Actionable insights — Pick Your Fighter: Precious Metals vs. Copper

1. Rank competing bullish stories by whose positioning is cleanest — the narrative is the tiebreaker, not the thesis

1. Start from the admission that you can be right on fundamentals and still lose. Set the weight explicitly before you look at anything: "if positioning and flows are 90%+ of a story today," then fundamentals are the remaining 10% and cannot outvote them.
2. Take every candidate long that shares a macro driver (here: a topping USD, a Triple Yasu Risk-Off) and score them against each other on positioning, not on how much you like the story. You are not asking "is this bullish?" — you are asking "which of these has the emptiest boat?"
3. Apply the veto rule from the last mistake: refuse anything "85% there," or where technicals and fundamentals conflict. Breaking or broken momentum, or a chart that simply will not confirm the bullish narrative, disqualifies the trade regardless of how good the story reads.
4. Separate liking an asset from owning it. Write down the temperamental bias you carry into the analysis so you can discount it — the asset you are philosophically drawn to is the one you will over-forgive.
5. Allow the answer to be "the less romantic one." The output of the screen is a ranking, not a veto on the runner-up: he stays constructive on gold while refusing to buy it yet.

• The moment the runner-up's positioning does wash out — that converts the ranking, and it is a scheduled re-check (weekly CoT/LME reports), not a vibe. And watch your own language: needing to argue that fundamentals will "eventually" win is the tell that you are at 85%.

2. The capitulation test — did specs buy the decline or get carried out on it?

1. Overlay the price drawdown on speculative net length across the same window. The question is not the level of positioning but its direction as price fell.
2. Score it: net length rising into a falling price = holders averaging down = no capitulation, no durable low. Net length collapsing faster than price = a washout worth buying.
3. Use the prior smash as the yardstick. If positioning is longer now, after a second decline, than it was at the trough of the first one, the market has absorbed no pain at all.
4. Read positioning three ways because they disagree — contracts, $ notional, and % of open interest. Contracts mid-range with $ notional at the high end means the same book is much bigger in money terms than it looks.
5. Check the fund flows separately from the futures: ETF/AUM outflows can show one cohort capitulating (the miners, the juniors, fast-money retail) while a different cohort (long-term physical holders) has not moved at all. Only the cohort that has not sold can still sell.
6. Note where the market's centre of gravity has moved. If physical trade has migrated to another region, the local exchange's absolute size is less meaningful — but its positioning remains a clean read on the local speculative cohort.

• The flip: a sharp drop in spec length on a down week (holders finally quitting), ETF stockpiles rolling over in earnest rather than flattening, and $ notional falling with contracts rather than diverging. That is the entry he is waiting for in gold.

3. The visibility tell — when a trade becomes consensus among people you respect, expect to wait a year

1. Track the inbound rate as a sentiment series: how many notes on this theme have you read in the last few weeks, and how many people have asked you about it? Rising inbounds are an early crowding indicator that no positioning dataset publishes.
2. Note specifically when the idea appears among analysts you rate highly. That is not confirmation — it means the smartest slice of the flow is already in, so the marginal buyer must now come from a less-informed cohort.
3. Retrieve the base rate from the last time this happened: the theses were valid, fundamentals improved, and it still took ~a year to work. Price that time cost into the trade — a year of opportunity cost is a real loss even when the drawdown is small.
4. Convert it into a sizing/timing rule rather than a veto: if the idea is visible and positioning is not washed out, wait; if the idea is visible and positioning is washed out, that's the squeeze.

• The idea reaching the tier below the people you respect (sell-side summaries, retail feeds) — usually the point at which the wait ends one way or the other. And the inverse signal: a good idea that nobody is emailing you about.

