Actionable insights — $83 Crack Spreads & Why Canadian Oil Became Valuable

4:41 1. Price the margin, not the average — "macro guys are average guys"

1. Refuse to reason from sector averages. "I don't care about the cost structure of the vast majority of producers" — the price is set by whoever is deciding at the margin, so find that decision-maker and model their tradeoff.
2. Ask the marginal question explicitly: who is the last producer that has to be induced to supply, and what price does that specific barrel/tonne/cargo need? That number is the floor, not the industry-average cost curve.
3. Treat a market as a negotiation between two people rather than a curve on a chart — the information content is in the tradeoff being made, not in the aggregate.
4. Expect the macro consensus to be systematically blunt here: "macro guys are average guys, micro guys are marginal guys. There's a lot more different information content found at the margin than there is at the average."

• Any consensus built on an average (average cost, average inventory, average capacity) in a market where one constrained marginal unit sets the clearing price.

14:16 2. Price the thing people actually consume — watch products, not the headline commodity

1. Start from consumption: "nobody consumes crude oil." Identify the refined/finished form the end user actually buys — diesel, gasoline, jet — and price that.
2. Measure the conversion margin between raw input and finished product (the crack spread; the 321 crack is one-third diesel, two-thirds gasoline). A margin that widens toward or beyond the raw material's own price is a capacity signal, not a demand signal.
3. Locate the constrained step in the chain — here refining (Russian plants bombed, capacity locked behind the straits, Chinese teapots idled) — and ask how long replacement takes. Refineries cannot be built globally on demand.
4. Check whether a strategic buffer exists for the constrained form. Crude has an SPR; refined products have nothing.
5. Own the constrained step, not the abundant one: buy the petroleum indices rather than picking one product, or on the equity side refiners, producers and integrateds ("you get a little bit of both").

• A conversion margin (crack, smelter TC/RC, fabrication premium) near record highs while the raw commodity is flat — and whether a strategic reserve exists for the constrained form.

15:15 3. Bottleneck rotation — bank the spikes instead of demanding a trend

1. Hold the structural thesis (chronic under-investment) constant and accept the expression rotates: crude in March/April → refined products → copper → gold and silver. "The bottleneck changes and rotates across these different markets, but the trend is the same."
2. Do not require a continuous uptrend to validate the thesis — "it's a sequence of spikes and rotating," which is exactly why critics score you wrong between spikes.
3. Harvest rather than hold: "the way these commodity investments work, you're banking those spikes." Take the spike in the market that broke, then re-deploy into the next constrained one.
4. Scan the whole complex, not one market — energy, metals, grains, softs, petrochemicals and fertilizers are all running the same shortage script.
5. Use the last spike's exhaustion as the entry cue for the next: he was short gold March→June, went flat, and re-entered precious just as copper's move matured.

• Which market is currently the binding constraint, and which one has just finished spiking — the rotation cue. Cross-complex anomalies (cocoa's fifth-largest move on record, a hard corn move) as evidence the script is running everywhere.

34:21 4. "Everything reprices" — stress-test an asset's IRR, not its commodity deck

1. When someone says a price "can't" get there, remember the cost side moves with it. Rising: 2000 oil at $20 earned mid-20s IRRs; by 2006–07 oil was 3x higher and returns were lower, because steel, labour, capital and services all repriced up.
2. Run it in reverse before underwriting a downside case. Take the actual asset, cut the commodity deck hard, and reprice every input alongside it — steel, copper, labour, FX, food, fertilizer.
3. Expect the IRR to be far more resilient than the price move implies. A Calgary upstream asset underwritten on a ~$110 deck at ~25% IRR still returned ~17% at $40 oil — a ~66% price collapse.
4. Draw the survival rule from it: "you basically don't want to hold on to anything when you go through these things. You'll live through it if you let it reprice." The businesses that die are the ones whose costs can't move.
5. Identify who can't reprice — the owner of the fixed physical asset. "Those companies like Chesapeake who own the land got killed… probably the same thing with the AI guys."

• FX (the ~60% Canadian-dollar move), steel and labour indices moving with the commodity; and, in any boom, which participants own inflexible physical assets they cannot re-cost.

