XJ International Holdings Co., Ltd.

The operating recovery is plausible. The investable recovery still depends on cash left after creditors, required school spending and minority interests.

Students pay for qualifications and teaching. The listed share receives the residual only after school obligations, contractual-control routes and financing claims.

XJ International, formerly Hope Education, is a Cayman-incorporated issuer with HKD ordinary shares on HKEX. Its financial statements use RMB and a 31 August year-end. The economic business is a campus operator: higher education, junior colleges and technician education, with overseas institutions in Malaysia, Thailand and Hungary. It is not primarily an online training platform. ¹

The operating equation is fee-paying student exposure x recognized tuition yield + boarding and ancillary revenue, adjusted for acquisitions and disposals. Teaching payroll, premises and depreciation determine gross profit; administration and recruitment determine operating profit. License, quota, teacher supply and usable accommodation constrain growth before theoretical demand can become revenue.

The principal PRC claim runs through offshore holding companies, a wholly foreign-owned enterprise and structured contracts with Sichuan Hope Education and affiliated schools. Malaysia and Thailand use separate subsidiary and trust arrangements. These routes support accounting control, but each has its own distribution, tax and creditor constraints. The immediate bottleneck in this investment case is funded, accessible cash rather than an assumed shortage of constructed places. ¹

The recent census is broadly flat before disposals, with fewer junior-college and technician students. Undergraduate growth alone does not establish group demand growth.

A census is a stock. Intake, graduates, transfers and withdrawals are flows, so the unexplained difference between two censuses is not a dropout rate. Earlier new-entrant observations of 48,789 and 72,830 belong to different acquisition-era source periods; they cannot establish a current recruitment trend. The next useful disclosure is registrations and fee-paying exposure within the retained portfolio, with disposals dated explicitly. ^6,8

Designed capacity is also different from permitted admissions and staffed capacity. The 2023 annual source reports 305,640 places and 290,927 students at 31 October 2023, implying 95.2% utilization within that defined scope. Subsequent disposals prevent using that old ratio as current headroom. The new October 2026 land purchase is staff accommodation and adds no announced student seats. ^4,11

The comparable FY2023-25 calculation shows an improving recognized yield. It does not isolate tariff increases from program, geography, delivery timing or student mix.

Tuition revenue increased 6.4% in FY2025 while the matched October census rose only 0.1%. The arithmetic therefore attributes about 6.2% to recognized yield. This is consistent with mix improvement, but it is insufficient to certify pricing power without retained-cohort fees, discounts, refunds and competitor tariffs. ^1,3

H1 FY2026 tuition was RMB1,853.3m, versus RMB1,889.9m a year earlier. Disposals during the six months mean a pre-disposal census cannot support a clean current fee calculation. We retain the reported tuition flow but leave that yield branch uncalculated. Historical calendar years, the eight-month transition and August fiscal years are separated in the appendix. ²

The audited PRC group disclosure describes legacy private non-enterprise units or nonprofit schools. A school upgrade or independent-college conversion is not completed for-profit registration.

At approval of the FY2025 statements, three PRC schools were incorporated as nonprofit and the remaining PRC schools were still undergoing classification registration as private non-enterprise units. The disclosure does not allocate the three nonprofit identities across the roster. The matrix therefore distinguishes a confirmed group boundary from individual legal subtype; it does not fabricate school-specific tax or dividend permissions. ¹

Classification key: ✗ L/N means the FY2025 group disclosure supports legacy-unconverted or nonprofit status, without assigning the individual subtype; ✗ NP identifies Jinzhong’s explicit nonprofit permit from February 2025. Neither symbol alone proves a cash-distribution route. Scope key: the table is the FY2025 principal named legal-entity roster, with later transaction flags; it is not an additive current campus count. The September circular reports 20 operating schools, 16 in the PRC, including the disposal target. Different dates, principal-entity coverage and brand/campus counts explain why those records should not be summed. The latest trading quote does not update school legal status. ¹²

