alirezarezvani/claude-skills / finance/skills/stock-analysis
stock-analysis
UnverifiedAnalyze, research or value a listed stock or company (US, global, NSE or BSE) with verified data, ratios, scores and valuation. Use whenever asked if a stock is a buy, what it is worth, or how a company looks.
references/sectors/metals-mining.md
View on GitHub# Metals, mining and commodity producers — sector playbook
Use this when: the company digs, processes, smelts or refines a physically traded commodity and takes the price the market gives it — iron ore, coal, copper, zinc, lead, aluminium, gold, silver, nickel, lithium, bauxite, alumina, integrated and secondary steel, ferro-alloys, sponge iron, and the mining-adjacent explorers and developers that have reserves but no earnings.
Two assumptions underneath the generic ratio checklist both fail here, and they fail in the same direction. First, the company does not set its selling price: the price is exogenous, identical for every competitor, and moves several hundred percent across a cycle. Second, the asset base is not a going concern of indefinite life — every tonne sold is inventory liquidation from a finite orebody, and the mine eventually stops and must be closed at the owner's cost. The consequence is that almost every standard ratio is not merely noisy here but *inverted*: it looks best at the top of the cycle and worst at the bottom, which is precisely backwards. Your job is to replace price-contaminated financial ratios with unit economics (cost per tonne against the global cost curve), physicals (grade, strip, recovery, volume) and through-cycle normalisation.
## Contents
1. [Why the generic ratio set fails here](#why-the-generic-ratio-set-fails-here)
2. [The metrics that actually matter](#the-metrics-that-actually-matter)
3. [How to value companies in this sector](#how-to-value-companies-in-this-sector)
4. [Peer set construction](#peer-set-construction)
5. [Sector-specific red flags](#sector-specific-red-flags)
6. [Cycle and structural context](#cycle-and-structural-context)
7. [India vs global notes](#india-vs-global-notes)
8. [Checklist](#checklist)
---
## Why the generic ratio set fails here
Work through these before you quote any standard ratio. In most cases the correct action is to suppress the metric or replace it with the through-cycle version, not to caveat it.
**P/E — actively inverted, and the single most dangerous number in the sector.** Earnings peak when the commodity peaks, so the stock screens cheapest (3–5x) at exactly the moment the cycle is about to roll, and screens expensive or loss-making (negative, or 40x+) at the trough when the risk-reward is best. Screening for "low P/E in metals" is a reliable mechanical method of buying the top. The correct read is the opposite of the generic one: **high or negative P/E at a cyclical trough is a potential entry; a mid-single-digit P/E on peak earnings is a warning.** Only normalised, mid-cycle EPS makes a P/E interpretable at all.
**OPM / EBITDA margin — a commodity-price artifact, not a measure of management.** A fourth-quartile-cost miner prints 40% margins at peak prices; a first-quartile miner prints 15% at the trough. Margin tells you where the *price* is, not how good the business is. Ranking peers on OPM in this sector ranks them on when you happened to look. Replace it entirely with **unit cash cost versus the global cost curve** and **EBITDA per tonne against the company's own history**.
**ROCE — doubly distorted, and anti-comparable across peers.** Self-discovered reserves are carried at near-zero (exploration expensed or written down); acquired reserves are carried at full purchase price. Two geologically identical mines can therefore show 40% and 8% ROCE purely because one was found and one was bought. Add fully-depreciated legacy assets flattering the denominator, and a just-commissioned greenfield with capitalised capex and no output crushing it, and the raw number carries almost no information. Compute ROCE on **mid-cycle EBIT** and ideally against **replacement cost of installed capacity**, not book.
**D/E and net debt/EBITDA — procyclical, and they lie in the same direction as P/E.** 0.8x net debt/EBITDA at the peak becomes 5x on *identical debt* when the price halves. Leverage looks most reassuring exactly when the risk is highest. Test leverage against **trough EBITDA**, and make sure the "debt" figure includes leases, acceptances/bill discounting, perpetual and hybrid instruments, mine-closure and rehabilitation provisions (ARO), and — India-specific — the discounted value of committed auction-premium and DMF obligations.
**Revenue growth — meaningless as reported.** Revenue is price × volume, and price moves far more than volume. A miner can post +40% revenue with *falling* production. Never report revenue growth in this sector without decomposing it; only **saleable volume growth** is a real KPI, and only realised-price-vs-benchmark explains the rest.
**FCF — the easiest number in the sector to manufacture.** Capex is lumpy and splits three ways: sustaining, capitalised waste stripping, and growth. A company that defers waste stripping, fleet rebuilds, furnace relines and tailings-dam spend prints excellent free cash flow for two or three years while quietly liquidating the orebody and accumulating risk. Split reported capex into sustaining vs growth, compare sustaining capex to DD&A and to life-of-mine average stripping, and only then treat FCF as a quality signal.
**D&A comparability is broken, which breaks EBIT, OPM and P/E together.** Units-of-production versus straight-line depreciation, and different assumed mine lives, change reported EBIT between otherwise identical peers with no cash difference at all. This is *why* the sector talks in EBITDA and unit cash costs rather than operating profit — use the sector's own language rather than forcing EBIT-based ratios onto it.
**P/B and book value — unstable and not a floor.** Serial impairments, revaluation of acquired reserves, and full-cost versus successful-efforts treatment of exploration make book value a residue of past accounting choices rather than a measure of asset value. A "low P/B" in metals frequently means the impairments have not been taken yet. Use EV per tonne of capacity against greenfield replacement cost instead — that is the real asset-value floor.
