Actionable insights — How drill core unlocks uranium discoveries

00:32 1. Treat a geophysics map as a hypothesis, not a discovery

1. When a release leads with a survey image, ask what physical property was actually measured — density, gravity, conductivity — and note that the answer is never "uranium."
2. Read the colour ramp as relative contrast between rock volumes, not as a map of contents: "this rock is denser than that rock," nothing more.
3. Distrust the apparent precision of the picture. The boundaries are gradients, not edges — "there's not distinct lines" — so any target outline drawn on top of them carries far more uncertainty than the image suggests.
4. Give partial credit where a method genuinely resolves something specific (a graphitic conductor), then immediately ask the geometry question the method cannot answer: how is that feature oriented, where does it start, where does it go?
5. Downgrade "we have identified a target" from a result to a testable claim, and wait for the drill to adjudicate it.

• Releases that describe a geophysical anomaly in language normally reserved for mineralisation — the tell that a hypothesis is being marketed as a finding.
• Whether the company states the survey method and the property it measures at all; omission is itself informative.

01:47 2. Judge early holes on geology learned, not on assays hoped for

1. Before the results land, write down what the company said it expected to intersect — rock type, structure, target depth, depth to the unconformity. That is the actual test.
2. When the hole reports, score it against that list rather than against the headline metal grade. A hole that confirms rock type, fault orientation and target depth has advanced the model even with no mineralisation.
3. Separate the two failure modes: a wrong model (the geology isn't there) is fatal to the target; a wrong intercept (the geology is there, the drill missed) is a positioning problem and usually cheap to fix.
4. Expect misses as the base rate, not as bad news, and calibrate your reaction to a "disappointing" first hole accordingly.
5. Only treat radioactivity or grade as the primary signal once the structural model has been confirmed by the earlier holes.

• Whether the release reports depth-to-unconformity, rock types and structural observations, or only radioactivity — the former means the model is being tested, the latter means the news is being managed.
• Language distinguishing "we didn't intersect the target" from "the target isn't there." They are very different outcomes.

02:05 3. Apply the three-hole rule before forming a view on a target

1. Count the holes on the specific target — not on the property — before assigning it any value in your own model.
2. Below three holes, treat the geology as unconstrained: you have points, not lines. Interpolation between holes is what turns observations into a structural picture.
3. Allow one exception, the one Frostad allows: an obvious hit against an explicit prediction is knowable immediately. Everything ambiguous requires the third leg of the stool.
4. When holes two and three land, check whether the company revises its structural interpretation. A model that survives triangulation unchanged is worth far more than a model that has been quietly redrawn each time.
5. Resist re-rating a stock on a single hole in either direction, and size accordingly until the third hole is reported.

• Programs that report hole 1 loudly and go quiet — the missing legs of the stool.
• Whether the next collar is a small step-out ("just turn the drill a bit") or a complete move; the size of the step reveals how much confidence the last hole actually created.

03:12 4. Remember that the deliverable is a 3D structural model, not a core photo

1. List what has to happen after the core leaves the ground — detailed logging, structural interpretation, geochemistry, lab assays, a gamma probe down the hole — and treat any conclusion drawn before that work as provisional.
2. Ask what the company is building with those inputs: rock types plus logging plus probe data assembled into a 3D image, then several holes joined to resolve fault and structural orientation.
3. Keep the actual exploration objective in view. The search is for structural complexity capable of hosting a deposit — not for a single spectacular interval.
4. Weigh a company's technical apparatus (modelling software, physical mapping, the ability to present the model so specialists can critique it) as part of its ability to convert spending into knowledge.
5. Reframe the whole budget accordingly: geophysics, surveys, geochemistry and drilling are expenditures that produce refined data, and data quality is what compounds across a program.

• Whether successive releases show an evolving 3D or structural interpretation, or just serial intercept tables — the former is a program learning, the latter is a program drilling.
• Mentions of downhole geophysics (gamma probe) and lab turnaround, which set the real gap between drilling and knowing.

04:25 5. Check that the structural data is orientable — the measurement is destroyed if it isn't

1. Understand the failure mode first: a structure visible in a core stick has no meaningful direction unless the stick's rotation relative to the ground is known. Rotate the core and the structure appears to strike somewhere else entirely.
2. Confirm the program is running oriented core — tools that preserve the core in the orientation it came out of the ground in — wherever structural targeting matters.
3. Recognise what that unlocks: with orientation you can "start chasing it," projecting a fault or contact to the next collar; without it you have a description, not a vector.
4. Check the on-rig recording discipline too: rock type, structures, faults, seams, all measured and marked by distance, and compared against the pre-drill prediction (target depth, depth to unconformity).
5. Treat this as a proxy for technical quality generally — a team that spends money preserving orientation is a team that intends to use the data.

• Explicit mention of oriented core or structural measurements in technical releases on structurally-controlled targets.
• Whether reported structures come with an orientation at all — an unoriented "strong structure" is close to uninformative.

05:35 6. Price a hole, then read a drilling pause as discipline rather than trouble

1. Put a cost on a single hole in the relevant jurisdiction and depth range — here $300,000–500,000 — and divide the company's treasury by it. That is the real measure of how many attempts the story has left.
2. Distinguish what can be decided at the rig (visual observations, small collar adjustments) from what requires the data to be digitised, manipulated and compared. Only the first is available "on the fly."
3. Read a scheduled hiatus as the interval bought to make the expensive decision properly: abandon the target, or spend another hole confirming that the last one merely missed.
4. Frame the recurring judgement explicitly — "did we kill it, or do we need to drill another hole just to make sure it didn't go in the wrong direction" — and see whether management articulates it that way.
5. Invert the usual reflex: a program that pauses to interpret before spending the next half-million is behaving rationally; a program that keeps turning the drill without a revised model is spending capital to generate news.

• Whether a gap in drilling is announced as a planned interpretation break or simply goes unexplained — and whether the program that resumes reflects a revised model.
• Treasury divided by cost-per-hole as the honest count of remaining attempts, and any financing that has to happen before the third leg of the stool is drilled.