Methodology

Overview

Every company page carries a set of scores on a 10-point scale, each worked out from researched figures, never typed in by hand:

• Management — how deep the current team's documented track record is.
• Financials — balance-sheet strength, judged differently for producers and for explorers.
• Deposit Quality — the size, grade, mine life and cost of the deposit. Royalty and streaming companies, which own no deposit, get a Royalty Portfolio Score instead.
• Location & Infrastructure — the jurisdiction, what is already built at the project, and whether a producing mine is nearby. Jurisdiction is not a separate card: it is the largest part of this score.
• Overall — one number combining the four above.
• Acquisition Susceptibility — how closely a company matches the profile of juniors that have been bought. It is kept out of the Overall score.

These scores are separate from Community Rankings, which are opinions submitted by JuniorMetrics users, not something we compute.

Unresearched inputs count at a neutral 5. Most of these scores blend several inputs, and for many companies some of them haven't been researched yet. Each unresearched input counts at the middle of the scale, 5/10, in proportion to its weight: a score whose researched inputs average 9 but cover only 40% of its weight publishes as 0.4 × 9 + 0.6 × 5 = 6.6. A thinly researched score therefore sits near 5, and a fully researched one is its plain weighted average. We used to leave unresearched inputs out and average the rest, which let the least-researched companies rank highest on whichever inputs happened to be easiest to score. An input that can never apply to a company (grade for copper, which has no calibrated grade curve, or Deposit Quality for a royalty company that owns no deposit) is left out rather than counted at 5. Where an input is missing, a hatched band on the score shows the range it could land in once researched, and the published number sits in the middle of it.

None of these scores are investment advice — they're a structured starting point for your own research. See the legal disclaimer for the full picture.

One number worth keeping in view while reading any of this: junior mining as a sector has a rough, oft-cited base rate of roughly 1–2 successes for every 10 companies that set out to become one — the large majority stall out, dilute existing shareholders into irrelevance, or go to zero somewhere along the way. That figure comes from industry commentary we reviewed, not from JuniorMetrics' own tracked-company data, so treat it as sector-level context for calibrating expectations, not a statistic we're claiming to have derived or verified ourselves.

Jurisdiction Risks

A 1–10 score for the political and regulatory stability of the place the company's flagship project sits in. Higher is safer. In the United States, Canada and Australia it follows the state, province or territory on the stated basis below. Everywhere else it is a researched judgement of the country, kept consistent with the other companies in the same country, weighing rule of law, permitting speed and predictability, indigenous/community consultation requirements, expropriation and resource-nationalism risk, and general political stability. It does not change with short-term news; it's meant to reflect the operating environment a company is stuck with for the life of the mine.

On each company page this score appears as the 40% jurisdiction factor inside Location & Infrastructure rather than as a card of its own.

US states, Canadian provinces, Australian states and Argentine provinces

A company whose flagship is in the United States, Canada or Australia is scored on one stated basis: the Policy Perception Index (PPI) from the Fraser Institute Annual Survey of Mining Companies, 2025 (published February 26, 2026; Table 2). The PPI rates each jurisdiction's mining policy climate from 0 to 100, from a survey of mining executives on regulatory uncertainty, environmental rules, taxation, land claims and protected areas, infrastructure, labour, political stability and security. A jurisdiction the 2025 survey dropped for too few responses (California, Michigan, Minnesota, Washington, Wyoming) uses the 2024 survey; a state Fraser does not rate (Oregon, South Carolina, Texas and others) stays on the country basis.

The mapping. A PPI of 70 or more scores 9. Below 70, the score drops one point for every 15 PPI points or part of them: 55–70 scores 8, 40–55 scores 7, 25–40 scores 6, 10–25 scores 5 and below 10 scores 4. It never gives 10, because the PPI is relative (the best jurisdiction in each year's survey scores 100), so its top means best of that year's field, not free of risk, and the leaders change from year to year. The mapping is calibrated to the scores this site had already given: every US, Canadian and Australian jurisdiction we scored 9 has a PPI of 73 or more, the four we already scored 8 for jurisdiction reasons (Yukon, Nunavut, Manitoba, Montana) sit between 62 and 70, and across 27 other countries our country scores fall about one point for every 15 PPI points. Outside these countries the scores stay country-level judgements, because a policy survey of mining managers only partly captures sovereign, security and currency risk, and many countries are rated on a handful of responses.

Argentina. Fraser rates Argentina by province (San Juan, Santa Cruz, Neuquén, Mendoza and Chubut in 2025; La Rioja in 2024). A province's PPI measures its own mining policy, not the national currency and capital controls, export duties and macro instability every Argentine project shares. So an Argentine province scores the same mapping less 3 points for that national risk: San Juan and Santa Cruz score 6, the level this site already gave Argentina, and Neuquén, La Rioja, Mendoza and Chubut follow Fraser's ranking below them. Salta, Jujuy, Catamarca and Río Negro had too few survey responses to be rated, so companies there stay on the country basis.

Project-specific adjustments. A company keeps a different score only for a fact about its own project or district, stated in its caption: Pebble in Alaska scores 4 because of the EPA's veto of the project, Westmoreland in Queensland scores 5 because Queensland does not allow uranium mining, and companies in Alaska's Ambler Mining District score 8 because the district's only planned access, the Ambler Road, was rejected by the federal government in 2024, approved again in October 2025, and is not yet built.

Management

For each current management team member, we track their prior company roles and tag each one SUCCESS, FAILURE, ONGOING, or NEUTRAL based on how that company actually performed (e.g. a mine reaching production and returning capital is a success; a company that went to zero or was delisted is a failure).

The Management score measures the depth of that documented track record, and is deliberately not a success rate. The reason is a property of the underlying data rather than a design preference: recorded outcomes skew overwhelmingly positive, because failures rarely make it into an executive bio in the first place. A ratio computed on that data would mostly be measuring what gets written into bios, and it collapses on small samples — a single documented setback next to no other evidence would sink a team's score, while a team with one researched win and nothing else would score perfectly.

So the score starts from a neutral base and adds credit for three separate things: breadth (how many of the current team have at least one documented prior win), depth (how many documented wins the team has in total), and quality (whether any of those wins were hard to achieve — a major discovery or a project taken to production counts for much more than a corporate milestone a rising market can hand to a mediocre team). A documented setback subtracts, but scaled by how much documented success sits alongside it: one failure next to eight wins is a very different fact from one failure next to nothing. Read a low score as "we have researched little about this team", not as "this team is weak" — which is why a score resting on two or fewer researched outcomes carries a note saying exactly that.

A team with no decided track record at all (e.g. all first-time executives, or roles still ongoing) shows — rather than a score — we don't invent a number when there's nothing to measure.

That track-record score is then reduced by a concurrent-role overextension penalty when a team member is spread thin across multiple companies at once — a real vetting concern, not just a display stat. Executive seats (CEO, CFO, COO, President) get a narrow allowance of 2 companies before the penalty starts, since those roles imply a full-time commitment; director/board seats get a wider allowance of 3, since serving on multiple boards concurrently is normal, often positive, practice. Each company over the allowance costs 0.3 points, capped at 2.0 points total.