4. Audit the geography of inventory before you read the inventory number

1. Never take global exchange stocks as one number. Decompose by exchange and ask what fraction sits in each venue — a headline total can be bearish while every tonne that is actually reachable has vanished.
2. Identify any policy wedge that immobilises one venue's stock: a tariff, an export ban, a sanction, a trust structure. The test is whether metal can physically and economically flow back out. If the answer is no while the policy stands, that inventory is sequestered and should be excluded from the tradeable balance.
3. Price the wedge: measure the inter-exchange premium that the policy sustains and treat it as the pump. As long as the premium persists, the flow continues in one direction.
4. Attach the sequester to its political duration, not to a forecast — "as long as the threat exists," i.e. as long as this administration's ideology holds. That, not a supply/demand model, is the timeline on which the position must be sized.
5. Re-plot the ex-sequester series against its own seasonality (5-year and 15-year). The comparison must be seasonal because inventories have a construction-cycle rhythm; the signal is a break in the shape, not the level.
6. Find the historical analogue where the same structure already played out and check how the price behaved then.

• Any softening of the policy (tariff carve-outs, an exemption, an administration change) — that unlocks the sequester and reverses the flow, which is the single biggest risk to the trade. And weekly ex-sequester stocks: a 4Q that fails to draw would break the thesis on its own terms.

5. The physical-premium inversion — when the marginal buyer pays up into strength

1. Establish the normal reflex of the regional physical premium first: in a well-supplied market, a futures rally makes price-sensitive buyers step away and the premium narrows.
2. Catalogue the historical episodes when the premium widened and record where the futures price was each time. If every prior blowout occurred with the benchmark near a low, the premium is a bargain-hunting signal, not a scarcity signal.
3. Then look for the inversion: the premium widening with the benchmark at the highs. That is a buyer who needs metal irrespective of price — a short in the physical, not an opportunist.
4. Corroborate with a second, independent read on the same region's stocks (a broker's inventory work) so you are not trading one series.
5. Then check the time-spread for confirmation: a cash-vs-3-month spread flipping into backwardation (nearby metal priced above deferred) means someone is paying to have it now. Compare today's whole curve against -1wk, -1m and -6m to see the shift rather than the level.
6. Only when premium and curve agree do you have a physical-tightness call rather than a chart.

• The premium narrowing back on a rally (normal reflex restored = thesis over); backwardation deepening or spreading down the curve (tightness generalising); and the same inversion appearing in the rest of the world, which is the step from "China is short" to "everyone is short."

6. When one exchange's book looks extreme, go find the other book

1. Steelman the crowded-trade objection with its own best data before answering it — state the extreme honestly ($ net long near record, % of OI stretched, the historical analogue when that flagged a top).
2. Then ask the structural question: is this exchange where the marginal player would express the trade? If the squeeze is coming from a region, look at the exchange that region trades on.
3. Pull the second venue's open interest and speculative net long (contracts and $ notional) and compare each to price. Crashed open interest with price at all-time highs is the configuration to hunt: it means the rally has happened without speculative sponsorship.
4. Aggregate the venues before concluding. One extreme plus one washout can net to "not particularly extended" — which is the answer that matters for risk.
5. Convert it into the squeeze mechanic: if specs are not long, they have no longs to sell to commercials who need to cover. Ask the closing question out loud — where does the paper or physical metal actually come from?

• LME open interest rebuilding on the way up (fresh spec money arriving = the free option decays); and the generalisable version — any market where price makes highs while its dominant venue's OI collapses deserves this treatment.

7. Cross-commodity ratios as the valuation anchor — solve for the implied price

1. Build the ratio between the asset you want and the one the crowd already owns, in consistent units on one exchange's pricing (state the units explicitly — $/t vs 1oz, or 1oz : 1lb — or the levels are meaningless).
2. Mark the historical range and, on the chart, the significant lows. Identify the threshold below which forward absolute returns have historically been good, and the mean-reversion band the ratio spends protracted periods inside.
3. Note the path, not just the level: once the ratio turns up from a major low it has tended to keep running back toward the top of its band. Trend from a low beats level.
4. Solve for the implied price by holding the other leg still ("assuming gold stays at $4,000," "assuming silver stays at $58"). This is a scenario, not a forecast — but it converts a ratio into a target you can size against.
5. Use two independent ratios and take the corroboration seriously only if they agree. Prefer the pair with a shared demand driver (industrial usage) as the "cleaner," more comparable read.
6. Record the exception that broke the rule (here 2011, when a secular metals bear began) — the ratio signal is invalid inside a structural downtrend, and knowing the exception is how you recognise one.
7. Then stress-test the conclusion against the strongest counter-case: find the historical instance where positioning was already extended and the ratio still worked, to see whether crowding actually caps the move.

