1318 — Mao Geping Cosmetics Co., Ltd. Class H

Deep-value: 1318.HK — Mao Geping Cosmetics Co., Ltd. (毛戈平化妆品股份有限公司)

Analysis date: August 24, 2026 · Reference price: … (close 08.24.2026, Yahoo Finance) · Market cap: … ≈ 21.65 bn CNY · Shares: 490,186,900 · Exchange: HKEX · ISIN: CNE100006PM5 · Reporting currency: CNY · Trading currency: HKD (rate used: … = 0.8574 CNY, i.e. 1 CNY = …, Yahoo CNYHKD=X at 08.24.2026)

Identification note — a correction to today’s earlier run. The morning-generated data pack (data-pack-1318-20260824.md) resolved the bare symbol “1318” as Listed Index Fund TOPIX Small (1318.T, a Japanese ETF) and came back completely empty. The “Exchange” column in the tracker (row 260 in Sumar) says HKEX, and the tracker’s name is “Mao Geping Cosmetics Co., Ltd. Class H.” I regenerated the data pack on the correct symbol — data-pack-1318.HK-20260824.md — and every figure in this report comes from there, from yfinance on 1318.HK and from cited public sources. This is exactly the symbol-collision pattern described in the instructions (SNN/EL/M), just on an Asian numeric code: bare “1318” resolves by default to Tokyo. Infrastructure recommendation: adding it to _BVB_ALIASES in deep_data_pack.py would prevent a repeat.

A second correction, also from verification: the research brief states the price and market cap are “contradictory, unreliable data” and lists 86.50 / … and market caps of 35.32 / …. The correct figure as of this analysis is … / …; the values 86.50 and 102.14 are old quotes (January and September 2025 respectively), and … corresponds to the ~… price from March 2026. The brief thus missed the file’s central fact: the stock has lost 60.6% from its all-time high and …% in 2026 alone.

Executive summary

The thesis, briefly. Mao Geping is the only 100%-Chinese luxury cosmetics brand accepted at premium-mall beauty counters (SKP Beijing/Wuhan/Chengdu, Hangzhou Tower), built on the founder’s personal IP — a celebrity makeup artist with 25 years of fame in China. The business’s economics are remarkable: 84.2% gross margin, 23.8% net margin, 29.5% ROE on average equity, zero bank debt, and 3.43 bn CNY of net cash, with revenue growing from 1.83 bn CNY (2022) to 5.05 bn CNY (2025), a 40.3% three-year CAGR. Since listing at … in December 2024, the stock rose to … (June 2025) and fell to … — a complete euphoria-and-disappointment cycle in 20 months, without reported profits ever missing.

What actually broke. Not the results. FY2025 delivered revenue of CNY 5,050.5 mil. (…) and net profit of CNY 1,204.1 mil. (…), ahead of the market’s pace. What broke is confidence in owner alignment. On January 6, 2026, exactly 12 months after listing, when the lock-up expired, six people — founder Mao Geping, his wife Wang Liqun, sisters Mao Niping and Mao Huiping, brother-in-law Wang Lihua, and executive chairman Song Hongquan — announced their intent to sell up to 17.2 mil. shares (3.51% of capital, ~… at the then-price of …), the stated reason being “supply-chain investment” and “improving personal life” (Tencent News, Guancha). On May 15, 2026, shareholders Ding Tao and Xu Kejun — the founder’s nephews — placed 9.873 mil. shares through JPMorgan at …, a 5% discount, netting ~…; the stock fell 16.88% intraday and closed … (Eastmoney, Sina Finance). From May until today, the quote has lost another 16%.

Estimated value. Real free cash flow to shareholders in FY2025 is CNY 969 mil. — CFO 1,199.4 mil. minus capex 187.0 mil., minus interest paid 2.3 mil., minus lease principal 41.3 mil. (the unallocated residual from financing). This is 4.3% below “simple FCF” of 1,012.4 mil. — a small divergence, so the choice of definition doesn’t flip the verdict, but 2025’s capex (187.0 mil.) is abnormally low versus 2024 (615.9 mil.), and CNY … mil. sits in construction in progress, so the normalized flow is lower than reported. The five triangulated models give a range from … (DCF bear with normalized capex of 300 mil. and r=12%) to … (DCF bull at 18% growth, close to consensus), with the median at …. Monte Carlo (20,000 scenarios, same central assumptions) confirms it: median …, P(undervalued) …%, P(MOS> …) only …%. Median intrinsic value: … versus … today.

Verdict. WATCH — don’t buy at …. Not because the business is bad: it’s one of the most profitable consumer companies in China, with an impeccable balance sheet. But because, after a 60% drop, the price has just barely reached value, not below it — a 4% margin of safety isn’t a margin of safety, it’s a rounding error. Three things need to happen to change the verdict: (1) the H1 2026 results on August 27, 2026 (three days away) need to show that the finished-goods inventory, which grew 73.6% against COGS growing 31.2%, was sales preparation, not sales that didn’t happen; (2) the family needs to stop selling — or, better, the company needs to start buying back, which it has never done; (3) the price needs to fall below ~…, where the base DCF offers a 30% margin of safety (the simulation’s P25 percentile). Below … the thesis becomes a buy; between 41 and 55 it’s a watch position; above … the model says you’re paying for the 18%/year growth only consensus sees.

The biggest thesis risk is neither accounting nor valuation: it’s that the brand carries the name and face of a 62-year-old man, and that man is selling shares to “improve his personal life.” No 84% gross margin compensates for that if the person disappears or gets compromised.