Financials

The Financials score is calculated differently depending on stage, because a producer and a pre-revenue explorer should not be judged by the same yardstick:

• Producers & construction-stage companies — scored on net cash (cash minus debt) as a percentage of market cap. Revenue is funding operations at this stage, so balance-sheet strength relative to market cap matters more than raw treasury size.
• Everyone else (explorers/developers) — two inputs, weighted equally. Net cash (cash minus debt) as a share of market cap, on a smooth curve: zero net cash scores 2, 10% scores about 5, 20% about 7, and more keeps scoring higher with no ceiling, while net debt scores below 2. Runway: months of cash at the recent burn rate (cash ÷ quarterly burn × 3), where 12 months scores about 5, 24 months about 7.4, and a company that isn't burning cash scores 10. Burn comes from trailing free cash flow (operating cash flow minus capital spending) where we have it, otherwise from operating expenses for the latest reported period, a rougher accounting proxy. Most explorers don't have a burn figure researched yet, so their runway counts at a neutral 5 (see above) and the card shows a range band until it is.

This replaced an explorer formula that read only cash as a share of market cap, ignored debt, and maxed out at 20%: a third of explorers scored 9–10, including companies with more debt than cash. A cash or debt figure we don't have is never counted as zero. With no cash figure there is no Financials score. With cash but no debt figure, net cash can't be computed, so it counts like any other missing input: at a neutral 5, except that it never counts higher than the cash alone would score, because debt can only lower net cash. Runway needs only cash and burn, so it still counts. With neither a debt figure nor a burn figure, there is no Financials score. A debt figure of zero that we have read from the company's accounts is a real figure and scores as zero debt.

Deposit Quality

How big, how rich, how long-lived and how cheap to mine a company's deposit is. It blends four inputs: deposit size (35%), grade (25%), mine life (20%) and all-in sustaining cost (AISC, 20%). Each is scored from 0 to 10 on its own scale, described below. An input that hasn't been researched yet counts at a neutral 5 until it is (see Unresearched inputs). An input with no calibrated scale for the company's metal is left out and the other inputs re-weighted. Neither is ever counted as zero.

Royalty and streaming companies and investment or holding companies own no deposit, so this score isn't shown for them. Royalty companies get the Royalty Portfolio Score instead.

What the deposit-size figure counts

• Contained metal, never an equivalent. Some companies report their resource as metal-equivalent ounces (gold-equivalent, silver-equivalent, copper-equivalent and so on). We record only the contained amount of the company's primary metal. An equivalent total converts the by-products into that metal at the company's own price assumptions, so it changes whenever those prices change, and two companies' totals can't be compared unless they used the same prices. The equivalent figure is kept in the notes; it is never the scored number.
• By-products stay out. A metal counts when it carries real weight in the deposit: it is the most valuable metal there, or roughly a third of the deposit's value or more. Gold or silver that is a by-product credit in a copper deposit, at roughly a quarter of the value or less, is not counted. In between, we follow how the company itself frames the deposit.
• Measured and Indicated only. Inferred resources are excluded from the scored figure for every company, and Reserves are never used in its place (both explained below).
• Company-wide. The figure is summed across every project the company holds an interest in, unlike mine life, grade and AISC, which come from its most advanced project (see Flagship vs. Secondary Assets).

Whose deposits count

• Owned outright or through a joint venture: counted. For a company with one asset we show the whole deposit and record its ownership percentage separately; P/NAV uses that percentage. A company holding several assets at different ownership levels is shown at its attributable total. A government's carried interest, common in West Africa, reduces the company's share the same way a joint-venture partner's does.
• Held under a mining lease: counted. A lease is a current right to mine, even when the company also holds an option to buy the ground.
• Held under an option or earn-in: not counted yet. While the option still requires spending, a study or other work, the deposit counts only once the option is exercised or the interest is earned. The exception is an option the company can exercise whenever it chooses, for a fixed price with no work left to do, when the company's own regulatory resource statement already reports the resource as its own.
• A royalty, net smelter return or profit share without title: not counted. That holds even when the company operates the mine. A royalty holder doesn't count the deposits its royalties sit on either.

Why Inferred is left out. Inferred is a real, compliant, disclosed category, but it is the lowest-confidence one and can't be converted to reserves. Blending it into the headline figure would reward a lightly drilled deposit at the same weight as a fully delineated one, and deposit size is the heaviest input to this score. Where a company has a material Inferred resource we still show it, broken out by category directly beneath the headline figure on its page and marked "not scored". A company whose only resource is Inferred shows no figure at all rather than a zero.

Why not Reserves. Reserves are the narrower, economically mineable part of the Measured and Indicated resource, so using them would understate deposit size for exactly the companies furthest along. Producers, including the largest, are held to the same resource standard as everyone else.

Deposit-size scales

Deposit size is scored on a log scale, because deposits span several orders of magnitude, with a separate scale for each metal: the same number of ounces or tonnes means very different things in silver and in gold. The table gives the size that scores 0, 5 and 10 on each scale, and how many tracked companies each one scores today.

Equal value, equal score. Every scale except nickel's is set so that a deposit scores about the same as a gold deposit worth the same in the ground. Its midpoint, the 5/10 column, is the amount of metal worth roughly what 1 million oz of gold is worth, and it spans the same range of value as the gold scale, so the scores stay within about a quarter of a point of each other from a US$250 million deposit to a US$50 billion one. Rare-earth oxide is valued through its neodymium-praseodymium content (a 22% share at about US$97 per kg, roughly US$21,000 per tonne of oxide), and a tonne is scored by mass even though heavy-rare-earth baskets are worth more per tonne. Lithium's price moves far more than gold's, so its match is only as current as that price. Nickel's scale hasn't been matched yet.

Why the ceilings sit so high. Gold's 10/10 is set above the largest gold resource base any company has reported (Newmont's 233.5 Moz at year-end 2024), so a 10 means bigger than any gold company has ever assembled, not merely large. Because the figure is company-wide, the biggest companies can exceed any single orebody (the largest, Muruntau in Uzbekistan, is about 170 Moz). The silver, zinc and rare-earth ceilings sit far above any real deposit of those metals: those deposits are worth far less in the ground than the largest gold and copper ones, and a ceiling just above them would undo the value matching.

Where a scale is thin or missing. Read a scale that rests on a handful of companies (the right-hand column) as provisional. Platinum-group-metal ounces are scored on the gold scale: too few companies report them, on different metal baskets, to build a scale of their own, and a PGM ounce is worth well below a gold ounce, so their deposit-size scores run high. Tin, vanadium, manganese, tungsten, iron ore, magnesium, mineral sands, cobalt, molybdenum, niobium (as Nb₂O₅) and titanium (as TiO₂) each have too few tracked companies to calibrate a scale, so deposit size is left out of their score rather than measured on a borrowed one. Bauxite is shown as tonnes of bauxite ore at its alumina grade, the way bauxite resources are reported, not as contained metal, and is not scored either. Diamond resources are shown in carats and not scored, because a carat of one deposit's diamonds can be worth many times a carat of another's. A resource reported in tonnes of copper or zinc, or pounds of nickel, is converted first, so the unit a company reports in never changes its score. When a resource with no size scale leaves too few inputs for a score, the Deposit Quality card says “Not scored (no size scale for manganese)”, or carats, titanium and so on, and shows the figure on file: the resource has been researched, it just can't be ranked.