• The ratio failing to turn up from the low (the 2011 failure mode — check whether a broad metals bear is starting); and the other leg moving instead of yours (if gold falls, the implied copper price falls with it — the target is a relationship, not a level).

8. Name the strongest argument against your trade — and answer it with location, not demand

1. Identify what the market has decided is the primary price driver of your asset (a broker's factor attribution will name it) and treat that driver as your largest single risk, not your thesis.
2. Go find the listed proxies for that driver and read their charts. The equities most directly geared to a theme break down before the commodity does — they are the early-warning system for your own position.
3. Retrieve the historical rhyme for a crowd owning a commodity as a hedge to something else (a growth book buying a "value" energy hedge). Themed crowds unwind for reasons unrelated to the commodity's own balance.
4. State the risk in plain terms and refuse to soften it — "this is a real risk," "the trade is not 'sleep at night'."
5. Then answer it on a different axis than the one being attacked. If the attack is on demand, the defence must be on supply, location or positioning — "even if projected global balances were to notably soften… the metal is simply in the wrong place."
6. Decide with the discomfort intact: "as a buddy once said, do the hard trade."

• The buildout proxies breaking down further while copper holds (the divergence that proves the location argument) versus copper following them down (the demand argument winning). And the AI-hedge crowd actually liquidating physical/futures exposure rather than just the equities.

9. The commercial squeeze template — a rally where spec positioning falls is the violent one

1. Find the most recent violent move in a comparable commodity and plot spec positioning through it. Note precisely where positioning peaked relative to the price move.
2. Recognise the signature: positioning peaking early and then declining for the rest of the advance means the buying was commercial — physical shorts covering — not speculative. Those are the moves that go multiples, because the buyers must have the metal regardless of price.
3. Test the current market for the same precondition: low open interest and low spec length, so that when commercials scramble there is nobody to sell to them.
4. Watch the curve for the trigger. Tightening time-spreads / backwardation are the mechanical tell that the scramble has begun. Treat that as the entry signal, not as confirmation to wait on.
5. Size for a dislocation rather than a drift: this is an accident-shaped payoff, so the position must survive being early and be big enough to matter if it ruptures.

• Spec length rising with the price (a normal, speculative rally — smaller and more fragile); and, on the way up, whether open interest stays low (commercials still covering) or fills in (the squeeze is being sold into).

10. Express a commodity call where the multiple gap is widest — and let liquidity dictate the execution

1. Once the commodity call is made, screen the equity expression by relative valuation, not by quality. Price the large caps first (forward PE, forward EV/EBITDA) and ask whether they already discount your target price.
2. If the large caps trade "rather full," step down the market-cap curve to advanced explorers and junior producers — the cohort whose valuations are "terribly depressed" and whose discount to the majors has "grown extreme." The discount gap is the edge, not the absolute cheapness.
3. Structure it as a basket with deliberate asymmetry of size: a small number of highly concentrated positions where the work is done, plus tracking positions at "inconsequential size for now" that keep you engaged and give you a foothold to scale into as each thesis confirms.
4. Let liquidity set the mechanics: in thin names, never use market orders. Assume the execution can cost more than the analysis is worth.
5. Keep the disclosure honest about what it is — a position, not a recommendation — and hand the work back: "please do your homework."

• The discount closing from the top (majors bidding for juniors — the M&A wave he has positioned for before) versus from the bottom (juniors re-rating on their own drill results); and the warning sign that the gap is justified — financings at ever-worse terms, which turn "depressed" into "diluting."