92730 — NeoPharm Co., Ltd.

NeoPharm Co., Ltd. (네오팜) — KOSDAQ 092730 — deep-value analysis, REFRESH

September 16, 2026 · reference price … · 15,641,256 shares outstanding (16,027,989 issued − 386,733 treasury) · market cap 340,197 mil KRW

Regime: REFRESH. The reference analysis is from 08.23.2026 (verdict: BUY, small position). The mechanical deep_delta.py triage of 09.15.2026 requested a refresh for 3 unverifiable falsifiers — not a broken falsifier. Delta pack: rapoarte/deep/delta-092730-20260915.md.

What I inherit, explicitly. Between 08.23.2026 and today there is no new filing. The DART query (python dart.py 092730 --ani 2) returns as the most recent periodic filing the 반기보고서 (2026.06), received 08.14.2026 — the exact same one the reference analysis relies on. The next primary document is the Q3 2026 quarterly report, expected around 11.14.2026. Consequently I inherit untouched: the Business and moat, Management and capital allocation, Balance-sheet analysis — Quality of Earnings, CEO profile chapters, and the factual structure of the Red flags (report of 08.23.2026, lines 35–87, 161–246, 247–270, 271–290). I summarize them here with their figures, but don’t re-derive them: nothing has been published that would move them.

What I re-derive, and why. Four things:

1. The falsifier set — the real cause of the refresh. Three of the six falsifiers in the old thesis are structurally unverifiable for this issuer, not temporarily unavailable. I replace them with observables that actually populate. Detailed in Verdict versus the tracker’s GBL score.
2. Access to net cash (λ) — step 5b of the method, which the August run only handled in prose and did not apply in the simulation (acces=1 in both the 08.23 and 09.15 JSON). At a ticker with 42.6% of price in net cash, that’s not a detail: it’s nearly half the intrinsic value awarded with zero variance, across all 20,000 scenarios. I now calibrate it from filings and apply it.
3. The entire valuation, at today’s price and with λ applied — the five triangulated models and Monte Carlo.
4. The FCF bridge, cross-checked: yfinance has since delivered the 2026-06 quarter, which it didn’t have in August. The manual reconstruction from back then is confirmed to the million (see below) — a validation, not a correction.

The price move isn’t the thesis. The delta pack records … from … to …. The … figure is the price recorded in the tracker when the thesis was written, not the report’s reference price (…, close of 08.21.2026). Versus the reference report, the stock fell 3.5% in 26 days, on zero new information. That’s context, not a reason to re-do the thesis, and I build nothing on it.

Executive summary

NeoPharm makes dermocosmetics for sensitive and atopic skin, under five brands — Atopalm, Real Barrier, Zeroid (the medical line, through hospitals and clinics), Derma-B and Ts — on two proprietary patented technologies: synthetic ceramides Ceramide-9S and the multi-lamellar emulsion MLE®. 196 employees, headquarters and plant in Daejeon, KOSDAQ-listed since 2007, 42.34%-controlled by the It’s Hanbul group.

Operating figures, unchanged from August, since nothing new has been published. Q2 2026 remains the best quarter in the company’s history: revenue 43,572 mil (… YoY), operating profit 9,789 mil (…), net profit 8,405 mil (…). For the half-year: 79,796 / 18,493 / 17,029 mil (… / … / …). TTM (Q3 2025 – Q2 2026): revenue 148,389 mil, operating profit 32,476 mil (margin 21.9%), net profit 30,648 mil, EPS ….

Today’s valuation, at …: P/E 11.1x, P/B 1.69x, EV/EBIT 6.01x after deducting net cash of … mil, FCFE yield 7.88%, simple FCF yield 8.16%. Net cash is … per share, i.e. 42.6% of price and 72% of equity.

What changed in the thesis, in one sentence: the business is the same, the price is 3.5% lower, but the valuation collapses when net cash is treated for what it is — a claim with uncertain access on a family that has never distributed it — instead of cash at face value. The August run had correctly written the problem in prose (“if that money never reaches the shareholder, its present value isn’t … mil, it’s 29,640”) and then reported the simulation marker that gave it full face value anyway. This refresh closes the inconsistency.

Calibrating access (λ = 0.35), from the filing, not from impression. The FY2025 annual report (chapter VI, dividend table) gives three years of total shareholder return: dividends 5,865 mil (FY2023), 9,384 (FY2024), 11,730 (FY2025) = 26,979 mil, buybacks zero, on cumulative consolidated net profit of 74,101 mil → 36.4% of cumulative profit reached the shareholder. The independent floor: the pile yields 2.25% net, and capitalized at …% is worth 31,051 out of …, i.e. an implied λ of 0.214. I use λ = 0.35, between the two, at the Samsung benchmark level (0.35), above Sunny Optical (0.22) and Geely (0.20) — because NeoPharm has a distribution policy that’s written, quantified (40… of standalone profit), honored and increased by 25%, “high-dividend enterprise” tax status, and a parent company structurally dependent on this dividend; and below i-Scream Media (0.40), because it has bought back not a single share in five years and hasn’t canceled the 386,733 treasury shares.

Estimated value. Five models, all with λ stated: DCF bear … (…), 5-year historical multiples … (…), EPV Greenwald … (…), base conservative DCF … (…), DCF bull … (…). Median of the five: …. Monte Carlo across 20,000 scenarios with λ = 0.35: median intrinsic value …, median MOS ** …**, but with P10 at … and undervaluation probability …% (versus 98.3% without λ).

Verdict: HOLD / accumulate only below …. I’m downgrading from BUY (08.23.2026) — not because the business deteriorated (nothing published shows that), but because the valuation model back then contained a methodological error I identified and didn’t correct: 42.6% of price was treated as cash at face value, across all scenarios. Corrected, the median of the five models moves from … to …, and the pessimistic end of the Monte Carlo distribution moves from … to …. The business remains excellent: 51… return on operating capital employed, a twenty-year niche-leading brand, zero financial debt for five years, uninterrupted dividend since 2007. The price, however, no longer offers a margin of safety once the unproven hypothesis about the treasury is removed from the equation.