Grade, mine life and cost

• Grade uses a log scale for gold (0.01 oz/t, about 0.3 g/t, scores 0; 0.5 oz/t, about 15.5 g/t, scores 10) and for silver (1 oz/t to 15 oz/t). No other metal has a calibrated grade scale, so grade is left out for them.
• Mine life uses one log scale for every metal, from 1 year (0) to 120 years (10). 120 years is the continuous operating history of Bingham Canyon, among the longest-lived mines anywhere, and longer than any mine life we track. On a log scale, going from 5 to 10 years earns as much as going from 40 to 80, because years far in the future add less value.
• AISC uses a straight-line scale for each metal, from the cost that scores 10 to the cost that scores 0 (below), and is compared like with like. Companies report it on different bases: a by-product AISC is per ounce or pound of the main metal, with the other metals' revenue credited against costs; a metal-equivalent AISC (AgEq, AuEq, CuEq) converts the other metals into the main one; a co-product AISC splits the costs across the metals. For the same mine a by-product figure is usually much lower, so each figure is labelled with its basis and scored on a scale for that basis. A negative AISC, where by-product credits exceed costs, scores 10. Where there is no scale for a metal or a basis, the AISC is shown as reported and not scored.

Mining method (open pit or underground, strip ratio) isn't tracked yet, so it isn't an input to this score.

Coverage is genuinely thin: about 47% of the companies we track have enough researched inputs to get a score here at all, and for roughly 62% of those, only one or two of the four inputs are in.

The minimum. A single researched input isn't a meaningful version of this score: it would be a different measure wearing its name. So the score is withheld (“Not enough info”) unless the researched inputs carry at least 35% of the blend between them, which is the weight of deposit size alone. A contained-metal figure can support a score on its own; a mine life, a grade or an AISC on its own cannot.

The range band. Above that minimum, whenever an input is still unresearched the gauge shows a hatched band: the range the score could land in once the rest is researched. Its low end assumes every missing input comes in at the bottom of its scale, its high end at the top, and the published score sits in the middle. A company with a contained-metal figure and a mine life but no grade or AISC, whose two researched inputs average 4.8, publishes 4.9 with a band of 2.6–7.1. A caption such as “2 of 4 inputs researched” names what is in and what is missing. An input we researched and found nothing for counts as researched there, with that number in brackets (“5 of 6 inputs researched (4 with nothing found)”); it still counts at the neutral 5 and stays inside the band until a figure turns up. An input that can never count for a metal (grade for copper, for example) isn't treated as missing and doesn't widen the band. The band is a different mark from the red flag described at the end of this page: the flag means a figure was assumed, while the band means every figure used is real and part of the picture simply hasn't been researched. Below 60% of the blend's weight, the caption also says to read the number as indicative rather than settled.

Location & Infrastructure

How easy a company's flagship project is to build and run where it sits. It answers two questions: may you build it there, and what is already built there? It blends three factors, each shown as its own row under the score on every company page:

• Jurisdiction (40%) — the jurisdiction score described above: permitting predictability, rule of law, and mining-friendliness. It is known for every company, so this score always has a value.
• Infrastructure (40%) — how much of the standard checklist is confirmed built at the flagship project: an existing mill, road access, and grid power count evenly; a tailings facility counts at half weight. Each item scores 10 if confirmed built, 0 if confirmed absent, and a neutral 5 until it is confirmed either way. Each real, sunk piece of infrastructure is capex a company no longer has to raise or build.
• Nearby producing mine (20%) — a producing mine within 50 km, which usually means shared roads, power, mills, and labor. Within 20 km (trucking distance to someone else's mill) scores 10, tapering to 8 at 50 km.

Anything not yet researched counts at a neutral 5/10 (see above) — never as zero, and never shown as “N/A”. The infrastructure checklist is researched for projects at every stage: whether a grassroots project already has a road and grid power is exactly what separates a cheap build from an expensive one. A score that rests on jurisdiction alone carries a range band and an “N of M inputs researched” caption, so it reads as provisional until the rest comes in. Today, of the 1,176 operating companies we track, 777 (66%) have their flagship's infrastructure researched, 221 (19%) have a listed producing mine within 50 km counted in their score (another 12 have one that isn't counted because it would lower the score; see below), and 360 (31%) rest on jurisdiction alone so far. Another 367 have had their flagship's checklist researched with nothing found: no source we could cite confirms or rules out any of the four items. Those read “No verified figure” with the date of the check and a recheck due 180 days later, and count at the neutral 5 like an unchecked list. It is a dead end for now, not a gap we haven't looked at.

A nearby mine can only help. The list of producing mines we measure against (545 mines, each with cited coordinates, built from official national lists of producing mines where they exist) still covers some countries better than others. A project in a well-listed district finds its neighbor; a project in an equally busy but less well-listed one would look remote when it isn't. So distance only ever counts in a company's favor:

• No listed producer within 50 km reads “Not yet researched” and counts at the neutral 5, never scored as remote.
• A listed producer within 50 km is counted only if it raises (or matches) the score the company's other researched inputs already give it. If it would lower it, the page still shows the mine, its distance and what it would have scored, marked “Not counted: would not raise the score”. For example, a fully built project in a 9/10 jurisdiction already scores 9.5; a producer 45 km away would score 8.3/10 on its own, so it is shown but not counted, and the score stays 9.5.
• A company's own mines are skipped, so a producer isn't credited for being next to itself; its own mill already counts in the checklist.

Confirmed-absent facts count too: a checklist item recorded as not built pulls the score down just as a built one pulls it up. Grid power specifically means a confirmed connection to a national grid — a “no” there can mean an already-built diesel or HFO power station rather than a real power gap, a known coarseness we'd rather disclose than hide. An item nobody has confirmed either way counts at the neutral 5, never as built: a project with confirmed road access and grid power, and nothing known about a mill or tailings, scores 7.9/10 on the checklist, not 10/10. So one or two confirmed facts can never stand in for the whole checklist, and the checklist figure on a company page is the one the score actually uses.

Like mine life, grade, and AISC above, the checklist reflects only the company's flagship (most-advanced) project, never a blend across every project it holds — see Flagship vs. Secondary Assets under Models below. A secondary project's own infrastructure (shown on its own project tab on the company page) is never averaged in, in either direction. For a royalty or streaming company, which holds interests in other operators' mines rather than building its own, and for an investment or holding company, which owns no mine at all, the checklist and the nearby-mine factor read “Not scored”: its Location & Infrastructure score is its jurisdiction score.

This score is one of the four inputs to the Overall Score below, in the place jurisdiction used to hold on its own. The peer median shown beside it is taken across every company in the same commodity.

Royalty Portfolio Score

A royalty or streaming company owns no deposit and files no technical report of its own — it holds a book of interests in OTHER companies' mines instead, so Deposit Quality above structurally can't apply to it. This score is the royalty/ streaming substitute, shown only for those companies: a weighted blend of portfolio scale — attributable gold-equivalent ounces per year (35%), balance-sheet net-cash ratio (20%, the same formula Financials uses for producers above), undrawn credit-facility firepower relative to market cap (15%), dividend yield computed live off the current share price (15%), and the producing share of tracked producing/development/ exploration assets (15%).

Portfolio scale and the credit-facility ratio both use a log scale, the same reasoning as Deposit Quality's size curve: both span orders of magnitude across the tracked royalty universe. Balance-sheet strength is not a separate formula — it reuses Financials' own producer branch directly, since a royalty company is revenue-generating and economically closest to that branch.