**Current ratio, inventory turns and working-capital rules — contaminated.** Inventory revaluation gains and losses run through cost of sales, so turns move with price rather than efficiency. Concentrate sales are provisionally priced: the final price is set on a quotational period one to four months after shipment, so a quarter's receivables and EBITDA carry marks that reverse next quarter with zero operational change. Do not read working-capital deterioration as distress without first stripping revaluation and provisional-pricing effects.
**DCF with a perpetual-growth terminal value is simply wrong for a mine.** Reserves deplete; the asset stops. A Gordon-growth terminal value assumes the opposite of the defining fact about the business. The correct model is a **finite life-of-mine NPV with an explicit closure cost at the end**. (For processors and recyclers with no orebody — rolled aluminium, EAF steel, secondary refiners — a going-concern DCF is defensible; for a mine it is not.)
**Trailing dividend yield — a peak-cycle mirage.** Most majors now run base-plus-variable payout policies explicitly designed to distribute windfalls. A trailing yield of 10–12% is therefore a signal that you are near the top, not that you have found an income stock. Value the **base** dividend as the sustainable component and treat the variable component as a cyclical windfall.
**"Consistency" screens systematically select the wrong companies.** Filters for 10-year EPS CAGR, no loss-making years, stable margins and steadily rising book value reject essentially every good miner (they all have loss years) and select the ones whose numbers were flattered by a peak. If your framework contains such a screen, disable it for this sector rather than applying it and noting an exception.
**India-specific blind spots in generic checklists.** A large share of the cost structure is statutory and non-negotiable — royalty, DMF, NMET, GST compensation cess on coal, and auction premiums that can exceed 100% of the IBM-notified average sale price. None of this appears in any standard ratio, yet it can consume half of gross revenue. Separately, many Indian producers do not publish JORC/NI 43-101/SK-1300-compliant reserves at all, so the reserve-based metrics that developed-market analysis leans on must be reconstructed from lease documents, approved Mine Plans and IBM filings, or flagged as unavailable.
---
## The metrics that actually matter
Ranges below are **indicative only**. They vary by commodity, market, jurisdiction, cycle position and reporting period, and several are quoted in nominal dollars that drift with inflation. A company's own 10-year history and its direct sub-sector peers at the *same point in the cycle* always override any absolute band quoted here.
| Metric | Definition / how to compute | Indicative healthy range | Why it matters |
|---|---|---|---|
| **Unit cash cost and position on the global cost curve** (C1; AISC for precious metals) | C1 = mining + processing + on-site G&A + freight, net of by-product credits, per unit of saleable output. AISC (World Gold Council definition) = C1 + sustaining capex + sustaining stripping + royalties + corporate G&A + reclamation accretion. What matters is the **quartile** on the global cost curve (Wood Mackenzie / CRU / AME), not the absolute number. | First or second quartile; ideally first. Reference points: copper C1 below ~$1.60/lb (first quartile ~$1.00–1.25/lb); gold AISC below ~$1,400/oz against a global average of roughly $1,450–1,600/oz; iron ore C1 of $18–30/t for the seaborne majors; large Indian zinc producers guiding cost of production ex-royalty near $1,000–1,100/t against LME $2,600–3,000/t. Above the 75th percentile is a survival question, not a valuation question. | Price is exogenous and identical for everyone, so the only durable competitive advantage in this sector is being low on the cost curve. Cost position determines who still makes money in the trough (when high-cost tonnes shut and set the marginal price), who survives without a rescue rights issue, and who compounds. **This metric replaces operating profit margin entirely.** |
| **EBITDA per tonne / per ounce (unit spread)** | EBITDA ÷ saleable volume *sold*, tracked by product and by asset. For processors also track the conversion spread separately: HRC price minus roughly 1.6t iron ore plus 0.7t coking coal; LME aluminium minus ~1.9t alumina, ~0.4t carbon and ~14 MWh of power. | Indian flat steel: mid-cycle roughly Rs 9,000–12,000/t; below Rs 6,000/t is stress; above Rs 18,000–20,000/t is an unsustainable peak. Global integrated steel $100–150/t mid-cycle. Indian primary aluminium $700–900/t mid-cycle, $1,100/t+ at peak. Rolled-products converters ~$500–550/t of shipments and largely spread-agnostic. | Normalises out price and volume mix and isolates operating performance, which reported margin cannot. It is the unit in which the sector guides, forecasts and values, and the input to mid-cycle EV/EBITDA. A company whose EBITDA/t consistently exceeds peers *at the same realised price* has real operating advantage (integration, scale, energy, yield); one whose EBITDA/t is only high when prices are high has none. |
| **Reserve life (R/P), resource life and reserve replacement ratio** | Proved + probable reserves ÷ current annual production = years of mine life; the measured/indicated/inferred resource behind it; and reserve replacement ratio = reserves added ÷ reserves depleted, on a rolling 3-year basis. Always check the reporting code: JORC (Australia), NI 43-101 (Canada), SK-1300 (US, mandatory since 2021), PERC (Europe) versus India's UNFC/IMIC 2019 framework, which is far less consistently applied. | Core assets: reserve life above 15–20 years. Below 8–10 years the company must find or buy reserves under time pressure — where value is destroyed. Replacement ratio above 100% on a 3-year average. India: captive leases run to 2030/2050/2070 depending on MMDR vintage — check the **actual expiry date**, not the geology. | A miner is a wasting asset. Short reserve life converts an apparently profitable business into a forced acquirer at the top of the cycle. Long life plus in-house replacement is what separates a compounding franchise from a run-off. Watch restatements: reserves can be "added" by raising the assumed long-term price or lowering the cut-off grade — an accounting event, not a discovery. |
