Company Overview
ROXX is a small Tokyo Growth-market HR tech company built around Z Career, a recruiting platform for non-desk workers in areas such as manufacturing, construction, transport, retail, and service. Its target user is not the affluent white-collar job switcher; it is the younger, lower-income, often non-regular worker trying to move into full-time employment. That niche matters, because it partly explains both the opportunity and the fragility.
The authoritative supplied snapshot is anchored to FY2025, the year ended 2025-09-30 and filed 2025-12-25, with market data as of 2026-07-23. I also found newer official data: H1 FY2026, period ended 2026-03-31, disclosed 2026-05-14. The latest clean official annual base is FY2025. More recent data is partial and unaudited.
Current market data in the supplied snapshot: share price ¥399 and market capitalization ¥2.902bn. The supplied deterministic snapshot leaves net cash/debt as unavailable, so I do not replace that field. On audited FY2025 official numbers, revenue was ¥4.513bn, operating loss was ¥721.9m, and ordinary loss was ¥767.2m. Reported FY2025 net income was +¥1.051bn, but that was not clean operating earnings: it was created by a ¥1.941bn extraordinary gain from selling the back check business. On newer official H1 FY2026 data, which is partial and unaudited for investment purposes, revenue was ¥1.876bn, operating loss ¥479m, and net loss ¥558m.
The short version: this is a company with a real niche, some growth, and a balance sheet that is not yet broken, but the clean earnings base is still negative. That is the right place to start.
How the Company Makes Money
ROXX operates a hybrid of marketplace, recruiting workflow software, and human advisory. Z Career serves three parties at once: job seekers, hiring companies, and partner recruitment agencies. Hiring companies can post jobs without an upfront listing fee and pay mainly when a hire is made. That lowers employer friction and pushes ROXX toward outcome-based revenue rather than subscription-only revenue.
The economic engine has three main pieces. First, ROXX earns performance revenue when a successful hire is made through its own channel. Second, it earns administrative or platform-related fees around that matching process. Third, partner agencies using the platform pay recurring usage fees, and the company is trying to add new recurring revenue through AI-related products such as AI interviewing and workflow tools.
The mix matters. In the official FY2025 business overview, management explains that when matching is handled through a partner agency, ROXX captures only a modest take rate, roughly around the high-teens; when matching is handled through its own direct channel, it captures far more of the economics. That means value creation depends heavily on channel mix, advisor productivity, and candidate acquisition efficiency.
Capital intensity is low in the traditional sense. This is not a factory business. Sustaining capex is small, and the formal capex line is not the central issue. The real capital intensity is commercial: hiring and training advisors, buying or engineering candidate traffic, and funding the working-capital strain that comes with growth and turnover in staff. In other words, ROXX can look asset-light and still be cash-hungry.
The business has a few real sources of durability. The niche is underserved because low-income, non-desk placements are less attractive to many traditional recruiters. ROXX also has a platform of roughly 400 partner agencies and a growing pool of historical selection data. But this is not yet a proven high-return compounder. The moat is still modest. Candidate acquisition channels are not proprietary, advisor productivity is not yet fully standardized, and the company has not yet shown that it can turn growth into repeatable free cash flow.
Why the Stock Fell
The stock is down because the market moved from valuing a growth story to valuing a still-unproven business model. Since listing in September 2024 at ¥2,110, the shares have fallen to ¥399 as of 2026-07-23, an 81% decline. That is not one bad quarter. It is a rolling reassessment of what the business is actually worth.
| Date | Observed event | What it told the market |
|---|---|---|
| 2024-09-25 | IPO on TSE Growth at ¥2,110 | The market initially paid for high growth and platform optionality. |
| 2025-08-13 | Announcement to sell the back check business for about ¥1.9bn | Focus improved, but future reported profit would be flattered by a one-off disposal gain and the long-term revenue story would be reset. |
| 2025-12-25 | Audited FY2025 filing: revenue ¥4.513bn, operating loss ¥721.9m, ordinary loss ¥767.2m, operating cash flow -¥1.024bn, net income +¥1.051bn | The “profit” was low quality. The core business was still loss-making and cash-burning. |
| 2026-02-12 | Q1 FY2026 official disclosure: revenue ¥926m, operating loss ¥314m | The post-sale business still needed heavy investment and near-term losses widened. |
| 2026-05-14 | H1 FY2026 official disclosure: revenue ¥1.876bn, operating loss ¥479m, net loss ¥558m, operating cash flow -¥866m, cash down to ¥2.297bn | The company kept full-year guidance intact, but the back-half recovery requirement became very demanding. |
The dominant market narrative is straightforward: FY2025 earnings were optical, H1 FY2026 still burned real cash, and the promised H2 inflection may be too aggressive. That narrative is substantially grounded in facts, not just fear.
