Actionable insights — $20 to $147 Oil: Super Spike vs. Super Vol

22:231. The top-projects screen — aggregate every announced project and score the forecast against delivery

1. Build a bottom-up inventory of the largest identifiable supply projects in the world (they did a "top 50 projects," then a "top 75") with each one's expected volume and start-up date. This is the supply side from the companies' own guidance, not an agency aggregate.
2. Sum it into a growth forecast for the non-OPEC (i.e. non-discretionary) part of supply. They forecast 3%.
3. The following year, score it against what actually arrived. It came in at zero.
4. Repeat, and treat the second miss as the datapoint. One year's shortfall has an excuse (a Gulf of Mexico hurricane); a second, with a different excuse (a Nigerian disruption), is a pattern. "At some point you have to take responsibility that stuff happens in the world."
5. Now ask the structural question the misses imply: why can companies no longer hit guidance? Their answer — the easy exploitation of nearby fields from the 70s/80s boom "was starting to peter out."

• Two or more consecutive years of collective (not company-specific) production-guidance misses, each with a different excuse. That's the signature of a resource-quality problem masquerading as bad luck.
• Run the same screen today: he says explicitly he is not seeing it — "shale's going to grow at least 300,000 barrels a day this year… even I would have said probably flat at best."

27:302. Watch the 5-year forward, not front-month inventories

1. Identify what regime you are in. In a stable, low-price commodity market, near-term inventories are the whole game and everyone is trained to watch them.
2. When a structural repricing is underway, the front end can look soft — precisely because buyers anticipating scarcity are building stock, creating a visible near-term surplus. The two effects look identical on an inventory chart and mean opposite things.
3. So move your attention to the long end. The 5-year forward is where a genuine change in the cost of supply shows up, because it prices the marginal barrel nobody has drilled yet.
4. Expect to be told the long end is "just speculation." Test it with a cost/returns study rather than a flow study: if the capital intensity of the industry is genuinely rising, a higher forward price is a cost signal, not a positioning signal.

• A widening gap between a soft front end and a rising 5-year forward while commentary explains it away as speculative flow.
• The inverse today: he anchors on the long-dated curve staying in a band as the reason Super-Vol, not Super-Spike, is the regime.

42:523. Returns on capital peak before the commodity does — the miss he still regrets

1. Separate the two calls. "Where does the commodity go?" and "should I be overweight the equities?" are different questions with different answers, and getting the first right does not carry the second.
2. Track sector return on capital against price, not earnings against price. Early in a cycle they rise together. Watch for the point where price keeps rising and returns flatten.
3. That flattening is cost inflation and capital intensity eating the price — capex per barrel, service costs, project scope. It is the top for the equities even while the commodity still has years to run.
4. When the two diverge, downgrade the equities. Do not let the correctness of the commodity call keep you overweight the stocks.

• Sector ROCE flat or falling for two years while the commodity price makes new highs — with capex per unit and service cost inflation as the mechanism to confirm it.
• The tell that the excess return has been competed away: incremental projects earning the same return as the trough-era ones.

44:524. Let someone else run your own numbers to their conclusion

1. Publish the underlying data behind a call, not just the conclusion — the returns series, the project inventory, the cost build-up.
2. Actively solicit what your best-informed readers conclude from your own numbers, especially when their conclusion contradicts your recommendation.
3. When a serious counterparty says "based on your analysis, the opposite follows," treat that as the highest-value feedback available — they have already accepted your data, so the disagreement is purely about inference.

• Any reader who accepts your data and derives the opposite recommendation — that is a structural error in your inference, not a difference of opinion.
• The reverse tell of consensus capture: everyone agreeing with the conclusion while nobody has interrogated the series.

46:105. The rock-star tell — treat your own celebrity as a top signal

1. Track the social reception of your call alongside the fundamentals: how you are introduced, how much media wants you, whether the call has a nickname.
2. When the analyst becomes the story — billed as an act rather than a source — the trade is crowded. That is a sentiment reading available to you and nobody else.
3. Act on it as a reason to check the position's asymmetry, not necessarily to reverse. The commodity call may still be structurally right while the near-term price is done.

• A thesis acquiring a brand name, a roadshow and an audience — versus the same thesis being ignored, which is where the money is made.
• The corollary: having missed a rollover, immediately shift the work to where does it trough and how fast does it come back, which is the part they did get right.