| **Head grade vs reserve grade, strip ratio, metallurgical recovery** | Grade = metal content per tonne of ore (g/t Au, % Cu, % Fe, % Zn). Strip ratio = waste tonnes moved per tonne of ore. Recovery = % of contained metal actually recovered in the plant. Compare mined grade and strip against the **life-of-mine (LoM) plan averages** in the technical report. | Mined grade at or below reserve grade, and strip ratio at or above LoM average, is normal and healthy. Grade materially above reserve grade, or strip materially below LoM average, for several consecutive periods is the classic high-grading / deferred-stripping pattern. Recovery stable or improving; a 2–3 point recovery decline is a real earnings event. | These are the physical drivers of everything downstream. Falling grade means more tonnes must be moved, crushed and processed for the same metal, raising unit cost mechanically regardless of management effort — the structural headwind in global copper and gold. Conversely, mining above reserve grade flatters today's cost and cash flow while stealing from future years. **No financial metric reveals this; only the physicals do.** |
| **Saleable production volume, sales vs production, capacity utilisation** | Volume actually *sold* (not produced), the gap between production and sales (stockpile build/draw), and utilisation of installed smelter, refinery, blast-furnace and mill capacity. Track debottlenecking-led volume growth and the ramp curve of new capacity against guidance. | Steel/alumina/refinery utilisation above 85–90%; below 75–80% fixed-cost absorption collapses. Sales should track production closely; a persistent finished-stock build is a demand problem or an evacuation/logistics problem. New projects should reach design capacity within 4–6 quarters. | Volume is the only part of revenue management actually controls. In a fixed-cost-heavy business incremental tonnes carry very high contribution margin, so a 5% volume miss can cost 15–20% of EBITDA. Serial volume guidance misses are the most reliable early indicator of orebody, equipment or permitting problems, and the number on which the market re-rates. |
| **Realised price versus benchmark (realisation discount / NSR)** | Achieved price per tonne or ounce against the relevant benchmark (LME/COMEX, Platts 62% Fe CFR China, HRC import parity, Newcastle/API coal indices), plus the reason for the gap: product mix, grade and impurity penalties (silica, alumina, phosphorus, moisture), treatment and refining charges (TC/RC) on concentrate, freight, provisional-pricing marks, and contract vs spot vs e-auction mix. | The gap should be **stable and explainable**. Widening discounts signal deteriorating ore quality or product mix. India: the large coal producer's realisation blends FSA notified price with e-auction volumes that can clear 40–100%+ above notified price — the auction share is a major swing factor. Indian steel NSR should track import parity within a narrow band. | Two producers quoting the same benchmark can realise materially different revenue per tonne. Persistent widening of the discount is a silent margin leak that no headline cost metric shows, and provisional pricing can flip a quarter's reported EBITDA in either direction with no operational change. |
| **Sustaining capex per unit, and sustaining capex vs DD&A** | Capital required to keep current production and safety standards intact — mine development, capitalised stripping under IFRIC 20, tailings dams, fleet rebuilds, furnace relines — separated from growth capex, expressed per tonne/ounce and compared with DD&A and the LoM plan. | Over a full cycle sustaining capex should be broadly in line with DD&A. Gold roughly $150–300/oz. Integrated steel roughly $25–45 per tonne of capacity per year, plus relines every 15–20 years. Sustaining capex below 60–70% of DD&A for multiple years is deferred maintenance dressed up as free cash flow. | The easiest lever for flattering short-term FCF and the one causing the most permanent damage — and, for tailings dams, catastrophic risk. Splitting sustaining from growth is also what makes FCF interpretable at all: growth capex is optional and can be stopped; sustaining capex cannot. |
| **All-in cash break-even commodity price** | The commodity price at which the company generates zero free cash flow after cash costs, royalties and levies, sustaining capex, interest and cash tax. Express per tonne/ounce and compare with spot and with the consensus long-run/incentive price. | Spot at least 30–40% above all-in break-even for a comfortable balance sheet. A producer whose break-even sits within 10% of spot has no cushion. Also compare break-even with the marginal industry cost (≈90th percentile of the cost curve), which anchors the long-run floor price. | Converts the entire income statement into the one number that matters in a downturn: at what price does this company stop self-funding and start drawing debt? Far more decision-useful than interest coverage or D/E, and it lets you size downside directly from a price deck. |
| **Through-cycle leverage: net debt / mid-cycle EBITDA and / trough EBITDA** | Net debt — including leases, acceptances, perpetual and hybrid instruments, discounted closure/rehabilitation provisions, and in India the NPV of committed auction-premium and DMF liabilities — tested against normalised mid-cycle EBITDA *and* a modelled trough EBITDA. Plus gearing = net debt / (net debt + equity). | Diversified majors run net cash to 0.5x mid-cycle and gearing under 20–25%. Indian integrated steel up to 2.0–2.5x mid-cycle is workable given growth capex; above 3.5x mid-cycle is dangerous. Real test: **net debt / trough EBITDA under 4–5x, and no refinancing wall inside 24 months.** | Almost every permanent capital loss in this sector is a leverage event, not an operating event — debt taken on at the peak against peak EBITDA, then refinanced dilutively at the trough. Reported net debt/EBITDA is most reassuring exactly when risk is highest, so only the mid-cycle and trough versions carry information. |
| **Raw-material integration ratio (captive ore, coking coal, bauxite, power)** | Share of key input requirements met from own mines or captive power. Indian steel: captive iron ore and coking coal percentages. Aluminium: captive bauxite, alumina and captive coal-based power (~14 MWh per tonne of aluminium). | Indian integrated steel spans near-100% captive iron ore with 20–25% coking coal at one end to roughly 40–50% iron ore and minimal coking coal at the other. Each ~10 percentage points of iron-ore integration is worth roughly Rs 500–900/t of EBITDA at normal ore prices. For aluminium, captive bauxite plus captive power is the difference between first- and third-quartile cost. | Integration converts a price-taking converter into a structurally low-cost producer and is the main reason Indian producers' EBITDA/t diverges so widely. It also changes cycle behaviour: an unintegrated converter is squeezed from both ends, while an integrated producer captures the mining margin. Verify integration is real and long-dated (lease expiry, evacuation approvals), not a signed MoU. |
