SRB — Serabi Gold plc

Deep-value report: SRB — Serabi Gold plc

Ticker: SRB (LSE AIM) / SBI (TSX-V) / SRBIF (OTC) · Yahoo symbol: SRB.L · Sector: Basic Materials / Gold Price as of analysis date: … (close 03.09.2026, yfinance SRB.L; the same-day tracker snapshot showed … — I use the HIGHER figure, i.e. more conservative for the margin of safety) Shares: 75,731,000 (31.03.2026 balance sheet) / 76,307,760 total voting rights (01.06.2026) Market cap: …M / …M (GBP/USD 1.3524) · Cash 30.06.2026: …M · Debt: ~…M (leasing only) → net cash ~…M (20.2% of market cap) Reporting currency: USD · Trading currency: GBp (pence)

⚠️ Identification and source limits. SRB is not an SEC filer — edgar_10k_downloader.py SRB again returned “CIK not found” (run 04.09.2026). No 10-K/20-F/10-Q, and no SEC-Filings\SRB\. Primary sources are the issuer’s AIM/GlobeNewswire releases (audited FY2025 results from 01.05.2026 + the 11.05.2026 correction, H1 2026 interims from 29.05.2026, Q2 2026 production from 23.07.2026, the AGM result from 19.06.2026, the 01.09.2026 management change); yfinance fills in the quarterly series. The raw SRB symbol returns the wrong company (“1314”) and an EMPTY data pack — the valid data pack for this analysis is data-pack-SRB.L-20260904.md.

⚠️ What does NOT exist yet: the H1 2026 financial statements. AIM requires them before 30.09.2026, and the 01.09.2026 RNS confirms they’ll be published by then. So the latest complete financial figures are Q1 2026 (31.03.2026); for Q2 2026 I only have production and cash. Any statement about Q2 margins is an inference, flagged as such.

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

Serabi Gold is a micro-cap gold producer with two high-grade underground mines in the Tapajós district, Pará state, Brazil — Palito (the central plant, in production since 2012) and Coringa (in production since 2022, no plant of its own, with ore pre-concentrated and trucked to Palito). In 2025 it delivered a record year: 44,169 oz (…), revenue …M (…), net profit …M (…), the first dividend in company history (…, ~…M), and, since January 16, 2026, a balance sheet with no financial debt. At 30.06.2026 it held …M cash and zero bank debt.

The central thesis, versus the 26.07.2026 report, has partly flipped — and the reason is arithmetic, not narrative. The prior report normalized gold at …/oz and came out with an intrinsic value of 165-… against a price of …, i.e. “expensive.” That normalization is no longer defensible in September 2026: gold’s annual average was … (2024), … (2025), and … in 2026 so far, with spot at … on 03.09.2026 (GC=F). Using … means assuming a 43% crash from spot — a legitimate bear scenario, not a base case. Renormalized at …/oz (roughly the 3-year average, … i.e. minus 20% versus spot) and with the already-observed elevated costs (Q1-2026 cash cost …/oz, AISC …/oz), normalized owner earnings come out at …M/year, and intrinsic value at ~… — 29% above the current price.

The reverse check is the figure that matters most: at …, the market is paying for a gold price of …/oz sustained forever (perpetual EPV, 50,000 oz/year, cash cost …/oz, 12% discount rate, plus net cash). I.e. the market already assumes a permanent 27% decline versus spot and 4% below the last-three-years average. That’s a real margin of safety on the single most important assumption for a commodity producer — not an extrapolation of the peak.

The five triangulated models give … (bear, gold …) … … (bull, gold … sustained 10 years), with the median at … (…). Monte Carlo (20,000 scenarios over the same DCF model as the tracker, with OE …M ± 25%) gives median … MOS, P10 …, P90 …, and ** … probability of undervaluation**. Earnings quality is HIGH: CFO has exceeded net profit in each of the last four years (Sloan ratio … in 2025), the effective tax rate rises (13.8% → 21.5% → 23.5%), amortization grows faster than revenue, and exploration was tripled, not cut.

What keeps the verdict in check are two concrete things, not vague ones. (1) Coringa’s licensing: the provisional GUIA license, capped at 100,000 t/year, expires January 29, 2027 — under five months away. The permanent installation license (LI) still needs the land-use-change approval from INCRA and the filing of the indigenous-component study with FUNAI. The 53,000+ oz guidance is explicitly conditioned on “LI granted by Q4-2026.” Coringa produced 14,988 of 23,049 oz in H1-2026 — 65% of the group’s production hangs on an authorization that doesn’t yet exist. (2) A cluster of governance issues, not an isolated incident: at the 18.06.2026 AGM shareholders rejected both the remuneration report (52.34% against) and the remuneration policy (50.37% against); the largest shareholder (Classe Roca Magma FIP, 24.99%, managed by Starboard Asset Ltda) abstained on ALL resolutions — only 20.2% of share capital actually voted; a director who sits simultaneously on the audit committee AND the remuneration committee took 28.87% against; and the 13-year CFO, re-elected with 95.83% on June 18, left on August 28, with no stated reason, four weeks before the half-year results, and the interim isn’t a board member.

Estimated value: 224-…, center 374-…. Verdict: PARTIAL BUY — small tranche (max. 1/3 of normal allocation), with explicit triggers. The valuation justifies a position; the governance cluster and the January 2027 deadline justify the small sizing and deferring the rest until (a) publication of the H1-2026 statements and (b) the LI is granted. Prior GBL score from tracker: 56.7% → MONITOR (evaluated 26.07.2026). The divergence is real, but smaller than it looks — see the final chapter.