1:18:486. The two-trigger test — pre-commit to what would change the regime call

1. State the current regime as a falsifiable claim. Here: Super-Vol — ~1 mb/d of oil demand growth, comfortably met by available supply, so volatility rather than a price level.
2. Name the specific, measurable conditions that would flip it, in advance, with the data source you will use.
3. Trigger 1 (demand): global GDP growth back above 4% — which he ties mechanically to oil demand growth of 1.5 mb/d. Source stated up front: Goldman Sachs' economic forecasts ("just so people know the data source I'm using").
4. Trigger 2 (supply): a genuine supply disappointment "like we did again 25 years ago" — i.e. the top-projects screen from insight 1 failing again.
5. Require both. Demand surprise without supply disappointment is met by shale/Canada/LatAm/Middle East; supply disappointment without demand growth is absorbed.

• Global GDP forecasts crossing 4%; oil demand growth prints trending above 1 mb/d toward 1.5.
• 2027 capital budgets being reset at $70–80 (the shale supply response that would confirm Super-Vol rather than break it).

56:267. When the expected spike doesn't show, look one link down the chain

1. When a widely-forecast price move fails to appear in the headline instrument, don't conclude the thesis was wrong — ask which part of the value chain absorbed it.
2. For oil the chain is crude → refining → product. Check the crack spread (the refiner's margin: product price minus crude price) against its own history before concluding "no spike."
3. Benchmark against normal: Gulf Coast 3-2-1 WTI cracks normally run ~$20/bbl, ~$30 in a bull market.
4. Check the single tightest product separately — diesel cracks tell you about industrial and freight demand and about which refineries are actually down.
5. Then attribute: which physical assets were hit, and is the damage repairable or structural (a segment nobody wants to add capacity to is structurally tight).

• Crack spreads at multiples of normal while crude is merely firm — the disruption premium has relocated downstream.
• Segments that are both attackable by drones/standoff weapons and politically impossible to expand: that combination is where a temporary premium becomes a structural one.

57:338. Decompose the price outcome into actors — and flag the unsustainable one

1. Ask why the price is where it is rather than where the model said, and attribute the gap to named actors with numbers attached.
2. Separate the moves everyone expected (Saudi redirecting exports through the east–west pipeline to Yanbu, ~5–6 mb/d; US and Japanese SPR draws; a bit more shale) from the one nobody modelled.
3. For the unexpected actor, ask the durability question: is it drawing a finite stock? A finite buffer means the force reverses, and its reversal is a coiled bid.
4. Refuse to time it. Name the direction with confidence and the timing with none.

• China's monthly crude import run-rate turning back toward 12 mb/d — the buffer exhausting and the bid returning.
• Any actor absorbing or releasing several mb/d that is absent from the consensus balance; and separately, the US SPR at a multi-decade low with 4 mb/d of Canadian heavy as the un-hedged structural dependency.

53:079. Buy the sector that already earns in a hostile policy regime

1. Find a sector whose profitability held up while policy was actively against it. Surviving hostility is the stress test; you get the result for free.
2. Underwrite it on that hostile-regime profitability — no policy improvement in the numbers.
3. Then treat any policy course-correction as unpriced optionality rather than the thesis. Watch what changes in the talk first (rhetoric, alliances joined or left), since that leads the permits.
4. Separate the analytic claim from the political one explicitly, so the call survives disagreement about the politics.

• Pipelines actually reaching FID and construction (talk converting to permits), and westbound routes that don't depend on which US president is in office.
• Canadian LNG export projects being announced/sanctioned — the single largest re-rating lever for Canadian gas.

1:07:4410. Test any policy narrative against revealed behaviour, not stated priorities

1. Grant the narrative its strongest form — assume the stated priority really is the number-one issue. Then ask what any country actually optimises for.
2. His empirical claim: "no person, no country is going to model their economy on a CO2 basis. That it is not the organizing principle for anything." The revealed hierarchy is reliability, availability, abundance — then affordability.
3. Follow the arithmetic of the poor-country case, which is where narratives break. India uses 1.4 barrels per person and imports 5 mb/d; at a rich-country 10 barrels per person that is 44.5 mb/d of imports. "There's no chance India is ever going to want to import 44.5 million barrels a day."
4. Draw the two-sided conclusion the arithmetic forces, not the one your side wants: that number is exactly why EVs, LNG trucking and domestic battery/lithium/cobalt/copper supply chains get built — and why oil demand doesn't peak.
5. Distrust the vocabulary. "All these terms like green, brown, clean, dirty — total nonsense terms, and it leads to really bad policy decisions."

• Policy that survives a price shock versus policy abandoned at the first one — that's the test of whether a stated priority is real.
• Import-dependence arithmetic for large developing economies as the leading indicator of where non-oil energy investment actually goes.