| **Energy and power cost per unit, and specific consumption** | Energy cost per tonne plus physical intensity: MWh per tonne of aluminium, GJ or kg of coke per tonne of hot metal (coke rate / fuel rate), diesel per BCM moved. Plus source mix — captive thermal, linkage coal, e-auction coal, imported coal, grid, renewables / round-the-clock PPAs. | Aluminium smelting 13.5–15 MWh/t (below 14 is efficient). Blast-furnace fuel rate 480–520 kg/thm. Energy is typically 30–40% of aluminium cash cost and 20–30% of mining cost, so a 10% energy move is a first-order earnings event. | The largest controllable cost line in smelting and a very large one in mining, and the axis on which decarbonisation capex, carbon pricing (EU CBAM phasing in on steel and aluminium from 2026) and coal-linkage policy will play out. Producers with captive coal and cheap renewables PPAs have a durable edge; smelters exposed to spot power do not. |
| **Royalty, cess and total government take as % of revenue** (especially India) | All statutory levies as a share of revenue: royalty (India: iron ore 15% ad valorem, coal 14%, plus bauxite and zinc rates), District Mineral Foundation (10–30% of royalty), NMET (2% of royalty), GST compensation cess on coal (Rs 400/t), and the auction premium bid as a % of IBM average sale price for post-2015 blocks. Globally: state royalties, resource-rent taxes, export duties. | For an Indian post-2015 auctioned iron ore block with a 100–150% premium bid, total government take can exceed 45–60% of gross revenue — such a block is uneconomic at low ore prices. Legacy/captive royalty-only leases are dramatically better. Developed-market take is typically 5–15% of revenue plus corporate tax. | The most under-modelled cost in Indian mining, and it is **fixed regardless of the price cycle**, which massively amplifies downside operating leverage. Companies that won auctions with aggressive premium bids bought volume, not value. This line alone explains large permanent EBITDA/t divergence between peers mining the same ore. |
| **Mid-cycle ROCE, measured against replacement cost** | EBIT at normalised mid-cycle commodity prices ÷ capital employed, ideally adjusted so capital employed reflects the replacement cost of installed capacity rather than a mix of near-zero self-discovered reserves and marked-up acquired reserves. | Above 15% at mid-cycle prices is a high-quality asset base; 8–12% is average; below the cost of capital at mid-cycle means the business destroys value across a full cycle no matter how good the peak year looks. Benchmark against the ~8–10% real WACC used for base-metal projects. | Answers whether the company earns its cost of capital through a full cycle rather than in one good year. The replacement-cost adjustment removes the distortion that makes acquirers look permanently bad and old fully-depreciated assets look permanently brilliant. It is also a discipline test: managements that report and are paid on mid-cycle ROCE historically avoid peak-cycle M&A. |
| **FCF yield at spot AND at mid-cycle, with price sensitivity** | Operating cash flow less sustaining capex, interest, tax and leases ÷ market cap or EV — computed **twice**, once at spot and once at a normalised long-run price deck. Add explicit sensitivity: EBITDA and EPS impact per $10/t change in the ore or metal price, per Rs 1 change in USD/INR. | Mid-cycle FCF yield of 6–10% on EV is attractive. A **spot** FCF yield of 15–25% is almost always a cycle-peak signal, not a value signal. Sensitivity should be in the annual report; if it is not, build it. | This is the discipline that stops you buying the peak. The spot-versus-mid-cycle gap tells you how much current cash flow is price windfall versus structural. The sensitivity table tells you the true equity risk: for a levered producer a 20% price fall can erase 60–80% of EBITDA and all of the free cash flow. |
| **Rehabilitation / closure provisions and tailings risk profile** | The balance-sheet ARO provision for mine closure, tailings and land rehab; the discount rate and cost-inflation assumption used; the funding status (cash-backed vs unfunded); and the physical inventory of tailings dams by construction type (upstream vs downstream/centreline), consequence classification, and independent GISTM conformance. | Provisions should grow with disturbed area and be funded or covered by dedicated reserve accounts where jurisdictions require it (Australia, Canada, Brazil). Upstream-construction dams in high-consequence locations are a distinct, material risk and are now banned in Brazil. Look for a stated GISTM conformance date and an independent tailings review board. | Closure liabilities are debt in disguise: unfunded, long-dated, inflation-linked, and easily understated with a high discount rate or a stale cost estimate. Tailings failures have destroyed more shareholder value in single events than most commodity downturns, and post-failure liabilities are effectively uncapped. |
| **By-product credit dependence and co-product mix** | Share of C1 cost offset by by-product credits (silver and gold in zinc/lead, molybdenum and gold in copper, cobalt in nickel), and whether the company nets credits against cost or uses co-product accounting. | Know the number. A "first-quartile" cost position that depends on 30–50% of cost being offset by a by-product is a leveraged bet on *two* commodities, not one. Prefer co-product presentation for comparability. | By-product netting is the most common way a mid-cost mine is presented as a low-cost mine. When the by-product price falls, the headline cash cost jumps with no operational change — and peers using co-product accounting will look artificially worse. |
| **Safety and social licence: fatalities, LTIFR, community/permit status** | Fatalities and lost-time injury frequency rate per million hours, plus the status of forest and environmental clearances, community consent, and litigation. | Zero fatalities is the only acceptable target; a rising LTIFR or a fatality cluster is a leading indicator of deferred maintenance and production pressure. | Safety performance is the cleanest available proxy for operational discipline, and it is causally linked to the cost-cutting that precedes both accidents and production misses. Loss of social licence — protest, blockade, permit revocation — has stopped large producing mines outright and is not priced in any financial ratio. |
**Sourcing notes.** Globally: reserves, grade, strip, LoM plans and unit costs are in the technical report (NI 43-101 / SK-1300 / JORC) and the annual production report, not the financial statements; US filers must give reserves and cost detail in the 10-K under SK-1300. In India: production and realisation come from the monthly/quarterly volume release and the concall; royalty, DMF and auction-premium detail from the annual report notes, lease documents and IBM filings; capitalised stripping from the fixed-asset and accounting-policy notes. If reserve disclosure is absent (common in India), say so explicitly rather than substituting resource figures of unknown confidence.