Each of the five inputs is researched independently and, like Deposit Quality, an unresearched input counts at a neutral 5/10 rather than zero. This score carries the same floor as Deposit Quality and for the same reason: we withhold it unless the researched inputs are worth at least 35% of the blend — the weight portfolio scale alone carries — rather than re-weighting one input to 100% and printing a precise-looking number. P/NAV and implied cash margin (spot price minus disclosed cash cost per GEO) are the two most obvious next candidates for a future v2 component, but neither has enough tracked-company coverage today to build on.

Overall Score

An equal-weight average of Location & Infrastructure (which includes jurisdiction), Management, Financials, and Deposit Quality. Location & Infrastructure always has a value; any of the other three not established yet counts at a neutral 5/10 rather than being left out (see above). Leaving it out used to rank the least-researched companies highest: an explorer with no resource yet was averaged on location, management and cash alone, the three easiest inputs to score well on, and grassroots explorers filled most of the top 50. Each input enters at its own published value, which already reflects its own unresearched share, so nothing is pulled toward 5 twice. A royalty, streaming or investment company owns no deposit, so Deposit Quality is left out for it and the other three are averaged.

If Management, Financials, and Deposit Quality are all missing, we don't publish an Overall score at all — it reads “Not yet researched”. Location & Infrastructure on its own is not an overall assessment of a company, and printing a single input as a confident-looking “Overall” would imply we know more about that company than we do.

Where at least one of the other three inputs is available, we do publish the average, but flag it as provisional and say which input(s) are missing. The same flag appears when an input we did use is itself provisional, and, separately, when an input we used is entirely real but rests on a minority of its own sub-inputs, in which case we say that rather than describing it as assumed, because it isn't. A clean-looking average shouldn't hide any of those three kinds of gap underneath it.

On the Financials side specifically, a producer's score is built from cash and debt together as a single net-cash figure, and nothing else. If we don't have a figure for either one, net cash can't be computed, so we don't publish a Financials score for that producer (until October 2026 the missing figure was assumed to be zero and the score flagged as provisional). Counting it at a neutral middle instead would give every such company the same number regardless of its actual finances.

Acquisition Susceptibility

A relative ranking of how closely a company matches the profile of juniors that have actually been bought, not a probability. Only a few dozen significant deals happen a year across thousands of listed juniors, so even a high score describes a long shot. It is deliberately kept out of the Overall Score.

It blends six inputs, weighted by how well our acquisition research evidences each one: a disclosed pending acquisition or merger (26.25%), an operating producer's existing strategic stake in the company (25%), distance from the company's project to the nearest producing mine (22.5%), lifecycle stage (15%), jurisdiction (6.75%), and primary commodity (4.5%). Any input that doesn't apply to a company is left out and the rest are re-weighted, rather than being counted as zero. A company with project coordinates on file but no producer within 200 km does earn a real zero for distance. Each company's page lists all six inputs with what we found, the sub-score and the weight. When an input hasn't been researched yet (most often distance, because no project coordinates are on file), a hatched band on the gauge shows where the score could land once it is. A pending deal or a producer's stake can only raise the score when found, so the bottom of that band assumes neither turns up.

The producer's stake. The strongest predictive signal in the research is a producer that already owns part of the company, like Agnico Eagle's 13.9% of Rupert Resources before it bid for the rest, or Gold Fields' 50% joint venture on Osisko's Windfall project before it bought Osisko outright. We read this from the strategic shareholders on each company's page, which are sourced from the company's own disclosures. Only operating producers count, so funds, individual investors, royalty companies and other juniors do not. A producer holding 5% scores 8/10, rising to 10/10 at 20%, the take-over-bid threshold in Canada and Australia. A holder that size is the natural acquirer, and because a scheme of arrangement needs two-thirds to three-quarters of the votes cast, a block that large can in practice veto a rival deal. A stake under 5% is shown but not scored. So is a company that is already 50%+ controlled, because its sale is then its owner's decision. A stake is also never allowed to lower a score: when a company's other inputs already average higher than the stake's own score, the stake is left out.

What a missing stake means. Having no producer on the register is not held against a company, even when we've researched its register. Most juniors that do get bought had no prior stake from their buyer, so its absence says little. Two things the research also flags are not captured yet. A producer that is a joint-venture or earn-in partner on the company's project, rather than a shareholder, isn't recorded in a structured way. And a fresh strategic investment often comes with a standstill that bars a bid for a period, which we don't track, so each stake is shown with its report date.

Community Rankings

Separately from the calculated scores above, logged-in users can submit their own 1–10 rankings per category for any miner. The numbers shown on the Community page and in the compare table's community columns are simple averages of user submissions, with the submission count shown alongside so you can judge how much signal is behind a given average. These reflect user opinion, not JuniorMetrics research.

How We Assign a Stage

Every company sits at one of eight stages. A stage is a claim we check, not a label we copy: it is the furthest milestone actually reached, with evidence, on the company's most advanced project. It is never a blend across projects, and it can go no higher than the most advanced study the company has actually published.

1. Grassroots Exploration — Early-stage prospecting on unproven ground — staking claims, soil/rock sampling, and initial geophysics to identify targets worth drilling.
2. Resource Definition — Drilling has confirmed mineralization; the company is defining the size and grade of the deposit toward a formal, NI 43-101-style resource estimate.
3. Preliminary Economic Assessment — A first-pass, lower-confidence study of whether the deposit could be economic to mine, used to justify further spending.
4. Pre-Feasibility Study — A more rigorous economic and engineering study narrowing down mine design, costs, and returns ahead of a full feasibility study.
5. Feasibility Study — Bankable-grade engineering and economics detailed enough to support a construction financing decision.
6. Permitting — Seeking the environmental and regulatory approvals required before construction can legally begin.
7. Construction — Building the mine — earthworks, processing plant, and infrastructure — ahead of first production.
8. Production — Generating real, audited sales revenue from its own operations — not just first metal or stockpiled output — no formal commercial-production declaration required. A producer whose operations are all idle with no sales for 12 months reverts to its most advanced active project's stage.

What counts as production. A company is a producer when it has real sales from its own operations: audited revenue, or its own account of a completed sale. A first pour, a formal “commercial production” declaration, an offtake agreement or a prepayment is not a sale on its own, and neither is output the company says is unsold or still being commissioned.

Idle producers. A producer whose operations are all paused or on care and maintenance, with no sales in the 12 months since its last sale, moves back to the stage of its most advanced active project, with a note saying when operations went idle. Under 12 months, it stays a producer with that note. A company with any mine still selling stays a producer, and selling inventory left over from before the pause doesn't restart the clock. It returns to production when sales resume.

The milestone under a company's name. The dated event shown beside the stage at the top of each company page is the latest milestone we have on record, cited to its source. On a producer, an event more than two years old isn't shown there, since the stage already says the company is producing. On any other company it stays, with its date, and says when nothing newer has been recorded.

The Lassonde Curve

A stylized model of how the market tends to value a junior mining company as it moves through its project's life, named after Pierre Lassonde, co-founder of Franco-Nevada. It doesn't measure anything precisely — it's a shape, not a formula — but it's one of the most widely cited mental models in resource investing for explaining why a technically exciting project can still trade sideways or lower for years.