The foreign perimeter has separate rules: INTI International University and the Subang and Penang college entities operate in Malaysia, Shinawatra University in Thailand, and Wekerle International University in Hungary. PRC classification labels do not apply to them. Historical principal structures include trust-held interests in the Malaysian and Thai holding companies. Wekerle’s regulator register confirms accreditation through 31 March 2029, with an action plan due 30 November 2026 and progress report due 30 November 2027. This is an operating permission, not proof of demand or remittance. ^1,29

The Gao County project shows the economic mechanism in practice. Structures completed in 2022 could not serve the intended private higher-education use after the project was excluded from Sichuan’s 14th Five-Year establishment plan. The September supplement describes a narrow eligible buyer pool and a long negotiation. This is a specific permission-to-cash risk, with stronger evidence than a generic sector-policy discount. ¹⁵

We use four degree-education peers to benchmark operations. Their legal rights, campus maturity and capital structures are too different to justify an unadjusted average P/E.

China New Higher Education (02001), China Education Group (00839), YuHua (06169) and Shanghai Gench (01525) share degree-granting campus economics. They are operating comparators rather than interchangeable securities. XJ’s cross-border portfolio, restructuring and legacy claims make headline earnings and distributions less comparable. Gench’s December year-end is explicitly different; no time alignment is implied. ^24,25,26,27

We reject Virscend as a direct multiple anchor because K-12 demand, policy and pupil economics differ. China East Education is useful vocational-training context but has a different credential and course-duration model. The best alternatives for a learner are local public and private programs with similar courses, fees and qualifications; stock-exchange peers are not necessarily direct recruitment competitors.

FY2021-25 scale growth did not preserve the original gross margin. The latest half-year recovery is encouraging, but retained-school costs must confirm it.

FY2025 selling and administration together consumed 23.8% of revenue, compared with 22.6% a year earlier. Selling expense rose faster than revenue, so a recovery driven solely by higher student yield would overstate operating leverage. The base model assumes 22.5% for these costs; reducing it is a forecast requirement, not a reported achievement. ¹

The campus maturity split is not sufficiently stable after exits to attribute the entire margin movement to ramping schools. We do not hold teacher or maintenance costs flat while assuming higher enrollment. The model retains growing maintenance spending, a continuing capital program and cash allowances for working capital and legacy creditors. Rankings and internship opportunities are not monetized separately.

The annual accounts contain large gains and losses in both directions. Adding back only the unfavorable items would produce a misleading recurring-profit estimate.

The subtotal of negative RMB54.2m is a limited pre-tax diagnostic, not recurring net profit. It still contains goodwill impairment, other impairments, investment fair-value losses and other operating items. Full normalization needs tax and ownership attribution for each item. We instead forecast recurring school operations and funded investment explicitly, while keeping the issuer’s adjusted group profit as a separate comparison. ¹

The latest H1 operating cash covered only 39.2% of owners profit. After PPE, intangible and development-land spending, cash interest and lease principal, the half-year residual was negative RMB145.3m. Autumn tuition collections create seasonality, so neither that deficit nor a mechanical doubling of H1 proves the full-year outcome. Contract liabilities fell from RMB2,105.6m in August to RMB1,399.9m in February; recognition timing and school exits must be separated from demand. ²

The convertible bond is gone, but the newer June statement shows RMB3,130m borrowings, separate leases and a retained disposal guarantee. February remains the latest full balance sheet, not the latest debt evidence.

The original zero-coupon US$350m bond combined a lending claim and a conversion option. Zero coupon did not mean zero funding risk: the 2024 put dispute and winding-up litigation exposed offshore liquidity and remittance dependence. The appeal was dismissed in June 2025; restructuring became effective in September. H1 FY2026 records RMB1,165.1m redemption cash and RMB132.5m restructuring cash. ^1,2

Other payables include RMB369.4m of PPE creditors, RMB458.1m of independent-college conversion payables, RMB260m of acquisition payables and RMB122.1m of third-party loans. They are not all bank debt, and not all are immediately due; they cannot be ignored simply because the bond balance is zero. At the same time, it would be wrong to deduct all operating payables again after funding the same cost in projected cash. ²

