2317 — Systena Corporation

Deep-value: 2317 — Systena Corporation (株式会社システナ), TSE Prime

Revisit (REFRESH) of the analysis from 16.08.2026. Reference price: … (close of 04.09.2026). Fiscal year: April 1 – March 31. Accounting standard: JGAAP. Japanese issuer with no sponsored ADR — there is and should be no SEC filing; the EDGAR step is intentionally skipped, as with a BVB ticker.

Symbol collision — reconfirmed. 2317 is Systena on the TSE and Hon Hai Precision / Foxconn on the TWSE. Radu’s tracker (Pregatire_investitii_21.xlsx) gives Exchange = TSE, and query_tracker.py indicatori 2317 returns name=Systena Corporation, exchange=TSE. The Yahoo symbol used throughout this report is 2317.T. The brief research-2317-20260816.md (about Foxconn) remains void and was not a source for any figure.

What I inherit and what I re-derived

The regime is REFRESH. The mechanical triage (delta-2317-20260906.md) called for a revisit not because the thesis broke, but because all 6 falsifiers came back “unverifiable”: the closed triage vocabulary is fed from yfinance, and for a Japanese issuer that reports quarterly only on a cumulative basis, yfinance has no gross margin, operating margin, net debt/EBITDA, or quarterly revenue growth. It isn’t a signal about the company, it’s a data coverage gap. Today’s data pack (data-pack-2317.T-20260907.md) confirms it: the “Cash flow — quarterly” section literally reads “(data unavailable),” and the quarterly income statement contains four lines, two of which are EPS.

Inherited unchanged, because no new filing touched them: the business description and segment structure, the thesis architecture (mix-shift toward high-margin engineering, net-cash balance sheet, EPV ≈ price), the three pre-mortem scenarios, and the controlling-shareholder position. There is no new filing after the thesis date: the last earnings document remains the Q1 FY3/2027 短信 (tanshin) from 30.07.2026, and the next reporting is scheduled for 29.10.2026 (Q2). I checked the issuer’s IR list directly: after 16.08.2026 there is a single release, the one from 19.08.2026 (Supermicro), covered below.

What I re-derived, and why it was worth it. The August report closed with six explicitly stated gaps, the first of which read: “有価証券報告書 (the full annual report, filed 23.06.2026) was not consulted… Marked gap — first item to verify in the next analysis.” I consulted it. I downloaded the full annual report (106 pages, 20260623r07.pdf), the prior annual report (FY3/2025, 105 pages, for the third balance-sheet year), the extraordinary report from 26.06.2026, and the two releases issued after the thesis date. The result isn’t cosmetic:

Plus three things the old report had declared impossible to verify from the outside and that are now verifiable from 有報: the segment map restated for both years (so comparable), the FY3/2025 buyback quantified at 9,912 mil., and the full terms of the March 2026 option grant — including a share-price hurdle the board set for itself, at ….

The price move (438 → 453, …) is context, not motive. The thesis below does not rest on it.

Executive summary

The thesis, reaffirmed and better documented. Systena is an independent Japanese systems integrator — no keiretsu parent, founded in 1983 by 逸見愛親, who still controls it with 29.03% — that has shifted its center of gravity over three years from hardware resale and programmer staffing toward high-margin software engineering: SDV for Japanese automakers, “execution” PMO, and modernization of core banking and insurance systems. The figures describing the shift, all from primary filings: gross margin 23.50% → 25.09% → 26.28% over three years, operating margin 12.62% → 14.43% → 16.28%, ROE 20.0% → 24.0% → 31.4%, F-Score 9/9, capex 0.51% of revenue, zero capitalized or expensed R&D, zero goodwill, zero impairments, no client above 10% of revenue [有報 FY3/2026, p.1, 52-53, 73-74].

What’s newly known, and changes the numbers. The “other current assets” line of 3,738 mil., which in August remained unbroken-down (26% of the year’s net profit, flagged as the first thing to verify) contains 預け金 2,997 mil. — funds placed on deposit, which the issuer itself includes in its “cash and cash equivalents” reconciliation: 27,377 − 555 (deposits over 3 months) + 2,997 = 29,819 mil. [有報 FY3/2026, p.62]. Hence net cash isn’t … but … (27,377 cash and deposits + 2,997 funds on deposit + 2,356 short-term securities − 1,601 gross debt), i.e. … per share, 19.2% of the 453 price. The difference versus the August assumption is … and is a FACTUAL move, not an opinion: the figure appears in the issuer’s own note, not in my judgment.

The thesis’ central point holds unchanged, and that’s the conclusion that matters. The earnings-power value with zero growth (Greenwald EPV, normalized EBIT 14,700, r 8%, plus the corrected net cash) is … versus a price of 453 — a margin of …. In August, with the then-current figures, it came out 434 versus 438, i.e. …. Two independent re-derivation rounds, with a cash base 3 billion larger and a price 15 yen higher, give the same result: at today’s price you buy current earnings at parity and get the growth plan for free. That stability is itself information.

The counter-thesis, tested by a real quarter. FY3/2026 was partly non-repeatable: Business Solutions (37.7% of revenue) captured the PC-replacement wave from Windows 10’s end of support, a wave management itself describes as “一巡” (having run its course). FY3/2027 guidance is revenue …, operating profit …, net profit …. But Q1 FY3/2027 measured exactly this risk: Business Solutions … revenue and … operating profit after the wave was consumed [短信 Q1, p.2-3]. The air pocket didn’t open. And the 6.0% net profit decline doesn’t come from the business: 840 mil. is the new, non-cash option expense, and the rest is the non-repeat of 778 mil. of net financial income from FY3/2026. On the basis management itself tracks — EBITDA+S — guidance is ….

Estimated value. Five triangulated models, all at the price of 453:

Monte Carlo (20,000 scenarios, same base assumptions): median … MOS, median intrinsic value …, probability of undervaluation …%.

Verdict: BUY — average position, not maximum. August’s verdict holds, but the reason for “average” has changed in part. It’s no longer the valuation, and it’s no longer August’s nepotism-governance suspicion either: 有報 explicitly states the director sharing a family name isn’t related to the founder. The real reasons are three, all documented below: (1) dynastic succession — since April 2026 the president/CEO is 逸見真吾, the founder’s son, with 16,300 shares, i.e. 0.0046% of the company; (2) 29.03% of capital in a single person’s hands, through a vehicle registered at an apartment; (3) 3 of 4 outside directors come from firms Systena does business with (Mizuho — a creditor and client for 925 mil.; Nomura — administers the shareholder associations), and the company states it has no independence criteria of its own. The reasonable three-year price target, if the FY3/2029 plan materializes, is ~… (EPV on the planned operating profit of 20,160, adjusted for option dilution) — plus … of annual dividend along the way.