The curve typically shows an early run-up as drilling turns a speculative idea into a real discovery ("discovery euphoria"), followed by a derating trough once that blue-sky optionality is priced in and the market instead has to sit through the long, capital-intensive, dilution-heavy grind of economic studies and permitting. Valuation tends to re-rate upward again once a construction decision meaningfully de-risks the project, continuing into production. Reading it well means recognizing that a company's stage — not just its news flow — shapes how the market is likely pricing it right now, without treating that pattern as a guarantee for any individual stock.

We've overlaid an illustrative version of the curve on our own 8-stage model on the Lifecycle Roadmap page, so you can see roughly where the classic derating trough and re-rating climb line up against the same stages used everywhere else on this site — including the Pre-Production Sweet Spot described next.

Pre-Production Sweet Spot

Independent analyst Lobo Tiggre's research (also called the "Golden Runway") into the window between a company's formal construction decision — positive feasibility study, permits secured, financing in place — and first production. It's widely cited as one of the highest-probability, best risk-adjusted-return windows on the Lassonde Curve: most of the geological, financing, and permitting risk is already behind the company, but the stock has often still had significant re-rating left to deliver.

In Tiggre's own published studies of 100+ first-time mine builders, roughly 90–95% of companies that made a formal construction decision went on to reach production, with average gains from that decision to first production historically landing well over 100% (rising further for those held through commercial production). Average time from decision to first pour has run around a year and a half. Past performance of the group as a whole doesn't guarantee any individual company succeeds — a favorable stage in the cycle doesn't override red flags in management track record, financing structure, or jurisdiction risk covered elsewhere on this page.

On JuniorMetrics, this window is marked on the Lifecycle Roadmap and on each miner's Lifecycle Milestones. In our current 8-stage model, that maps to the Construction stage specifically — by definition, a company there has already locked in a positive feasibility study, permits, and financing. Permitting-stage companies are marked as approaching it. We deliberately don't extend the window into Production — we don't yet track how long a company has been producing, and including long-mature producers would misrepresent what the window actually means.

Prospect Generator Model

A business model where a company's main product is geological ideas rather than a single flagship project. A prospect generator uses its own technical team to stake or acquire a portfolio of exploration properties, then options or joint-ventures individual prospects out to major and mid-tier partners, who fund the drilling in exchange for an earn-in stake. The generator typically keeps a minority interest, back-in right, or royalty rather than paying for the drill program itself — spreading exploration risk across several properties and funding most of it with partners' capital instead of its own treasury.

Resource investor Rick Rule is one of the most prominent public advocates of this approach, arguing that it lets a junior generate several "at bats" a year while a partner's own due diligence and balance sheet absorb most of the funding risk, and that investors are generally better served by a basket of several prospect generators than a bet on any single one. We haven't fabricated quotes or figures from his commentary here — see his own framing directly:

• Rick Rule: The Prospect Generator Model — Investing News Network
• Investigating Prospect Generators with Rick Rule — Investing News Network

On JuniorMetrics, this is a real, structured classification rather than something you have to infer from a company's Documents and project description: every miner carries a business-model tag, and a company actually run this way is flagged Prospect Generator (or Hybrid PG for one that also directly funds and operates part of its own project pipeline rather than JV'ing out its entire portfolio — the same hybrid distinction Rick Rule and Jeff Phillips draw), shown as a badge and in Compare and head-to-head views. Its own page adds a dedicated Prospect Generation & Joint Ventures panel alongside its normal operator sections — showing every named JV project's partner, partner tier, retained interest or royalty, and whether partner-funded exploration spending is currently active — and the full roster lives on its own Prospect Generators directory rather than the main miners list.

The Graham & Dodd Method

Benjamin Graham and David Dodd's value-investing framework, laid out in their 1934 text Security Analysis. Its core idea: a security has an intrinsic value grounded in analyzable facts — assets, earnings power, financial strength — that can diverge from its market price. Buying only when price sits meaningfully below that estimated intrinsic value leaves a margin of safety that protects against bad luck or errors in the analysis itself.

Junior miners rarely have the earnings history Graham and Dodd built their original method around, so applying it here means substituting proxies. A technical-report-derived NAV — the after-tax NPV from a PEA, PFS, or Feasibility Study — stands in for intrinsic value, and P/NAV (price divided by that NAV) becomes a margin-of-safety-style ratio: the lower the multiple, the bigger the gap between what the market is charging and what the study says the project is worth, all else equal. This is exactly why each stage's Lifecycle Roadmap detail page shows a P/NAV band alongside "how this phase gets valued" — it's the same lens, tied to our real stage data rather than a company-by-company guess. Balance sheet strength matters too, which is the same reasoning behind the Financials score above — a company with a thin margin of safety on cash is a company one bad financing away from diluting that NAV per share down significantly.

None of this replaces judgment — a low P/NAV can reflect a genuine bargain or a genuinely troubled project (bad jurisdiction, weak management, an unfinanceable capex bill), which is exactly what the Location & Infrastructure, Management and Financials scores elsewhere on this page are for.

One thing worth knowing about how this figure is built: a technical report's NPV is the project's NPV, not the company's share of it. Dividing market cap straight into that number would silently assume the company owns the whole thing — for a 25%-interest joint venture partner, that understates P/NAV by 4x. So P/NAV here divides market cap by the company's attributable share of NAV, using a researched project-ownership percentage where we have one.

The honest caveat is now about coverage rather than arithmetic: that ownership percentage is only populated where a real, disclosed partial interest was found, and a company with nothing recorded is treated as owning 100% — which is correct for the large majority of miners, since they are single-asset and wholly-owned, but would understate P/NAV for an undisclosed or not-yet-researched JV. If a company's own filings describe a joint venture and its ownership percentage isn't shown on its page, treat that P/NAV as a lower bound.

EV/NPV vs. P/NAV

EV/NPV is what the market pays for the whole company, net of cash and debt, per dollar of the project's estimated value. The numerator is enterprise value: market cap plus debt minus cash, the same figure shown in each company's Enterprise Value box and used for EV per ounce. The denominator is exactly the one P/NAV uses: the study's NPV, scaled to the company's attributable share of the project. Below 1.0x means the market values the company at less than its study NPV.

Why show both? P/NAV divides market cap alone into NPV, so it ignores the balance sheet. A company carrying heavy debt looks cheaper on P/NAV than it really is, because its lenders are paid out of the project's cash flows before shareholders are. A company sitting on a large cash balance looks dearer than it really is, because part of its market cap is simply that cash rather than a bet on the project. EV/NPV corrects for both.

A few rules for reading it. If net cash exceeds market cap, enterprise value is below zero and we say so in words rather than printing a negative multiple. If the study NPV is zero or negative, neither ratio is meaningful, and both P/NAV and EV/NPV say "Not meaningful" rather than "Not yet researched". If a company's debt or cash figure hasn't been researched yet, it is counted as $0, exactly as in the Enterprise Value box, and both say the figure is an estimate and which input is missing; with both missing, EV/NPV equals P/NAV. The NPV is the study's figure as of the study date, while market cap, debt and cash are current, so an old study can leave either ratio stale. EV/NPV does not feed any score on this site; it is the ratio the Risk/Reward Asymmetry screen ranks on.