The RMB766.1m difference between the two borrowing totals is not established as new borrowing. February also reports RMB122.1m third-party loans and RMB75.8m of construction loans within other payables; these are outside its borrowing-note total. June’s unsecured other-borrowing category has no directly corresponding line there, and the medium-term-note security labels also differ. Instrument-level classification, additions, repayments and scope changes must be reconciled before attributing the difference to financing flows. ^2,12

The June circular is a newer indebtedness statement, not a full replacement balance sheet. We do not subtract February cash from June debt to claim a current net-debt figure. Its RMB76m leases are separate from RMB3,130m borrowings; neither total includes the following deconsolidated-business guarantee. ¹²

At 30 June 2026, XJ still guaranteed RMB179.9m bank borrowings of the subsidiary sold on 29 January. The interim identifies that subsidiary as Xingtai Group and reports RMB210m guaranteed at February. The newer exposure is 19.9% of the report’s RMB906.1m basic market value. Selling the subsidiary removed its borrowing from consolidation but did not extinguish XJ’s guarantee. Release or repayment remains a funding checkpoint. ^2,12

This guarantee is a contingent exposure, not a confirmed loss. The base assumes no call; an immediate full-call stress charges RMB179.9m once, with no tax benefit or recovery credit. It is additional to the forecast on-balance-sheet legacy settlements. We do not add it to consolidated borrowing and deduct it again. The valuation chapter quantifies the stress, and monitoring requires evidence of release, repayment or a funded settlement. ¹²

The observable allocation is defensive debt reduction combined with further property and connected investment. Ordinary-share distributions are not yet the priority.

The target is Sichuan Yonghe and Sichuan Vocational College of Culture and Communication. Its FY2025 revenue was RMB213.1m and after-tax profit RMB43.2m. Selling the equity at RMB377.2m equates to 8.7x that unaudited profit, before adjustments; total consideration adds RMB18.52m of shareholder debt. The transaction can release scarce funding but also removes earnings. Shareholder approval on 18 September is a completed stage; cash collection and control transfer are different stages. ^12,13

The RMB100m borrowing originally matured on 9 October 2026; RMB240m matures on 22 December. The first date has passed, but the circular alone certifies neither repayment nor default. Settlement or refinancing evidence is required before releasing the funding reserve. ¹²

The same circular budgets about RMB800m through February 2028 for compliance and development: construction/repair, equipment and settlement of specified payables. Its proposed Jintang student-accommodation land, about RMB37.2m before construction, is distinct from the October Nanchong staff-housing land. The forecast capital and legacy allowances cover these uses by category, rather than adding every disclosed budget on top of the same expenditure. ¹²

The new staff-housing land costs RMB115.4m, around 12.7% of the current basic market value in RMB. A RMB23.1m deposit is within that price. Contract instalments are RMB57.7m by 14 November 2026 and RMB57.7m by 14 October 2027; the settlement record must show deposit credit. For valuation we reserve the RMB92.3m remaining total early in FY2027, a conservative timing assumption, and add no new tuition seats. ¹¹

The INTJ subscription is especially relevant to allocation discipline. XJ subscribed US$7.5m for 600,000 Class B shares in a controller-linked financial public-relations company. The SEC ownership amendment says the placement was consummated, and INTJ’s October results say the cash arrived after its May balance date. Thus the transaction is beyond a proposal even though the earlier HKEX wording was conditional. ^16,22,23

At the reported transaction denominator, the new stake represented 28.1% of INTJ’s economic shares, despite the announced 85.2% voting control. INTJ’s H1 revenue and net profit were HKD15.5m and HKD2.8m, respectively; a single improved half-year does not substantiate the proposed education synergies. We assign no standalone platform premium or full-group profit to XJ shareholders. The subscription is sunk cash before the model cutoff, so it is not deducted again as a future cash call. ^16,17,22,23

Connected ancillary services require a different test from the INTJ investment. Under the Tequ agreement, the related operator pays Hope Education for premises and resources used for canteens, driving school and vocational training. Historical payments fell to RMB4.39m in FY2025; the FY2026 cap is RMB10m, not earned revenue. Monthly settlement and comparable-rent controls reduce collection/pricing risk, while cash receipts and actual resource use remain the relevant evidence. ^18,19

At August 2025 the parent itself had RMB121.5m cash, against RMB2,319.6m current liabilities, largely intra-group balances. Consolidated school cash cannot simply replace that parent balance. FY2025 and H1 FY2026 declared no dividend. Debt reduction can protect minorities, but until cash upstreaming produces distributions it supports survival rather than a cash-return thesis. ^1,2

Dominant pattern: a potential operating recovery constrained by legacy funding and cash access. The balance sheet has improved enough to study, but not enough to presume a durable minority return.

