.04 | Derived |
| FY25 Q3 | 0.54 | 3.83 | 4.36 | Derived |
| FY25 Q4 | 5.55 | 10.18 | 15.73 | Derived |
| FY26 Q1 | 0.49 | 8.88 | 9.37 | Disclosed, rounded |
| FY26 Q2 | 0.67 | 5.31 | 5.98 | Derived |
| FY26 Q3 | 1.78 | 4.90 | 6.68 | Derived |
| FY26 Q4 | — | — | 20.96 | Derived from revised forecast; not actual |
Inference: FY25 Q4's Product jump around Sky1 and the FY26 Q3 pre-launch shipment pattern ahead of Kyoto are consistent with shipment-based recognition. v9 therefore puts Japanese first-year economics at launch, rather than spreading them across the first year. This is a timing approximation: treating all first-year Japanese economics as Product does not prove that Japan has no digital sales, and the model does not reproduce every pre-launch shipment date.
Derived / model: the FY25 Q4 implied non-named revenue residual falls from 8.40 under the old timing to 4.02 under v9. Lifetime economics are unchanged. The remaining FY26 Q4 residual above the nine-month run-rate is approximately 4.86. Stronger Sky2/Kyoto sales, MG timing and one-off licences are possible explanations, all Inference; none is a disclosed breakdown.
The model's fixed annual Base has a separate FY26 gap of 7.88 to the revised revenue forecast. That is not the Q4 residual: the latter first uses actual-minus-model non-named revenue for the first nine months and extrapolates its quarterly run-rate. The two calculations answer different questions.
Three drivers, with assumptions kept visible
- Other-revenue level — Assumption: Base is 13.5/year: catalogue licensing 9.0, localisation/ports/new licences 2.5, other Product 2.0. The model-derived FY25 non-Sky1 amount is 14.42; the FY26 nine-month non-named run-rate is about 16.5 annualised. These are calibration references, not proof of a recurring floor. Named titles already carry their own long tails. The −5%/year risk version is shown alongside Base.
- Cost path — Assumption: five additional hires each year, cumulatively, at ¥7.5 million fully loaded per person-year. Costs become 15.875 / 17.250 / 18.125 in FY27–29. This is an investor input, not company hiring guidance, and R&D is not deducted a second time.
- New-title cadence — Assumption: one major and one medium title each year. Sky1 lifetime units of one million and Sky2 of 750,000, model receipts of ¥5,400 / ¥3,500 per copy and an overseas MG share of 25% remain assumptions, not confirmed sales or contract terms. The model-derived customer ceiling of 22.7% creates tension with 25%; the 20% MG sensitivity tests timing without claiming to know contracts.
Valuation: two multiples side by side
Model / Assumption: scenario value today = (multiple × mean FY27–29 EBIT × 100,000,000 + estimated excess cash × 100,000,000) ÷ 9,867,090 shares. The share count is Derived from disclosed issued shares less treasury shares. The model's estimated 2026-09-30 cash input is used for the 2026-10-20 research date; it is not a disclosed year-end balance.
| Scenario | FY27 EBIT | FY28 EBIT | FY29 EBIT | Mean EBIT | 10× | 8× |
|---|---|---|---|---|---|---|
| Base: other 13.5, +5 hires/year | 23.59 | 24.32 | 21.19 | 23.03 | ¥3,127 | ¥2,660 |
| Risk: other revenue declines 5%/year | 22.92 | 23.00 | 19.27 | 21.73 | ¥2,994 | ¥2,554 |
| Bear | 18.59 | 18.82 | 14.69 | 17.37 | ¥2,552 | ¥2,200 |
| Bull | 26.87 | 27.60 | 24.47 | 26.32 | ¥3,459 | ¥2,926 |
Scenario definitions — Assumption: Bear combines other revenue of 9.0 with costs above Base by 0.5 / 1.0 / 2.0 in FY27–29. Bull combines other revenue of 15.0 with a 33% uplift to first-year Japanese units for all titles. The risk version applies a 5% annual decline from the FY26 other-revenue base, including a first decline in FY27.
Model bridge at 10×: ¥3,026 → +¥25 → +¥152 → −¥76 → ¥3,127. The steps are the v8 reference under the current cash convention, recognition timing, other revenue rising to 13.5, and cumulative hiring. Rounded steps can create rounding differences. The article does not reuse the withdrawn future-cash method. Earnings enter once through the multiple; future earnings are not also rolled into the cash balance.
| Component | ¥ oku |
|---|---|
| Disclosed cash, 2026-06-30 | +108.00 |
| Derived Q4 net income: revised full-year forecast less nine months | +10.11 |
| Forecast dividend cash outflow | -5.92 |
| Assumption: working-capital timing | -4.00 |
| Assumption: operating cash reserve | -30.00 |
| Estimated excess cash used in valuation | 78.18 |
The cash bridge uses forecast Q4 net income as an approximation, not a cash-flow statement. The dividend is a company cash outflow; it is not added to the scenario as a payment available to a new buyer after the record date. The reserve and working-capital deduction are assumptions. A lower multiple reflects the possibility that post-FY29 earnings weaken; 10× and 8× remain parallel assumptions.
What we do not know, and what comes next
- Not public: per-title sales, Japanese physical versus digital sales, and title-specific MG amounts, licence dates, royalty rates and recoupment terms.
- Open inference: Konami's precise role and the FY26 Q4 residual. Strong launch sales cannot be separated from accelerated recognition with the present disclosures.
- Next: absorb FY26 results and FY27 guidance around November, replace estimated year-end cash and Q4 income with actuals, and rerun the quarterly backtest. The supplied research uses roughly 8.0 of FY26 Product as a thesis check, not company guidance. A large shortfall would challenge launch timing; weaker FY27 guidance alone would not prove permanent deterioration.
- Then: inspect the FY26 securities-report customer table, expected around December, including Konami, NISA, GungHo and CLE. Revisit MG limits, other-revenue composition and hiring assumptions. Expected dates remain provisional.
Sources, model version and disclosure
See the source ledger, including the 7 October revised forecast, FY26 Q3 filing, and the reproducibility record. Draft tables are generated from the v9 model at research commit 4c73b5cc7f34be7a2a8e491ee6d35e8e9f7e9143; the pinned model is copied unchanged into this article’s model folder. This source snapshot depends on modules from the research repository and is not a standalone executable.
Public version pending: falcom-3723-v9-public-YYYYMMDD will be created only after the FY26 results update and model freeze. It does not exist yet; no tag or public model link is invented here.
Much of the writing here is drafted with AI assistance. The research, judgements and final edits are mine.
Personal research, not investment advice. Figures are scenarios built on stated assumptions.
Comments
Comments are not open yet.