---
## How to value companies in this sector
The sector is valued on **mid-cycle cash flow, asset value and replacement cost — never on trailing earnings multiples.** Present the answer as a range across a bear/mid/bull commodity price deck with explicit EBITDA and FCF sensitivity per unit price move. A single point-estimate target price in this sector is close to meaningless.
**1. EV/EBITDA on normalised mid-cycle EBITDA — the primary working multiple.** It strips out the D&A distortion (units-of-production vs straight-line, acquired vs self-discovered reserves) and the very different leverage structures. Indicative anchors: diversified majors 4.5–6.0x mid-cycle; Indian integrated steel 6–7x (a structural growth premium over global steel at 4–5x); gold producers 6–9x; thermal coal 3–4x because of terminal-value risk. **The multiple must contract as you move up the cycle** — paying 6x on peak EBITDA is paying 12x on mid-cycle. If you find yourself applying a constant multiple to a rising EBITDA estimate, you have built a momentum model, not a valuation.
**2. P/NAV — the standard method in developed markets**, especially Canada and Australia for precious and base metals. Build an explicit life-of-mine DCF from the technical report: annual tonnes and grade, recovery, unit costs, sustaining and expansion capex, taxes and royalties, and closure cost at the end — with **no perpetual-growth terminal value**, because the asset simply stops. Discount at roughly 5% real for gold, 8–10% real for base metals and bulks, and 12–15%+ where jurisdiction or terminal-demand risk is severe. Then add corporate items and net debt. Benchmarks: producing majors 0.9–1.3x NAV, single-asset producers 0.5–0.8x, developers and juniors 0.3–0.6x. Above 1.5x you are paying for exploration optionality or for a commodity price above your deck — say which.
**3. EV per tonne of installed capacity / price-to-replacement-cost — the cycle-trough anchor**, particularly for processors. Greenfield integrated steel replacement cost is roughly $900–1,200 per tonne globally and roughly $550–800 per tonne for brownfield expansion in India. Buying well below replacement cost is the classic deep-value entry; paying a premium to it is only justified where the asset is captive-ore-integrated and genuinely cannot be replicated (mining leases are no longer freely available in India post-auction).
**4. EV per unit of reserve or resource** — EV per reserve ounce of gold, per pound of contained copper, per tonne of contained iron ore. Very rough and highly sensitive to grade and jurisdiction, but it is the only tool for pre-production explorers and developers with no earnings, and it is useful for cross-sectional screening.
**5. Sum-of-the-parts — essential for diversified and holding structures.** Value each commodity stream on its own appropriate multiple, then subtract a holding-company discount (typically 20–40% where a listed parent's main asset is a stake in a listed subsidiary). Typical structures requiring SOTP: a parent holding a listed zinc subsidiary plus aluminium, power and other unlisted assets; an aluminium producer whose downstream rolled-products arm should be valued on $/t of shipments rather than on the upstream multiple; an Indian steel producer whose domestic business deserves Indian multiples while a loss-making European business deserves a heavy discount or negative value.
**6. Mid-cycle EPS × a mid-cycle P/E** is a legitimate cross-check — but only with normalised EPS, never trailing.
**7. For dividend-oriented names**, model the base-plus-variable policy explicitly. Value the **base** dividend as the sustainable component and treat the variable component (historically taking total payout to 50–60%+ of underlying earnings in good years) as a windfall that should not be capitalised.
**What NOT to use:** trailing P/E; P/B as a value screen; DCF with a Gordon-growth terminal value for any depleting asset; EV/EBITDA on spot or peak EBITDA; trailing dividend yield; PEG; and any multiple applied uniformly across sub-sectors with different cost structures and terminal-demand risk.
**Developed-market vs India convention.** DM analysts lead with P/NAV and published long-term price decks because technical-report disclosure is mandatory and audited. In India, JORC/IMIC-standard reserve disclosure is patchy and LoM plans are rarely public, so P/NAV is seldom used outside a couple of large zinc and coal names. The practical Indian toolkit is: **mid-cycle EV/EBITDA + EBITDA per tonne × capacity + EV per tonne of capacity against greenfield replacement cost**, with an SOTP overlay for conglomerates and an explicit deduction for auction-premium-burdened leases.
