2555 — Sichuan Baicha Baidao Industrial Co., Ltd.

Deep-value: 2555.HK — Sichuan Baicha Baidao Industrial Co., Ltd. (“Chabaidao” / ChaPanda brand)

Analysis date: 28 August 2026 · Reference price: … (intraday quote, HKEX still open) · Market cap: … mn ≈ RMB 5,561 mn · Reporting currency: RMB · Trading currency: HKD · FX used: 0.8568 CNY per …

Identity note (mandatory): the bare ticker “2555” does NOT resolve correctly. On the Tokyo Stock Exchange, 2555 is a Simplex ETF on Japanese REITs; the company in Radu’s tracker (column 553 of Analiza, Exchange = HKEX) is Sichuan Baicha Baidao Industrial Co., Ltd., listed in Hong Kong under 2555.HK. It is not SEC-registered (no CIK, no companyfacts) — EDGAR was intentionally skipped. Today’s first generated data pack came out completely empty for this reason; I regenerated it after adding the alias "2555": ("2555.HK", ...) to _BVB_ALIASES in deep_data_pack.py.

Primary sources used (all downloaded directly from hkexnews.hk, not from aggregators):

Corrections to the research brief (treated as testimony, not fact — and three of its figures are wrong):

1. The brief says the H1 2026 results are “NOT YET PUBLISHED… the company is due to report on 31 August 2026”. False: they were published today, 28 August 2026, at 12:01 Hong Kong time, three hours before the session closed. The entire analysis below is built on them. This is the report’s single most important piece of information, and the brief lacked it.
2. The brief reports a “current price … (TipRanks)” and flags an unresolved discrepancy against … 6.33 is the 31.12.2025 close — an eight-month-old figure served by the aggregator as “current”. The real price today is …, with an intraday low of 4.370 that is also the new 52-week low. Verified series: 4.645 (24.08) → 4.545 → 4.545 → 4.445 → … (28.08), today’s volume 3.16 mn shares, ~1.7x average.
3. The brief cites “Chabaidao revenue of RMB 4.918 billion in the first half of the year”, taken from Futunn. Wrong as to the period: RMB 4,918.0 mn is the revenue for the full year 2024, not a half-year. The third-party source confused a year for a half-year, overstating the company’s scale by ~100%.
4. The brief says “final dividend RMB 0.18/share (versus RMB 0.2 the prior year)” — correct, confirmed in note 7 of the FY2025 results.
5. The analysts’ average target of …, which the brief flags as “unreliable, don’t use”, checks out: yfinance gives a targetMeanPrice of 8.057 on 3 analysts, range 6.342-9.896 — practically identical to the brief’s …-9.90 range. The figure the brief correctly flags as unusable is “… from 3 Wall Street analysts”; I found no source for it and ignore it.
6. The GBL tracker (row 4, indicator date 23.08.2026) writes “Mixue ~20-23x”. Wrong: Mixue (2097.HK) trades today at a TTM P/E of 11.56x. Chabaidao’s discount to Mixue is real, but it’s 6.8x vs 11.6x, not 6.8x vs 22x.

Executive summary

Chabaidao is China’s third-largest fresh-tea chain by store count (8,863 at 30 June 2026), an almost pure franchise model: 8,778 of the stores are franchised, only 13 are self-owned, and the company makes money by selling franchisees raw materials, packaging and equipment, plus a royalty on their revenue. The company listed in Hong Kong on 23 April 2024 at …/share. Today the stock costs … — down 74.8% from the IPO price over 28 months — and just hit a new 52-week low in the very session it published its half-year results.

Statistically, the paper looks very cheap. TTM P/E 6.82x, P/B 1.43x, a 9.2% dividend yield on the last 12 months’ distributions, and net cash of RMB … mn represents 52.5% of market cap — … of the … price is money in the bank, with no bank debt, no pledges, no covenants. Implied enterprise value is RMB 2,634 mn, i.e. 4.8x the 2025 shareholder cash flow and 2.4x TTM adjusted EBITDA. The FCFE yield on market cap is 9.8%.

My thesis is that this cheapness is real, but far less free than it looks, and today’s published results show exactly why. H1 2026’s … revenue growth is entirely, and disproportionately, coffee-equipment sales to the company’s own franchisees, on 18-month credit: equipment revenue jumped from RMB 48.5 mn to RMB 266.8 mn (…), while merchandise revenue — the core business, 85% of the total — fell 2.3%, and royalty-and-franchise-fee revenue fell 19.5%. Excluding equipment, company revenue was … versus last year, on a store network 4.7% larger. Trade receivables rose from RMB 17.9 mn (30.06.2025) to RMB 194.1 mn (30.06.2026), of which RMB … mn is classified as non-current assets because maturity exceeds 12 months. Gross margin fell 2.6 points, to 30.0%, and the company acknowledges the reason in text: the coffee campaign “returned profits to franchisees”.

In parallel, two indicators no one tracks point the same direction: royalty per store, a direct proxy for store-level sales, fell 9.1%, and the initial franchise fee collected per new opening fell 52.6% (RMB … thousand → RMB 47.1 thousand), on 43.6% more openings. And advances received from franchisees (contract liabilities), the cleanest forward-looking indicator of their demand, collapsed from RMB 324.2 mn (Dec. 2023) to RMB 96.0 mn (June 2026) — down 70% in two and a half years.

Capital allocation is the second problem, and it’s worse than operations. In 2024, the IPO year, management spent RMB 1,053.6 mn of shareholders’ cash buying listed shares of other companies (FY2025 cash flow statement, the “Purchase of equity instruments at FVTOCI” line). The cumulative loss recognized in equity on this portfolio is, at 30 June 2026, RMB 494.1 mn, i.e. ~47% of the amount invested, with a remaining carrying value of RMB 174.6 mn. The loss flows through OCI, so it never touches the income statement. In the same period the company has bought back no shares at all — explicitly confirmed in both filings (“neither the Company nor any of its subsidiaries purchased, sold or redeemed any securities”) — and today proposed an incentive plan that could issue up to 10% of capital as new shares.

Estimated value. Five triangulated models give a range from … (bear: coffee fails, royalty keeps falling, cash with a 30% access-risk haircut) to … (bull: reverts to the 2022-2025 average FCFE). The five-model median is …, and the DCF base case (OE = RMB 546 mn, g1 = 2%, r = 11%, gt = 1%, net cash −…) gives … i.e. …. A Monte Carlo simulation over 20,000 scenarios gives a median of … and a …% probability of undervaluation — but that probability is an artifact of the simulation varying only four operational parameters and treating net cash as certain, and net cash is half of the price. Don’t believe the …%.

Verdict: SMALL BUY / WATCH. This is a cheap company with a fortress balance sheet and a covered dividend, run by a founder who owns 82.7% and has demonstrated he’d rather bet cash on foreign shares than buy back his own at 1.4x book value. The position is justified by the ~8% dividend yield plus re-rating optionality, not growth — because there is no growth. It isn’t a conviction position. The trigger that would turn it into one: a real buyback, or two half-years in which equipment receivables collect on schedule with no provision. The trigger that closes the thesis: the first provision on the 18-month receivables portfolio.