Expected Value (Kelly) Screen

The Min expected value filter on the company directory treats buying a stock as a bet that it re-rates to the equity value of its project, and asks whether that bet has a positive expected payoff, using the Kelly-style formula EV = p·b − (1−p). p, the win probability, is the company's overall score out of 10, read as a probability (an 8/10 is p = 0.8). b, the net odds, is the profit per $1 staked if the re-rate happens.

The re-rate target is equity value: the study NPV scaled to the company's attributable share of the project (the same figure P/NAV and EV/NPV use), plus cash, minus debt. b is equity value divided by market cap, minus 1. Debt and cash are the same figures as in each company's Enterprise Value box. Until October 2026 the target was the attributable NPV alone, which ignored the balance sheet the same way P/NAV does: a company owing more than its project is worth could still screen as a bargain.

Rules for reading it. A debt or cash figure not yet researched is counted as $0, the same rule as the Enterprise Value box; the directory's CSV export has an "inputs counted as $0" column naming which. If debt exceeds NPV plus cash, equity value is floored at zero (a shareholder can lose the stake, not more), so EV bottoms out at −1. A company with no overall score, or no positive study NPV, has no expected value and drops out once the filter is on. So does a company whose latest share price is more than 2 trading sessions old (see the "as of" date), since its market cap, and so b, rests on that older price. This is a screen only: it does not feed any score on this site.

Risk/Reward Asymmetry Screen

The Risk/Reward filter on the company directory flags companies trading unusually cheap against peers at the same stage with the same primary commodity: the cheapest 10% of that peer group, in peer groups of at least 10 companies. The yardstick is EV/NPV: market cap plus debt minus cash, per dollar of the company's attributable share of its study NPV. A flag is a lead, not a verdict. Each one is labelled Known risk: jurisdiction when the jurisdiction score is below 6/10, a known risk that may explain the discount and could re-rate the stock if it resolves, or Unexplained when no driver is visible to the screen. A flagged company's page shows the flag and any researched, cited reason for the discount.

Until October 2026 the screen ranked companies on P/NAV. P/NAV ignores the balance sheet, so a company carrying heavy debt could screen as cheap even though its lenders are paid out of the project first, and a company whose market cap is largely cash could be missed. Ranking on EV/NPV removes both distortions.

Rules for reading it. A debt or cash figure not yet researched counts as $0, as in the Enterprise Value box, and the flag names it (for example "Debt not yet researched (counted as $0)"). A company with neither figure on file is not ranked on EV/NPV, because its EV/NPV would just be its P/NAV under another name. It is ranked on P/NAV against every peer's P/NAV instead, and its flag reads "P/NAV basis: no cash or debt on file". So every ranking compares like with like: EV/NPV with the EV/NPVs of peers that have a balance sheet on file, P/NAV with P/NAV. If net cash exceeds market cap, enterprise value is below zero and the company ranks as the cheapest in its group. A study NPV of zero or below makes both ratios not meaningful, so that company isn't screened. Nor is a company whose latest share price is more than 2 trading sessions old (it shows an "as of" date): both ratios rest on the price, so it is neither flagged nor counted in its peers' rankings until a current price arrives.

Old studies stay in. A company whose only study is old is still ranked on it, because that study is its current one; each flag carries the study-year chip, tinted amber when the study is more than 5 years old, so you can see when a cheap ratio rests on an old price deck. The screen is a filter only and does not feed any score on this site.

Flagship vs. Secondary Assets

A recurring theme across investor-interview coverage of junior miners we reviewed in early August 2026 — unlike the named frameworks elsewhere on this page, this wasn't one analyst's stated rule, just a habit repeated across several interviews: don't let a secondary or pipeline asset inflate how you value a multi-project company. The reasoning given was blunt — those assets "almost never" pan out the way an investor deck implies, so a credible valuation should really be anchored to the flagship (or furthest-along) project, with everything else treated as free optionality rather than something to underwrite.

JuniorMetrics already works this way structurally for a company with more than one tracked project, though not because we built it in response to this specific framework — it falls directly out of the "stage is a claim, not a label" rule the whole site already enforces. stage, and the non-additive Miner-level figures behind Deposit Quality and the NAV used in Graham & Dodd's P/NAV above (mine life, grade, AISC, NPV) all track the company's single most-advanced evidenced project, never a blend averaged across every project it holds. Contained ounces is the deliberate exception: it's a company-wide total, summed across every current Measured+Indicated resource the company holds (Inferred excluded, everywhere, no exception) — legitimately additive because two disclosed resources really do sit in the ground at once, unlike two competing NPV studies or mine-life estimates that can't honestly be combined into one number. A second or third project shown on a miner's page can add real color, and for contained ounces specifically, real additional scale; it never inflates the mine life, cost profile, or NAV figures that actually drive that company's scores, which stay pinned to the flagship project alone — the same discipline this interview commentary argues for.

One nuance worth being precise about, for the figures that do stay project-specific — mine life, grade, AISC, NPV: "most-advanced project" and "market-narrative flagship" aren't always the same project. Occasionally a company's own investor materials center on a large, earlier-stage project while a smaller, further-along project is the one actually driving its stage and scores today — in that case, it's the further-along project's numbers feeding this page, not necessarily the one dominating the company's own marketing. Contained ounces isn't affected either way, since it already pulls from every project the company holds rather than just one. Check a company's Documents and project tabs directly if you want to see how a less-advanced flagship's own numbers compare.

Kaiser's Fair Speculative Value

John Kaiser, publisher of Kaiser Research Online, built a valuation framework around a figure he calls Fair Speculative Value — an attempt to put a number on what a speculative exploration project should be worth right now, given both its ultimate geological potential and how far it has actually progressed toward proving that potential is real.

The model has two separate parts. First, Kaiser estimates the project's "size of the prize" — what the deposit could plausibly be worth if the geological thesis pans out in full. That starts as a geometric estimate (strike length × width × thickness × specific gravity, scaled by an assumed grade) to arrive at a tonnage and contained-metal figure, which is then converted into a potential revenue stream using a metal price assumption and an expected recovery rate, and netted against capex and opex estimates drawn from comparable technical reports on similar deposits and jurisdictions. The result is a theoretical value for the project if the deposit is real and gets built.

Second, Kaiser applies a certainty percentage that discounts that theoretical prize down to what it's actually worth today, given how much of the geological and economic case is still unproven. He ties this percentage to the project's stage of maturity, on a ladder that runs from a fraction of a percent at the earliest grassroots stage up to 100% once a mine is actually producing:

• Grassroots — 0.5–1%
• Target Testing / Drilling — 1–2.5%
• Discovery Delineation — 2.5–5%
• Infill & Metallurgy — 5–10%
• PEA — 10–25%
• Prefeasibility — 25–50%
• Permitting & Feasibility — 50–75%
• Construction — 75–100%
• Production — 100%

This is Kaiser's own progression and doesn't map one-to-one onto the 8-stage model used elsewhere on this site — he splits early-stage exploration into three rungs where we use a single Grassroots stage ahead of Resource Definition, and his "Permitting & Feasibility" rung combines what we track as two separate stages. Multiplying size of the prize by the stage-appropriate certainty percentage produces a Fair Speculative Value range, which Kaiser then compares against a company's actual market capitalization to judge whether it's trading cheap or rich for where it sits on that ladder — the same "is the market pricing this correctly for its stage" question the Lassonde Curve above addresses more qualitatively.