What the Market Is Assuming
| Concern | What the market is likely assuming | Quantitative check | Evidence quality |
|---|---|---|---|
| Front-loaded losses are temporary | H1 FY2026 losses reflect deliberate hiring and advertising investment that should convert into H2 revenue. | Official standalone quarterly operating loss improved from -¥314m in Q1 to -¥165m in Q2. That is real improvement, though still negative. | Good for the quarter trend; still partial. |
| FY2026 guidance is too hard | The market doubts the company can actually hit its unchanged full-year guidance. | After H1 FY2026, ROXX still needed ¥3.124bn of H2 revenue and ¥524m of H2 operating profit to reach the company’s full-year guidance of ¥5.0bn revenue and +¥45m operating profit. H2 FY2025 total revenue was ¥2.533bn and H2 FY2025 operating profit was only ¥61m. | Strong. |
| Reported H1 revenue decline means demand is weak | The market may read the official -5.2% H1 revenue change as a core business slowdown. | That reading is too simple. The prior-year base still included back check. Using management’s Z Career-only quarterly disclosures, core revenue was +2.2% in Q1 and +28.2% in Q2. By my arithmetic on those two management updates, Z Career H1 revenue grew roughly 14% year on year. | Moderate; company update, not audited segment accounts. |
| Moat is thin and acquisition costs are fragile | Even if demand exists, ROXX may not own durable economics. | Official FY2025 revenue was 85.9% concentrated in Z Career. Management says SNS rose to roughly 80% of acquisition mix and interview unit cost improved 25.2% year on year in Q2, but ROXX does not disclose a full audited cohort-level LTV/CAC bridge. | Mixed; some support, major evidence gap remains. |
| Growth story moved further out | The business is less diversified and the old scale narrative was pushed right. | After selling back check, management moved the timing of its ¥10bn revenue aspiration from FY2027 to FY2029. | Company guidance / management update. |
My read is that the market is assuming three things at once: first, the H1 pain is partly temporary; second, the H2 recovery hurdle is high enough that guidance deserves a discount; third, the long-term economics remain unproven even if the niche is real. That is a fairly rational set of assumptions.
Temporary or Structural?
| Plausible risk | Damaged economic mechanism | Realistically reversible within 3 years? | Classification |
|---|---|---|---|
| H1 hiring and advertising losses | Near-term reported profit and cash flow | Yes. If new advisors become productive on schedule, this is repairable. | Not truly structural |
| Candidate acquisition cost inflation and channel dependence | Unit economics, payback period, operating leverage | Partly. ROXX can improve mix and conversion, but it does not control the traffic environment. | Structural but survivable |
| Heavy concentration in Z Career after selling back check | Earnings resilience and diversification | Only partly. New AI products may help, but not quickly enough to change concentration today. | Structural but survivable |
| AI disintermediation of human recruiting | Take rates and the need for human advisors | Unclear. In ROXX’s user segment, human trust still appears important, but if self-serve matching works well enough, margins can compress permanently. | Structural damage if realized; not yet proven |
| Regulatory or license failure | Right to operate | No. If the placement license were impaired, the damage would be immediate. | Structural damage, low probability today |
| Future dilution | Per-share value capture | Yes, but only through profitability or non-dilutive funding. | Structural to shareholder outcome, not to customer demand |
The right diagnosis is not “temporary” or “structural” in isolation. The earnings valley is temporary. The question of whether ROXX has durable, scalable, cash-generative economics is structural. I would classify the situation as structural but survivable: the company likely survives, but the quality of the franchise is still being tested.
The time-as-a-moat test is not flattering. In two years, a well-funded competitor could copy much of the front-end offer, buy traffic, and hire recruiters. In five years, a large incumbent could likely recreate most of the workflow stack if the economics prove attractive enough. In ten years, a real moat would require data-driven matching advantages, sticky partner relationships, and measurably better employer ROI. ROXX may eventually build that, but it has not yet proved it. Today the moat is niche position plus execution know-how, not a hard barrier.
Is the Market Wrong? By How Much?
The supplied worth-dead and worth-alive snapshots are useful, but only if handled skeptically. On the asset side, NCAV is ¥1.608bn, or about ¥221 per share, which is only 55.4% of the current market capitalization. That means ROXX is not a classic net-net. Book value is ¥1.955bn, or about ¥269 per share, and the stock trades at 1.48x book.
On the earnings side, the supplied model says owner earnings are ¥922m, implying a 31.8% owner-earnings yield and a 3.15-year payback. I do not think that is decision-grade as a continuing-earnings proxy. Why? Because audited FY2025 net income was driven by a one-off disposal gain, while the same supplied model shows free-cash-flow conversion of -98.8%. Those two facts together are a warning, not confirmation.
So the right valuation is a hybrid: use asset value as the floor, then layer in only a cautious amount of continuing-business value. My valuation base is FY2025 audited results adjusted with H1 FY2026 official updates. Per-share values use 7,273,170 shares, the count implied by the supplied market-cap snapshot, so the per-share bridge reconciles to the supplied current market data.