---
## Peer set construction
A valid comparable here shares **cost-curve position, commodity, integration level and jurisdiction** — not merely the label "metals". Compare at the same point in the cycle, or normalise both sides to mid-cycle first.
**Splits that must not be mixed:**
- **Miner vs processor vs converter.** A pure iron-ore miner earns the mining margin and has a cost curve; an EAF steel mini-mill converts scrap and earns a spread; a rolled-products aluminium converter earns a fixed conversion margin per tonne and is largely metal-price-agnostic. Their margins, capital intensity and cyclicality are structurally different. Putting them in one table produces nonsense.
- **Integrated vs unintegrated.** An integrated steel producer with captive ore and a merchant converter buying ore at spot are different businesses with opposite cycle behaviour. Never compare their EBITDA/t without stating integration percentages.
- **Bulk vs base vs precious vs battery metals.** Iron ore and coal are freight- and quality-driven with China-concentrated demand; base metals are LME-priced with global inventories; gold is a monetary asset priced off real rates rather than industrial demand; lithium, cobalt and rare earths are thin, policy-driven markets with immature price discovery. Cost-curve logic applies to all, but valuation multiples and demand drivers do not transfer.
- **Producer vs developer vs explorer.** Producers are valued on cash flow; developers on risked NAV with a discount for financing and permitting risk; explorers on EV per resource unit and optionality. Applying a producer multiple to a developer, or vice versa, is a category error.
- **Diversified vs single-asset.** Diversification is worth a real premium because it smooths the cycle and spreads geotechnical and political risk. A single-asset producer must trade at a discount to NAV; do not treat that discount as an opportunity without underwriting the single-asset risk.
- **Jurisdiction tiers.** Australia/Canada/US/Chile-tier stability versus DRC, Indonesia, West Africa, Zambia and parts of Latin America. Jurisdiction shows up in the discount rate (5 points or more) and in the willingness of financiers to fund the next expansion — it is not a footnote.
- **Thermal coal vs met coal.** Thermal carries terminal-demand and financing risk and trades at 3–4x; metallurgical coal is tied to blast-furnace steel and has a longer runway. They are not interchangeable.
- **India-specific:** post-2015 auctioned leases versus legacy/captive royalty-only leases are economically different assets even for the same ore body. State-owned producers with administered pricing, government stake sales, and social obligations do not compare cleanly to private peers.
**Practical rule:** if you cannot state each peer's cost-curve quartile, integration percentage, reserve life and jurisdiction, you do not yet have a peer set — you have a list of tickers.
---
## Sector-specific red flags
**Capital allocation at the top of the cycle**
- Large debt-funded acquisitions or greenfield sanctions announced near a cycle peak. This is the sector's single most reliable value-destruction pattern — the 2007 mega-deals and the 2011–12 Indian capex wave both produced a decade of deleveraging. Watch for management describing peak prices as a "new structural floor" or a "structural shift in the cost curve".
- Buybacks, special dividends or upstreaming to a parent funded at the top while net debt is rising. In Indian promoter structures, watch specifically for a cash-rich listed subsidiary funding an unlisted parent via inter-corporate deposits, loans, or outsized "brand fee" and management-service charges.
- Aggressive auction-premium bids on Indian mineral blocks (at or above 100–150% of IBM notified average sale price). These convert a mine into a fixed-cost liability that is loss-making at low commodity prices. Check the premium percentage on **every block** before valuing the volume it adds.
**Borrowing from the future (invisible in financial ratios)**
- Capitalised waste stripping running well above the LoM average strip ratio (IFRIC 20): shifting operating cost into capex flatters unit cash cost and EBITDA today and guarantees higher costs later. Compare the capitalised stripping charge to the disclosed LoM strip every period.
- High-grading — mining materially above reserve grade to boost near-term output. Shortens mine life, raises future unit costs, and is invisible in every financial ratio. Detect it only by comparing mined grade to reserve grade across several quarters.
- Sustaining capex persistently below DD&A while free cash flow and dividends are showcased. Deferred relines, fleet rebuilds, mine development and tailings spend are borrowed cash flow, repaid later with interest.
- Reserve restatements achieved by raising the assumed long-run price or lowering the cut-off grade rather than by drilling. Reserves added by a spreadsheet assumption reverse when prices normalise, and typically trigger impairments.
**Disclosure and definitional games**
- Changing the cost metric definition: switching between C1, "cash cost", "total cost" and AISC; changing by-product credit treatment (netting versus co-product); or dropping AISC disclosure entirely once it deteriorates.
- Extending assumed mine life, or switching depreciation from units-of-production to straight-line, to reduce the DD&A charge — a pure accounting boost to EBIT and EPS with no cash effect.
- Serial "exceptional" and "one-off" items — impairments, restructuring, closure charges — appearing in most years. In this sector recurring impairments are the operating model of a bad asset base, not exceptions. Always look at reported earnings over a full cycle, not just underlying.
- Inventory revaluation gains and provisional-pricing (quotational-period) marks presented as operating performance. Both reverse, and both can swing a quarter's EBITDA by 15–25% with zero change in the business.
- Guidance or investor decks extrapolating peak EBITDA/t into forward years, or quoting enterprise value against spot-price EBITDA. Ask for the mid-cycle number; **if management cannot produce one, that is itself the finding.**
**Balance sheet and financing**
- A debt maturity wall inside the next 18–24 months while spot sits near or below the company's all-in cash break-even. This is the precise setup for a heavily dilutive equity raise at the bottom.