37:32 5. Buy the cheapest asset inside the cheapest sector — then roll it up for cash

1. Two-step the value screen. First find the cheapest sector on the screen ("oil is the cheapest thing on your screen"), then find the cheapest, most-disliked sub-segment inside it ("shallow water is the cheapest asset inside oil. People just don't want it, don't like it").
2. Price it on reserves in the ground, not on production multiples — here roughly $4 a barrel of reserves.
3. Prefer mature assets on the flat tail of the decline curve where a small maintenance spend restores output nobody has funded since COVID (~3,000 boe/d expected up to "four and change" quickly).
4. Underwrite for cash returned, not growth: "growth is a dirty little four-letter word." Fund a dividend (5% planned) and grow only by accretive roll-up, with the next targets already identified.
5. Check the plumbing before the geology: existing infrastructure removes the offtake question (US Gulf Coast refineries, ~250 miles of pipe with the assets).
6. Back the operator who has already built the same thing in the same basin — CEO Tim Duncan, who built Talos.
7. Own the picks-and-shovels leg alongside the asset leg: if the marginal barrel comes from shallow water, the rig fleet that drills it is the second expression.

• Sub-segments priced per unit of reserve rather than per unit of cash flow; capital that stopped flowing at a datable event (COVID) and never resumed; an operator with a prior exit in the identical basin.

46:51 6. The 120%-of-cash-flow top signal, and the management-turnover confirmation

1. Measure the beloved sector's capex against its own cash flow. When it crosses ~100% and heads for 120%, the cycle is late: "when you're at 120% and the thing dies, it's game over."
2. Confirm with the behavioural cycle, which runs in a fixed order: overspend → bust → "we will never do that again" → capital discipline while prices rise → prices scream → a new management team arrives that never learned the lesson and starts spending again.
3. Use management turnover as the hard evidence of where you are: at the 2014 LME dinner, "out of 13 companies, only one management team was still the same from 2012."
4. Apply the multiple consequence: a business that has to put steel in the ground eventually earns a steel-in-the-ground multiple. "They're going to have to learn how to get much lower multiples, more like what the energy guys did."
5. Position on the other side — the sector still in capital-preservation mode, paying cash out rather than burning it.

• Capex/cash flow crossing 100% in the crowded sector; wholesale management turnover after a bust; a capital-heavy business still carrying an asset-light multiple.

24:22 7. Trace a chokepoint all the way to the funding cost — the Bretton Woods chain

1. Don't stop the analysis at the barrel. Follow the chain: physical chokepoint → the security guarantee that keeps it open → the reserve-currency bargain that guarantee bought → the sponsor's cost of funding.
2. State the bargain plainly so you can test it: since 1945 the US kept the sea lanes open with an inherited 400-year network of ports (Malacca, Diego Garcia), and the world used the dollar in exchange. Break the guarantee and "it's not only the end of globalization, it's the end of Bretton Woods."
3. Price the exorbitant privilege with a live comparison, not theory: Switzerland's 30-year fixed mortgage at ~50bp because "Swiss franc is good as gold." Demand for a currency lowers its issuer's funding cost — and the US is running a 7% fiscal deficit into that arithmetic.
4. Find the transmission channel that moves first. Here it's the yen carry — oil goes to Japan, the yen was the world's cheapest funding currency and biggest short, and the proceeds bought US assets. "Why is Bessent fighting the yen? Because all that yen comes out of the US, interest rates go up."
5. Read official rhetoric as management, not information: since Carter's sweater speech the rule has been "never admit to the scarcity — create the illusion of abundance." Every president since Bush senior has used the SPR and its language to talk the market down, so discount calm official messaging in a physically tight market.
6. Then hold the asset that benefits from the chain breaking rather than the one that benefits from it holding — precious metals against a reserve-currency and funding-cost impairment.

• Yen/USD-JPY and official intervention as the fast channel; 30-year funding-cost spreads between reserve and safe-haven currencies; SPR rhetoric versus SPR draw data; First Nations agreements and Canadian pipeline/tidewater capacity as the physical re-rating.