Moat-and-mispricing score: 4/10. There is a real niche and some upside if H2 execution works. But the stock is no longer obviously mispriced once you strip out the optical FY2025 earnings.
| Case | Earnings / owner-earnings base | Normalization | Required yield | Balance-sheet adjustment | Intrinsic equity value | Intrinsic value / share | Vs. current ¥399 |
|---|---|---|---|---|---|---|---|
| Bear | No dependable positive continuing owner earnings | Turnaround fails; value anchored to asset floor and likely further burn | n/a | Implicit in asset-floor approach | ¥1.5bn to ¥1.8bn | ¥205 to ¥245 | -49% to -39% |
| Base | My estimate: ¥280m to ¥330m operating owner earnings | Strips the FY2025 disposal gain; assumes Z Career reaches modest real operating leverage but not a heroic margin | 14% to 15% | Plus roughly ¥0.3bn to ¥0.4bn of estimated net cash from the 2026-03-31 official balance sheet | ¥2.3bn to ¥2.8bn | ¥315 to ¥385 | -21% to -4% |
| Bull | My estimate: ¥430m to ¥520m operating owner earnings | Assumes advisor productivity improves, hiring investments pay off, and AI tools add genuine recurring revenue | 11% to 12% | Plus roughly ¥0.3bn to ¥0.4bn of estimated net cash | ¥4.0bn to ¥5.0bn | ¥550 to ¥690 | +38% to +73% |
The bear case is easy to see: the stock can drift toward the NCAV zone if H2 misses, cash burn continues, and the market stops underwriting a successful scale-up. The bull case is also real: if ROXX proves that H1 losses are just the cost of building productive advisor capacity and that AI meaningfully lowers training and acquisition friction, current valuation can look too low. But the base case still sits below the current market price.
That leads to the central conclusion: the market is not obviously wrong. It has already crushed the IPO narrative, but it is still paying for a meaningful chance that ROXX converts a real niche into real earnings. At ¥399, that is not crazy, but it is not a cigar butt either.
Key Facts, Estimates, and Judgments
| Item | Value | Type | As of |
|---|---|---|---|
| Core business | Z Career recruiting platform for non-desk workers; single-segment, standalone reporting | Official fact | FY2025 annual filing |
| Latest clean annual base | FY2025, filed 2025-12-25 | Official fact | 2025-12-25 |
| Newer official data | H1 FY2026, revenue ¥1.876bn, operating loss ¥479m, net loss ¥558m | Official fact, partial / unaudited current-year data | 2026-05-14 disclosure |
| Current share price | ¥399 | Current market data | 2026-07-23 |
| Current market capitalization | ¥2.902bn | Current market data | 2026-07-23 |
| FY2025 audited revenue | ¥4.513bn | Official fact | FY2025 |
| FY2025 audited operating result | Operating loss ¥721.9m; ordinary loss ¥767.2m | Official fact | FY2025 |
| FY2025 reported net income | +¥1.051bn | Official fact | FY2025 |
| One-off item distorting FY2025 earnings | Extraordinary gain of ¥1.941bn from sale of back check | Official fact | FY2025 |
| NCAV | ¥1.608bn; about ¥221/share | Supplied model estimate | Period ended 2025-09-30 |
| Book value | ¥1.955bn; current P/B 1.48x | Supplied snapshot / current market ratio | Book: 2025-09-30; market: 2026-07-23 |
| Owner earnings | ¥922m; owner-earnings yield 31.8%; FCF conversion -98.8% | Supplied model estimate | Period ended 2025-09-30 |
| Estimated balance-sheet adjustment used in valuation | Roughly ¥0.3bn to ¥0.4bn net cash | My estimate from official H1 FY2026 cash and borrowings | 2026-03-31 |
| Potential option dilution | 695,500 shares, about 9.6% of shares outstanding | Official fact | FY2025 annual filing date |
| Recent tracked super-investor filings | None found in supplied context | External supplied context | As supplied |
| Overall thesis | Real niche, weak proof of moat, earnings still optical, valuation no longer obviously cheap | Judgment | Current |
- Strongest falsification check: FY2026 needs a dramatic H2 improvement. If Q3 does not show clear operating leverage, the base and bull cases weaken quickly.
- Cash conversion check: If official operating cash flow remains deeply negative even after the hiring cohort matures, the “owner earnings” case is overstated.
- Core-demand check: If management’s continuing-business growth narrative does not show up in future official revenue and margin progression, the niche thesis is weaker than it looks.
- AI reality check: More customer announcements are not enough. What would change the thesis positively is disclosed, recurring, material AI revenue with limited extra headcount.
- Dilution check: An equity raise or material option dilution would lower per-share value and push the stock closer to the bear case.
- What would make me more constructive: two consecutive quarters of positive operating profit excluding one-offs, better cash conversion, and evidence that advisor productivity can rise without equally rising acquisition spend.