- Understated rehabilitation and closure provisions — a high discount rate, a stale cost estimate, no inflation escalation, or wholly unfunded obligations. Compare provision per tonne of disturbed ground against peers in the same jurisdiction.
- Hedging that has become speculative: large fixed-price forward sales locked in near the lows, or unhedged provisional pricing and currency exposure described as a "natural hedge". Read the derivative note, not the commentary.
**Operations, risk concentration and governance**
- Volume guidance missed repeatedly, or a new project ramping far slower than the feasibility study assumed. Serial volume misses almost always mean an orebody, metallurgy or evacuation problem management has not yet acknowledged.
- Widening realisation discount to benchmark (impurity penalties, falling Fe or Cu content, deteriorating product mix, rising TC/RCs) that is not explained in the MD&A — a silent, permanent margin leak.
- Upstream-construction tailings dams in high-consequence locations, absent or delayed GISTM conformance, or no independent tailings review board. Low probability, uncapped severity, and no financial metric will surface it.
- Single-asset, single-commodity, single-jurisdiction exposure combined with leverage. Any one is manageable; all four together mean one geotechnical, permitting or political event is terminal.
- Concentrated regulatory and jurisdictional risk with no mitigation: expiring mining leases, pending forest/environmental clearance (Stage-I vs Stage-II), NGT or Supreme Court proceedings, retrospective demands (the Odisha excess-production demands under the Common Cause judgment), state-level mining bans of the kind seen in Goa and Karnataka, and resource nationalism (Indonesian export bans, DRC, Chile, Mexico lithium, Panama).
- **India:** promoter share pledging, and related-party off-take, marketing or logistics agreements routing product to promoter-linked entities at below-market realisations.
---
## Cycle and structural context
**Where you are in the cycle changes which metrics are informative.** At the peak: leverage ratios, margins and P/E are all flattering and should be discounted; focus on capital-allocation discipline, the spot-vs-mid-cycle FCF gap, and whether management is sanctioning capex. At the trough: earnings-based metrics are useless; focus on cost-curve position, all-in break-even versus spot, liquidity and maturity profile, and EV per tonne of capacity against replacement cost. State explicitly in your output where you believe the cycle is and on what evidence — inventory levels (LME/SHFE/port stocks), the shape of the cost curve, incentive price versus spot, and industry capex intentions.
**The supply-side clock is long and that is the whole game.** A greenfield copper or iron ore project takes 7–15 years from discovery to first production and is subject to permitting risk throughout. This is why underinvestment during a trough sets up the next upcycle, and why capex announced at a peak reliably arrives into a glut. Track industry-wide capex and project pipelines, not just the company's.
**Structural cost inflation from declining grades.** Global copper and gold head grades have trended down for decades; the same tonnage of metal requires more material moved, more energy and more water. This raises the whole cost curve over time and supports higher long-run incentive prices — but it also means a company merely holding unit cost flat in real terms is outperforming.
**Demand concentration and the China factor.** Roughly half of global consumption of several base metals and bulks is China-linked. Chinese property, infrastructure and export-steel policy therefore drive iron ore, met coal and copper demand more than any Western macro variable. Chinese steel export volumes directly compress margins in India, South-East Asia and Europe. Model the demand side with that concentration explicit.
**Energy transition — a two-sided structural force.** Copper, aluminium, lithium, nickel and rare earths gain demand from electrification, grid build-out and storage; thermal coal faces terminal-demand and financing risk (hence its 3–4x multiple); blast-furnace steel faces decarbonisation capex it cannot fully pass through. Do not apply a single "commodities" narrative across the sector.
**Carbon and trade regulation.** EU CBAM phases in on steel, aluminium and cement from 2026, taxing the embedded carbon of imports and directly disadvantaging blast-furnace and coal-power-based producers selling into Europe. Scope 1 and 2 intensity per tonne is therefore becoming a cost variable, not an ESG footnote. Anti-dumping duties, safeguard measures and export taxes (including India's episodic export duties on iron ore and steel) can reprice a company's economics with weeks of notice.
**Scrap, recycling and secondary supply.** EAF and secondary-refining capacity grows as scrap pools mature, capping long-run prices and shifting the cost curve. A high-cost primary producer facing a growing secondary alternative has a structural, not cyclical, problem.
**Regulatory architecture (India).** The MMDR Act and its amendments moved India from discretionary allotment to auctions, creating a permanent economic split between legacy captive leases and post-2015 auctioned blocks. Forest and environmental clearance (Stage-I in principle, Stage-II final), state pollution board consents, evacuation infrastructure (railway sidings, conveyor corridors) and gram-sabha consent under FRA/PESA are all binding constraints on stated capacity. Treat announced capacity without clearances as an option, not an asset.