This is a manual, judgment-heavy process, not a mechanical formula you can run off public data. Kaiser builds the "size of the prize" estimate project by project, drawing on his own geological read of the property and hand-picked comparable deposits and technical reports as analogs — two analysts could reasonably land on different numbers from the same drill results. We don't (yet) compute or display a per-company Fair Speculative Value on JuniorMetrics; this section explains the framework so you can apply the same logic yourself when reading a project's technical report, the same way the Lassonde Curve above is explicitly illustrative rather than a per-company measurement.

One related habit worth carrying into how you read any NPV figure on this site: the after-tax NPV a company reports in its own press releases and technical reports — including the NAV used in the Graham & Dodd section above — is typically discounted at 5%, a rate junior companies tend to favor because it produces a larger, more flattering number. Kaiser has pointed out that major producers actually underwrite acquisitions at closer to a 10% discount rate, which is worth keeping in mind as a more conservative sanity check on any headline NPV figure.

Pure-Play vs. Byproduct Producers

Rick Rule draws a valuation distinction between a pure-play producer — one whose revenue is overwhelmingly concentrated in a single metal — and a byproduct producer, such as a copper miner that also sells the gold or silver recovered alongside its copper as a credit against costs. His framing: investors underwriting a pure play are effectively betting on one commodity's price and tend to reward it with a richer market-cap-to-free-cash-flow multiple, while a byproduct producer's economics blend more than one commodity cycle, so the market tends to judge it more on return-on-capital-employed than on that same multiple. Neither structure is inherently better — a byproduct credit that meaningfully lowers a mine's all-in sustaining cost is a real, valuable thing — but reading a byproduct producer's valuation multiple against a pure-play peer's as if they were directly comparable misses why the market prices them differently in the first place.

A company's detail page shows a Pure-Play badge only when we have a cited figure showing its primary commodity is at least 85% of its revenue (for a producer) or of its value (a study's life-of-mine revenue, otherwise), and the badge's note states that share and its basis. Below that share it shows Multi-Commodity with the share stated. We don't have that split for most companies — many juniors and even mid-tier producers don't disclose one — so where it hasn't been researched, a company tracked with one or more secondary commodities still shows Multi-Commodity (real exposure, though not a revenue split), and a company with neither shows no badge at all: the absence of a recorded by-product is not evidence that there isn't one, so the page just shows the primary commodity.

Optionality Scale

Rick Rule also draws a minimum-size line under what counts as a genuine optionality play — a deposit that isn't economic to mine at today's commodity price, but would become economic at a meaningfully higher one. His framing is a floor, not a ceiling: he wants roughly a million-plus-ounce deposit before he considers that bet credible — "a deposit where if the price goes up somebody has to buy it." Below that rough size, his own point is that a price recovery alone doesn't make the deposit a real acquisition target — it's too small to move the needle for anyone who would actually buy it, so it isn't a genuine call option on the commodity price at all, just an unproven resource. It's worth stating the direction plainly since it's easy to get backwards: a small deposit isn't automatically "the optionality play" by virtue of being small and speculative — in Rule's own framing, it can be too small to qualify as one.

This only applies to companies that aren't already producing. A producer's deposit is by definition already economic at current prices, so it isn't a price-optionality bet in the first place — it's cash flow now, not a call option. Construction-stage companies are excluded on the same logic: reaching that stage already means a positive feasibility study, i.e. the project is already underwritten as economic at current prices.

Rule's interview only speaks to gold ounces, so for other metals we use the midpoint of each metal's deposit-size scale from Deposit Quality above (the “Scores 5” column of its table). For gold that midpoint is almost exactly 1 million oz, Rule's own figure, and we hold it there whenever the gold scale is recalibrated. Every other scale except nickel's was built around the amount of metal worth about the same in the ground as 1 million oz of gold, so the threshold means the same thing in value terms for each: about 300,000 t of copper (661 million lb), 70 million oz of silver, 60 million lb of U₃O₈, 1 million t of zinc, 250,000 t of lithium carbonate equivalent and 200,000 t of total rare earth oxide. Lithium's price moves far more than gold's, so its threshold is only as current as that price. Nickel's threshold (about 790,000 t) simply falls out of its scale and hasn't been tested against Rule's benchmark.

Softer and more qualitative than a hard cutoff — this is one resource investor's stated heuristic, not a peer-reviewed economic threshold, and we surface it as a labeled badge on a miner's detail page rather than as a filter that removes companies from view anywhere on the site.

How to Read a Technical Report

Most of the numbers junior miners are actually valued on — NAV, AISC, mine life — come from a single document: the technical report filed alongside a resource estimate or economic study. We link out to a lot of these in each miner's Documents section, and they run 100–300+ pages, but a retail read genuinely only needs the Executive Summary — usually the first 10–30 pages — which restates every figure below in plain prose before the report descends into geology and engineering detail aimed at other professionals.

PEA vs. PFS vs. FS

Under NI 43-101 (the Canadian disclosure standard most of these companies report under — JORC is the Australian equivalent and SK-1300 the US one; all three share the same broad logic), a project's economics get studied in three progressively stricter stages, and they are not interchangeable levels of confidence:

• PEA (Preliminary Economic Assessment) — the earliest and by far the least reliable study. It can be built partly or entirely on Inferred Resources — the lowest-confidence resource category — and on conceptual mining and processing assumptions that haven't been engineered in detail. NI 43-101 is explicit that a PEA's results cannot be used as the sole basis for a production decision. Treat a PEA as a "this could work, here's the rough shape" sketch, not a number to anchor a valuation on by itself.
• PFS (Pre-Feasibility Study) — a materially more rigorous pass, leaning mostly on Measured and Indicated Resources with engineering taken to a level detailed enough to screen the project and choose a preferred development option. Still not considered bankable, but the assumptions behind it are far harder to wave away than a PEA's.
• FS (Feasibility Study) — the highest-confidence study, engineered to a level lenders and boards actually use to commit real construction capital. It requires Reserves (Measured/Indicated Resources converted through a positive economic test), locks in vendor quotes and detailed engineering rather than factored estimates, and is the study a company's formal construction decision — the same milestone behind the Pre-Production Sweet Spot above — is actually built on.

In short: the closer a company is to production, the more its numbers are worth trusting at face value. A PEA is a hypothesis: an early, if geologically promising, estimate not yet a plan.

The numbers worth finding

• NPV (Net Present Value) — the project's discounted after-tax cash flows, i.e. what the study says the mine is worth today. Always check what discount rate it's calculated at — reports commonly show it at both 5% and 8% (sometimes more rates in a sensitivity table), and a lower rate produces a bigger, more flattering headline number. This is the same NAV figure discussed in the Graham & Dodd section above, and the same 5%-vs-10% caution raised at the end of the Fair Speculative Value section.
• IRR (Internal Rate of Return) — the discount rate at which NPV would hit zero; a rough proxy for the project's own rate of return. Compare it against the discount rate used for NPV — an IRR only a few points above the discount rate is a much thinner margin than headlines suggest.
• Payback period — how many years of production it takes to recover the initial capex. Shorter is generally lower-risk, all else equal, since less of the return depends on assumptions holding up years out.
• AISC (All-In Sustaining Cost) — the fully-loaded cost to produce one unit of metal, including sustaining capital. This is the same cost figure behind the Deposit Quality score above; the lower it sits relative to spot price, the more margin (and downside cushion) the project has.
• Initial capex — the upfront cost to build the mine. Compare it against the company's actual market cap and treasury — a project that costs several times the company's current market cap to build is a project that will need a lot of dilution or debt before it produces anything.
• Mine life — years of planned production at the modeled rate. A short mine life leans harder on exploration upside or a future expansion to be worth more than the study's own numbers.
• Strip ratio (open pit only) — tonnes of waste rock moved per tonne of ore. A high strip ratio means more of every dollar spent on mining goes toward moving rock that isn't paying for itself.