The price appears consistent with three fears: reported profit overstates recurring school earnings; remaining capital and legacy uses absorb operating cash; and any residual stays within the group rather than reaching ordinary holders. These are analytical interpretations, not claims to know investor motives. The strongest counter-evidence is the latest gross-margin recovery and completed bond redemption. The strongest confirmation is the disposal circular’s explicit cash priorities. ^1,2,12

The deserved discount is not one blanket percentage. It arises from removing nonrecurring earnings, excluding unverified excess cash, retaining capital and legacy funding, and reserving dilution. Potentially excessive discount exists only if the retained schools sustain margins while these cash demands taper. Our base case illustrates that possibility; the bear case shows why it is not established value protection.

This is not a pure earnings-quality failure: annual OCF substantially exceeds annual reported profit. It is also not a proven structural value trap, because senior bond risk has been removed and operating margins improved. Watchlist, Low confidence best captures the unresolved balance. The controlling variables are retained-school cash after reinvestment and actual parent access, with settlement timing connecting them.

A funded owner-cash model gives HKD0.000 / 0.275 / 0.643 per reserved share. The base is an operating state, not expected value or a 12-month target.

Primary method: three explicit years of post-interest, post-reinvestment cash attributable to the ordinary-share claim, with a finite terminal cash multiple. Independent checks are the actual FY2025 cash residual, the parent cash position, disposal proceeds’ uses and the latest creditor balances. We reject a reported-profit P/E target and campus book value as a liquidation floor. We do not mechanically average these methods.

Valuation control: HKEX close HKD0.122 on 27 October; 8,673.936m issued shares excluding treasury at 30 September; CFETS fixing RMB0.85621 per HKD on 28 October. Basic market value is HKD1,058.2m, or RMB906.1m. Reserving all 702.792m existing option shares produces a conservative ceiling of 9,376.729m shares. Many options are out of the money; no exercise proceeds are credited. Extinguished bonds add no further shares. ^10,20,21

The revenue anchor is FY2025 group revenue less RMB213.1m from the approved Culture and Communication disposal: RMB3,747.2m. This is an illustrative retained-revenue anchor, not a fully reconstructed pro-forma retained-school statement; other in-year exits and INTJ accounting still require reconciliation. We express volume as an activity index, rather than inventing a post-disposal student count. No model output is presented as verified group guidance. ^1,12

Cash tax is 20% of positive EBIT less cash interest, an analyst blended allowance rather than a legal rate for every school. D&A is 17% of revenue; maintenance starts at RMB500m and rises 2% yearly. Growth spending is RMB200m / 175m / 150m, legacy funding RMB250m / 200m / 150m. These are expenditure categories, with disclosed compliance spending allocated once. The remaining land consideration is reserved early; any larger construction plan requires a new funding deduction.

Opening group cash is treated as necessary operating/financing liquidity, so the model adds no excess cash or sale receivable at par. School-disposal proceeds pay named debt and proposed land; its revenue is removed. Annual cash interest remains an explicitly conditional RMB180m assumption, not a demonstrated conservative allowance. The going-concern cases require the bank, note and other borrowing obligations to remain serviced or refinanced, including any June obligations previously classified in other payables. The same obligation must not also be charged as a new legacy claim. Principal funding and the June-to-February scope bridge need confirmation; the terminal is not debt-free.

Year-end cash is discounted from 28 October 2026 to the August FY2027-29 dates. The base terminal is 6x FY2029 owner cash, producing 74.6% of present equity value; the bull terminal share is 77.4%. That dependence warrants low confidence and operating sensitivities. No statistical scenario probabilities are defensible from the present evidence, so we publish no expected value or probability-weighted annual return.

