3998 — Bosideng International Holdings Limited

Bosideng International Holdings (3998.HK) — deep-value analysis

September 18, 2026 · reference price HKD 4.10 · market cap HKD 47,821 mil. (RMB 40,801 mil. at 0.8532 CNY/HKD) · 11,663.6 mil. shares · fiscal year ended March 31

Primary sources used: 2025/26 Annual Report (HKEXnews, 2026-07-27, audited by KPMG, statements at p. 126-233), 2023/24 Annual Report (2024-07-26, for FY2023 and FY2024), 2025/26 Interim Report (2025-12-18, H1 at 2025-09-30), data pack data-pack-3998.HK-20260918.md, research brief research-3998-20260918.md, tracker (query_tracker scor/indicatori). All figures in RMB thousand unless stated otherwise.

Methodological note on the data pack: the pack initially generated for the symbol 3998 came out EMPTY (yfinance doesn’t resolve numeric HKEX tickers without a suffix — a known trap). I regenerated it with 3998.HK. Even so, yfinance doesn’t have FY2025/26 in the annual income statement and has nothing quarterly (a semi-annual issuer, per HKEX Main Board rules). All FY2025/26 figures in this report come from the audited statements, not from yfinance.

Executive summary (1 page: thesis, estimated value, verdict)

Bosideng is the undisputed leader in Chinese down jackets: RMB 23,560.1 mil. revenue in the core segment in FY2025/26 (86.1% of the group, … versus the prior year), 3,647 stores, a 69.1% gross margin on the flagship brand, and a segment producing 5,592.6 mil. of operating profit at a 23.7% margin. The rest of the group — OEM (3,093.6 mil., …), ladieswear (558.3 mil., …, a loss of 190.4 mil.), and “diversified” (138.0 mil., …, a loss of 68.1 mil.) — is, in economic terms, a 258.5 mil./year tax paid for failed diversifications. The consolidated group made 27,350.0 mil. revenue (…) and 3,994.4 mil. attributable profit (…), at a 14.6% net margin and a 21.9% ROE.

The thesis isn’t about growth, it’s about the price paid for cash. At HKD 4.10 the company is worth RMB 40,801 mil. Of that, RMB … mil. — 38.2% of the market cap — is net financial assets: 8,251.1 mil. cash at bank (of which 403.4 mil. pledged), 8,708.6 mil. of wealth-management products at Chinese banks, minus 957.0 mil. of bank loans. Enterprise value is left at RMB 25,201 mil., i.e. 4.76× operating profit and 6.6× NOPAT — a 15.1% earnings yield on operating capital. A 50-year-old brand with a dominant market share and a 69% margin is not usually bought at 4.8× EBIT.

The price of this apparent cheapness is a governance question, not an accounting one: does that money reach the minority shareholder? I answered “partly,” with λ = 0.70 (see the valuation chapter), calibrated from three documented things: the 80.2% payout ratio out of FY2026 profit (RMB 3,203.2 mil. of declared dividends), the 75.9% total return on cumulative four-year profit, and the company’s own admission that the drop in the effective tax rate from 31.0% to 28.2% comes “from the decrease in PRC withholding tax on dividends, following the efficient planning of the offshore capital structure” — meaning bringing money up from the Chinese subsidiaries to the Cayman holding COSTS money, and the company says so.

The estimated value — a range, not a point. The five triangulated models give from … (Graham Number, which credits no cash at all) to … (DCF bull), with the median at …. Monte Carlo over 20,000 scenarios, with λ = 0.70 randomized Beta around the median, gives a median intrinsic value of HKD … (MOS …), with P10 at … and P90 at …; the probability the stock is undervalued is …%, and the probability of a margin above 30% is …%. The consensus of 17 analysts is HKD 5.4672 (…). The shape of the distribution is unusual: even the 5th percentile sits at …, meaning almost no reasonable combination of assumptions produces a permanent loss of capital from this price.

Verdict: INTERESTING with a lean toward BUY, moderate position. My recalculated GBL score is …% gross (Graham 8.0 · Buffett 7.0 · Lynch 7.0 out of 60), versus 65.8% in the tracker (evaluated 2026-09-06). Perfect convergence on Graham, divergence upward on Lynch (the segment data supports a market-share gain the mechanical score couldn’t see) and downward on Buffett (margin normalization and the acquisition track record). What caps the position isn’t the valuation, it’s the concentration: a single product category, a single country, a single 74-year-old founder who is simultaneously chairman and CEO, and a single weather variable. The 7.6% dividend pays you while you wait, but absorbs ~90% of normalized owner earnings — there’s no room left for an increase without touching the cash pile.