Red flags

• Commodity price assumptions well above spot or consensus long-term price. Every economic study has to assume a metal price to generate its revenue line, and a study built on an optimistic price will produce an inflated NPV and IRR even if every other input is sound. Always check the assumed price against where the metal actually trades — and against consensus long-term forecasts, not just today's spot — before trusting a headline number.
• No disclosed price assumption at all. If a press release headlines an NPV or IRR without stating the commodity price it's built on, that's a gap worth chasing down in the Executive Summary before taking the number at face value.
• Stale studies. A PEA or FS announced several years ago, with metal prices, capex inflation, and financing conditions having moved a lot since, can be badly out of date even if no single input was unreasonable when it was written. Check the report date, not just the headline figures.

How we label a study's age

Every figure that comes out of a technical study (NPV, P/NAV, EV/NPV, IRR, initial capex, NPV:Capex, mine life, and a pre-production AISC) carries the year of the study behind it, for example "2020 study · gold deck US$1,820 vs spot US$4,182/oz". The year is the study's effective date, taken from that figure's own citation; where the study states a metal price, the label puts it next to today's spot (or periodic reference) price, the same comparison as the Study Price Assumption card. A figure whose citation has no date gets no label rather than a guessed one. On /compare, /miners and each Risk/Reward flag the label is a small year chip: hover it for the details, and a chip tinted amber marks a study more than 5 years old.

Peer medians for these figures use studies from the last 5 years only: a peer whose study's effective date is older than that is left out of the median, so a company's P/NAV or IRR is measured against economics struck at broadly comparable metal prices. The old study is not hidden. It stays on its company's page, in every sort and in the Risk/Reward screen, because an old study the company still relies on is its current study: nothing newer has replaced it, and calling it wrong would be a claim we can't support. When a newer study from the same issuer does replace it, we update the figures themselves. If you'd rather not see old studies in a ranking, /compare and /miners both have a "Hide studies older than 5 years" filter, off by default.

Where We Source Our Data

JuniorMetrics pulls from a small number of real sources rather than one black-box feed, and the reliability of a number on this site tracks pretty directly to which of these it came from:

• EODHD (market data + fundamentals) — a single paid EODHD account on their ALL-IN-ONE plan (100,000 API calls/day) covers everything we pull programmatically: real-time quotes, EOD/historical prices, FX rates, spot Gold and Silver prices, and EODHD's Fundamentals endpoint, which feeds company financials, the Holders data behind institutional ownership pages, insider transactions, and company news. This is the same scheduled daily job described in How Our Data Is Verified below — a repeatable pull, not a manual re-type.
• Regulatory filings and company disclosures — SEDAR+ for Canadian issuers, SEC EDGAR for US issuers, and ASX/AIM announcement platforms for those exchanges. This is where technical reports live — the PEAs, PFSs, and Feasibility Studies filed under NI 43-101, JORC, or SK-1300 described in How to Read a Technical Report above — along with press releases and MD&A, which is where most of our hand-researched cash, debt, and project-economics figures are pulled from.
• Company investor-relations and management pages — the source for the current management-team data behind the Management score above: who's currently on a team, and what other companies they've held roles at.
• Per-metric source citations — structured numbers like NPV, AISC, and contained ounces can carry a citation back to the specific document or URL they were pulled from (the document icon described in How Our Data Is Verified below). This is genuinely a work in progress, not full coverage: 1203 of the site's 1203 miners currently have at least one cited metric. We're backfilling this over time rather than claiming it's done.
• Community-sourced data — Community Rankings, covered above in Community Rankings, is the one category on this page that isn't drawn from filings or disclosures at all. It's user-submitted opinion, plainly labeled as such wherever it appears, and it should be weighed differently from everything else in this section, which traces back to a real filing, report, or company page.

Share Prices and the "As Of" Date

Share prices refresh every hour from our market-data feed. Each price carries the market date of the trading session it came from, and we compare that date with the latest session on the company's own exchange (Toronto, New York, Sydney, London and so on), counting sessions on that exchange's real trading calendar, so a public holiday is not a missed day. A price more than 2 trading sessions behind is shown with a small amber "as of" date, for example "as of Aug 22, 2026": it is the latest price available to us, and our data feed has no newer quote. Usually that means the feed has stopped carrying the company, for example a listing it does not cover or a ticker change it has not picked up yet. A current price shows no date, so most pages never show the label.

The same date appears beside every figure built on that price: market cap, enterprise value, EV per ounce, P/NAV, EV/NPV and P/CF, on the company page, the directory cards, the comparison table, head-to-head and favourites. A company with an older price is also left out wherever an old price would mislead others: the peer medians for those figures (each median's hover says so), the Risk/Reward Asymmetry screen (neither flagged itself nor counted in its peers' rankings) and the expected-value screen. It returns to all of them as soon as a current price arrives. Pages are rebuilt at least every ten minutes, so a label can trail a new price by up to that long.

How Our Data Is Verified

Different pieces of data on JuniorMetrics come from different places, and go through different checks before they show up on the site:

• Live market data — price, volume, and FX rates are pulled directly from EODHD's market data feed on a scheduled daily job, not manually re-typed. The same job backfills and extends each miner's real trailing price history and 52-week range from actual exchange data.
• Company fundamentals — cash position, debt, jurisdiction rationale, and technical-report figures (mine life, cut-off grade, NPV, AISC) are hand-researched from primary sources: company filings, press releases, and NI 43-101 / JORC-style technical reports. None of this is scraped or generated wholesale — it's entered figure by figure, source by source.
• Source citations — for a growing set of metrics, a small document icon next to the number links straight to the original source document (and page, where we've recorded one) that figure was pulled from, so you can check the primary source yourself instead of taking our word for it. Coverage is partial and expanding — the icon only appears once a metric has actually been sourced.
• Internal data-quality audit — behind the scenes, we run a repeatable audit across every miner that flags missing or incomplete data and ranks the gaps by how many companies they affect. That's how we decide what to research next, and it's the same discipline that produces the unverified-data flags described below — we'd rather surface a gap than quietly paper over it.
• Drift detection — missing data isn't the only way a number goes wrong: a figure can be entered correctly and then the real world moves on (a construction decision, a first pour, a suspension, a financing close) without anything on this site noticing. We never let incoming news automatically rewrite a company's data — a mis-parsed headline could introduce a new error — but we do scan every incoming headline for language that signals exactly this kind of change and flag the affected company for a fresh human or research-agent look, and we separately track how long it's been since a company's hand-researched content was last verified, weighted by how fast that company's stage typically moves. 26 of the site's 1203 miners currently have a pending recheck flag from this system.