Interest check: RMB180m including lease interest implies a 5.49% non-lease cash yield on June’s RMB3,130m balance, after using FY2025 lease interest of RMB8.1m as a proxy. This is a plausibility test, not a measured June weighted rate or annual average debt balance. February loan rates span 2.50%-9.14%; the June unsecured category’s rate and instrument overlap need confirmation. ^1,2,12

An illustrative mix of 4.25% on bank loans, 5.30% on notes, 8.00% on secured other borrowing and 7.50% on unsecured other borrowing, plus the lease proxy, gives RMB183.9m annual cash interest. These are analyst rate assumptions informed by earlier disclosures, not reported June rates. They do not establish that RMB180m is too low; the higher-interest cases below show the valuation consequence if it is.

The immediate full-call guarantee stress reduces value by HKD0.0224 per reserved share, to HKD0.253 at RMB180m interest. Keeping base value at HKD0.275 therefore requires both the interest allowance and the no-call assumption to hold. No probability is assigned to a call, and no subsequent release is assumed. A call requiring additional refinancing could have a larger effect than this funded-payment sensitivity.

Reverse valuation: keeping every base assumption fixed, the quote requires activity at 86.4% of the base path. This is an index result, not a forecast of actual headcount. The observed discount could alternatively reflect weaker margins, slower legacy settlement or poorer cash access; price alone cannot identify which fear is correct.

Two horizons: intrinsic values above discount FY2027-29 cash. For a separate 12-month no-convergence illustration, retain today’s price/base-value ratio of 44.3% and roll each intrinsic state forward at its own equity hurdle, assuming no distribution. The resulting marks are HKD0.000 / 0.140 / 0.322. These are conditional marks under a persistent discount, not market forecasts. A tighter actionable range requires settlement, retained-school and parent-cash evidence by 28 October 2027.

Entry arithmetic alone would require about HKD0.069 for 3x base-upside/bear-downside, using the same fragile model. This is a hurdle illustration, not an entry recommendation. Survival, liquidity, legal access and a credible expected-return case must pass separately; a lower quote does not cure an unverified cash claim.

The earliest useful signals are settlement and retained-school collections. A profit beat without these would not resolve the investment case.

The near-term upside mechanism is not simply “the market notices low P/E.” It is a sequence: sale collections settle named debt; retained-school collections cover compliance and legacy cash; parent receipts become observable; then a funded distribution or executed buyback validates the minority route. Until then, the operating and creditor cases can improve without a proportional equity re-rating.

Execution is a separate constraint. On 27 October HKEX recorded 498,000 shares and HKD60,252 turnover, with HKD0.122 bid and HKD0.124 ask. The quoted spread is about 1.6% of midpoint. A single day is not average daily volume or a defensible exit-capacity estimate. We make no institutional-position or short-squeeze claim from this observation. ²⁰

There is evidence of recovery, but the current information does not establish a protected downside or a repeatable cash-return route.

Upgrade: documented repayment/refinancing of named maturities, release or fully funded resolution of the Xingtai guarantee, a matched retained-school operating baseline and positive annual cash after required reinvestment and legacy settlements. A funded parent distribution must validate the legal cash path. Recalculate the model after those events; do not preserve the old target mechanically.

Downgrade: retained-school collections weaken while land, compliance or legacy uses consume more cash than modeled. A 3-point margin shortfall or another RMB100m annual cash call each removes roughly HKD0.075 per reserved share in the base calculation. Those are material changes relative to the quote.

Kill / rebuild: loss of control or license, an adverse senior-claim event, or recapitalization that destroys the old-share claim. No accounting book-value floor or historic share price overrides that event. The zero bear already recognizes discontinuous loss; it is not a prediction of insolvency.

Position discipline: this report establishes no new-position case. The 1.26x base reward/risk is below the price hurdle, probability-weighted return is unestablished and execution capacity needs a multi-day liquidity study. The classification is a research judgment, not a personalized trade instruction.

The conditional base is lower than our prior model despite a lower quote. We remove the approved disposal target’s revenue, reframe volume as a retained-activity index, fund the remaining staff-housing price without double counting its deposit and use actual August discount dates. The conclusion remains Watchlist; these changes narrow what the apparent upside actually proves.