---
## India vs global notes
| Dimension | India (NSE/BSE, Ind AS) | US / global (10-K, GAAP/IFRS, EDGAR/ASX/TSX) |
|---|---|---|
| Reserve disclosure | UNFC / IMIC 2019; patchy, rarely audited, LoM plans usually not public. Reconstruct from lease documents, approved Mine Plans and IBM filings — or state that it is unavailable. | Mandatory and standardised: SK-1300 (US, since 2021), NI 43-101 (Canada), JORC (Australia), PERC (Europe). Technical reports are public, signed by a qualified/competent person, and are the primary input to NAV. |
| Government take | Royalty (iron ore 15% ad valorem, coal 14%), DMF (10–30% of royalty), NMET (2%), GST compensation cess on coal (Rs 400/t), plus auction premium on post-2015 blocks that can exceed 100% of IBM ASP. Can exceed 45–60% of gross revenue on an aggressively bid block. | State/provincial royalties, resource-rent taxes (e.g. Australia), free-carry and state participation in some jurisdictions. Typically 5–15% of revenue plus corporate tax. |
| Pricing conventions | Steel quoted in Rs/tonne on import parity; coal on FSA notified price plus e-auction premium; iron ore on IBM notified average sale price for royalty and on domestic/seaborne benchmarks for sales. Figures in crore/lakh — convert consistently and state the unit. | LME/COMEX/SHFE for base and precious metals; Platts/Argus 62% Fe CFR China for iron ore; Newcastle/API indices for coal; USD per tonne, pound or ounce. |
| Currency | Most commodity prices are dollar-linked while a large part of the cost base is rupee-denominated, so INR depreciation is usually EBITDA-accretive for exporters and import-competing producers — but it inflates USD-denominated debt. Model both legs. | Local-currency cost base against USD revenue is the same structure in Australia, Brazil, South Africa and Chile — the AUD/BRL/ZAR/CLP effect is a standard part of cost-curve analysis. |
| Ownership and governance | Promoter and government holding matter: check pledge levels, related-party off-take and marketing arrangements, brand/management fees to unlisted parents, and minority protection in listed subsidiaries. PSU miners carry administered pricing, disinvestment overhang and social obligations. | Widely held; the governance issues are different — dual-class structures, state golden shares in some jurisdictions, and controlling-family stakes in a few majors. |
| Disclosure venues | Concall transcripts and quarterly investor presentations carry volume, realisation, cost and integration detail that never appears in the financials. Annual report notes carry royalty, DMF, stripping policy and closure provisions. CARO reporting flags statutory dues and related-party issues. | 10-K MD&A and segment notes, quarterly production reports, technical reports on SEDAR/ASX/EDGAR, and reserve tables under SK-1300. Sell-side publishes explicit commodity price decks. |
| Valuation practice | Mid-cycle EV/EBITDA, EBITDA/t × capacity, EV per tonne of capacity against replacement cost, SOTP for conglomerates. P/NAV used for only a handful of names. | P/NAV leads, especially in Canada/Australia; EV/EBITDA on consensus deck as cross-check; EV per reserve unit for developers. |
| Accounting detail | Ind AS mirrors IFRS, including IFRIC 20 on stripping and IFRS 16 leases. Watch for capitalisation policy differences on exploration and on pre-production revenue. | IFRS filers similar. **US GAAP differs**: no IFRIC 20 equivalent (US practice generally expenses post-production stripping), different impairment reversal rules (IFRS permits reversal, US GAAP does not), and successful-efforts vs full-cost choices. Do not compare capitalised-stripping-adjusted cost between a US GAAP and an IFRS filer without adjusting. |
---
## Checklist
- [ ] State where you believe the commodity is in its cycle, and on what evidence — do not quote a single ratio before doing this.
- [ ] Suppress trailing P/E, or report it only alongside mid-cycle normalised EPS; never call a 4x peak-earnings P/E cheap.
- [ ] Replace OPM with unit cash cost (C1 / AISC) and the company's **quartile on the global cost curve**.
- [ ] Compute EBITDA per tonne/ounce by product and compare to the company's own 10-year mean and to peers at the same realised price.
- [ ] Decompose revenue growth into price and **saleable volume**; report volume growth separately.
- [ ] Check reserve life (R/P), 3-year reserve replacement ratio, and whether reserves were added by drilling or by assumption changes.
- [ ] Compare mined grade to reserve grade and current strip ratio to LoM average across several periods — this is the high-grading test.
- [ ] Split capex into sustaining vs growth; check sustaining capex against DD&A and against capitalised stripping.
- [ ] Compute all-in cash break-even price and compare with spot and with the 90th-percentile marginal industry cost.
- [ ] Test leverage against **trough** EBITDA, not trailing; include leases, hybrids, ARO and (India) auction-premium/DMF commitments in net debt.
- [ ] Map the debt maturity profile against the break-even price — flag any wall inside 24 months.
- [ ] Quantify integration (captive ore, coal, bauxite, power) and verify it is long-dated and cleared, not an MoU.
- [ ] Compute total government take as % of revenue; for Indian post-2015 blocks, pull the auction premium % on every block.
- [ ] Check realised price versus benchmark and explain any widening discount.
- [ ] Compute FCF yield at spot **and** at mid-cycle; build the EBITDA sensitivity per unit price and per unit FX move.
- [ ] Compute mid-cycle ROCE and compare with the ~8–10% real WACC used for base-metal projects.
- [ ] Review closure/rehab provisions (discount rate, escalation, funding) and the tailings inventory with GISTM status.
- [ ] Check by-product credit dependence and whether cost is presented net or on a co-product basis.
- [ ] Value on mid-cycle EV/EBITDA and, where technical reports exist, a finite life-of-mine NAV with **no perpetual terminal value**.
- [ ] Cross-check against EV per tonne of installed capacity versus greenfield/brownfield replacement cost.
- [ ] Apply SOTP with an explicit holding-company discount for diversified or holding structures.
- [ ] Build the peer set on cost quartile, commodity, integration and jurisdiction — never on the "metals" label alone.
- [ ] Scan for peak-cycle M&A or greenfield sanction, definitional changes to cost metrics, and serial "exceptionals".
- [ ] India: check promoter pledge, related-party off-take/marketing, lease expiry dates, and clearance status (Stage-I vs Stage-II).
- [ ] Present the conclusion as a bear/mid/bull range across a price deck, not as a single target price.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |