SLP — Sylvania Platinum Limited

Deep-value analysis — Sylvania Platinum Limited (AIM: SLP)

September 18, 2026 · reference price … GBp · market cap GBP 243.2 mil (USD 325.1 mil at GBP/USD 1.3368) · 258.72 mil shares outstanding · fiscal year ended June 30

The methodological note that must be read before any figure

This run began with a symbol collision resolved along the way, and the reader needs to know that in order not to compare this report with older artifacts generated on the wrong company.

The bare SLP symbol resolves on Yahoo Finance to Simulations Plus, Inc. (Nasdaq, pharmaceutical simulation software, USD reporting, August fiscal year, P/E 46, Healthcare sector). The data pack generated automatically ahead of this session (data-pack-SLP-20260918.md, the 12:50 version) and the EDGAR archive downloaded to SEC-Filings\SLP\ are for that company — a software maker with 20.2 million shares and … of annual revenue. The SLP ticker in Radu’s tracker is, per query_tracker.py indicatori SLP, Sylvania Platinum Limited, AIM exchange — a platinum-group-metals producer with 258.7 million shares and … of revenue. The two have absolutely nothing in common.

What I did: I added the alias "SLP": ("SLP.L", "Sylvania Platinum Limited") to _BVB_ALIASES in deep_data_pack.py and regenerated the data pack. The current version on disk (12:51) is correct and carries the header Sylvania Platinum Limited (SLP.L) · Price: … · Currencies: reporting USD / price GBp ⚠️ DIFFERENT. I left a _COLIZIUNE-SIMBOL.md file in SEC-Filings\SLP\ warning that the 10-K/10-Q filings there belong to Simulations Plus, so a future run doesn’t read them as the Sylvania file.

The consequence for primary sources, explicitly declared: Sylvania Platinum is a Bermuda-registered company, listed on AIM, with operations in South Africa. It does not and will never file with the SEC. So there’s no .txt + _index.md package with chapters and line ranges for this ticker, which I’d normally use as a primary source. The sources the figures below rest on, in order of confidence:

1. Aggregated consolidated financial statements from XBRL/yfinance for SLP.L — a full annual balance sheet over four fiscal years (FY2023–FY2026, June), an annual income statement over four fiscal years, an annual cash flow statement over four fiscal years, a semi-annual balance sheet over four periods. These are the audited figures reported by the company, mechanically aggregated. They form the report’s backbone.
2. The H2/FY2026 earnings call transcript (investing.com, ~2026-09-16) — for dividend policy, capex structure, the loan to the Thaba JV, cost per ounce, and capital-allocation priorities.
3. The FY2026 results RNS announcement (via ADVFN, 09-15–16, 2026) — to confirm headline figures.
4. The research brief research-SLP-20260918.md — treated as testimony, not truth. Two corrections made below.

What I could NOT verify in the primary source, and why, declared here once for the whole report: (a) the breakdown of FY2026 capex between sustaining and expansion capex — the transcript states verbatim that the “breakdown [is] not separately disclosed,” so the company doesn’t publish it; (b) the identity of the asset that generated the … impairment in FY2026 — it appears in the accounting tags, but not in any of the sources accessible within the three-web-search cap; (c) the person-by-person breakdown of the 11.5% held by insiders. These three gaps are flagged as such where they matter, not filled in with assumptions.

Two corrections to the research brief, both checked against the financial statements:

• The brief states “EPS: 25.66 pence.” This is wrong: it’s 25.66 U.S. cents. Net profit of … divided by 258.8 mil shares gives …/share, and at the GBP/USD rate of 1.3368 that means … pence. Cross-check: … GBp / … GBp = P/E 4.90, exactly the P/E of 4.947 calculated independently in the data pack. Had someone used 25.66 pence, they’d have gotten a P/E of 3.66 and a false earnings yield of 27% — an error of a third on the thesis’s central metric.
• The brief states a buyback program of “up to …lion,” with no date. The transcript gives the exact figures: … executed in FY2026, a new … program announced alongside the results. The cash flow statement confirms … on the Repurchase Of Capital Stock line for FY2026.

A third observation, not a correction: the brief couldn’t reconcile analyst targets of 107.61 GBp (fintel/nasdaq), 164 GBp (stockanalysis), and 175 GBp (RBC, initiation June 2026). I didn’t reconcile it either — I spent the three-web-search cap on capex, dividend policy, and confirming the headline figures, which matter more for valuation than consensus. I treat consensus as a wide, uncalibrated range, not an anchor.

A currency error in third-party data, flagged because it propagates silently: yfinance reports an enterpriseValue of 176.8 million for SLP.L. The figure is obtained by subtracting cash in USD (67.2 mil) from a market cap in GBP (243.2 mil). The correct enterprise value is … − … of net cash = … = …. Any EV multiple calculated from the yfinance figure is ~25% too low. This is exactly the class of error documented by the working note on mismatched currencies in the data pack’s header.

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

What Sylvania Platinum is. A company that doesn’t dig. It buys and reprocesses chrome tailings — other companies’ mining waste — from the Bushveld Complex in South Africa, extracting from them platinum-group metals (platinum, palladium, rhodium, gold — the “4E basket”) and, since 2025, chrome concentrate. It operated six reprocessing plants (Sylvania Dump Operations) through August 2025, when it commissioned a seventh, Thaba JV, in partnership with Limberg Mining Company. FY2026 produced 95,885 4E ounces, a record, at a cash cost of …/ounce group-wide, in a year the 4E basket averaged …/ounce.

The thesis in three sentences. Sylvania is an unusually clean balance sheet — zero goodwill, zero intangibles, … thousand of total debt, … of net cash (20.5% of market cap) — attached to a violently cyclical cash flow, which the market prices, on the peak year’s earnings, at 4.8× profit and 2.3× EBITDA. On the average earnings of the last four years, the same figures become 9.5× profit and 4.4× EBITDA — i.e. a cheap company, but not absurdly cheap, and the entire difference between “historic bargain” and “reasonable price” sits in the answer to a single question: where we are in the rhodium cycle. What makes the case interesting isn’t the multiple, it’s the fact that the years of expansion capex have ended — building Thaba consumed … in FY2025 and … in FY2026, and the asset is already producing revenue (… of attributable chrome revenue, 4.2% of net revenue), so future free cash flow no longer has to finance a new plant.

The estimated value. Five independent models, all starting from normalized owner earnings of … (not the … reported in FY2026, and even less the … this year would give adjusted for working capital), give a range of … to … margin of safety, with a median of …. The Monte Carlo simulation over 20,000 scenarios, with λ = 0.75 applied to net cash, gives a median intrinsic value of 116 GBp versus the price of 94 GBp, a P10–P90 range of … to …, and an undervaluation probability of …%. The range is honestly wide: not because the model is poorly calibrated, but because a company whose revenue has swung between … and …lion over four years, on the same physical asset base, has a wide intrinsic value.

Verdict: INTERESTING — gradual accumulation. Re-scoring the 30 GBL criteria, with the mandatory cyclical normalization applied, gives …% (Graham 9.0 · Buffett 6.5 · Lynch 6.5 · F-Score 7), down from 75.0% in the tracker; the entire difference comes from the PEG and profit growth, calculated on peak earnings, dropping to zero after normalization. The arguments in favor: a debt-free balance sheet with no intangible assets, high earnings quality (negative accruals in three of four years, operating flow exceeding net profit in FY2026 by …), a continuously falling share count (267.6 → 258.8 million over four years), a formal dividend policy of a minimum 40% of adjusted free cash flow that the company has consistently exceeded, and … returned to shareholders over the last four years — 64% of cumulative net profit and 120% of cumulative free cash flow. The arguments against, which justify gradual accumulation instead of a conviction position: exposure to the rhodium price is smooth and undiversified, the dividend was cut by roughly 85% in FY2024 and isn’t a coupon, the … impairment in FY2026 remains unexplained in accessible sources, and management just announced its intent to diversify into manganese, cobalt, nickel, and lithium and into Namibia, Botswana, and Zambia — the kind of announcement that, at a company with a proven niche model, spends capital on unfamiliar businesses.

What would make me change my mind. A second year of raw-material inventory growing faster than cost of goods sold (DIO has already climbed from 31.8 to 42.0 days), a real acquisition outside the Bushveld PGM niche, or identification of the … impairment showing a plant with depleted tailings resources — the latter would mean the business’s life is shorter than any model with positive terminal growth assumes.