Actionable insights — Record Diesel Margins and the Refining Squeeze

1. When a commodity's price chops but its derivatives trend, own the spread, not the raw material

1. Split the chain into its stages — raw material, conversion, distribution — and chart each one's price separately rather than treating "energy," "grains" or "metals" as one exposure.
2. Find the stage where price is trending while the others chop. Directionless raw material plus rising output price means the bottleneck is conversion capacity, and the conversion margin is the tradable variable.
3. Express the view through the owner of the conversion asset (the refiner, smelter, crusher, fabricator), whose earnings are the spread, rather than through the producer, whose earnings are the raw price.
4. Verify with the margin series itself — the crack/processing spread — and check whether it is making records rather than merely being high.

• 3-2-1 and distillate crack spreads vs the crude flat price; utilization rates at the conversion stage; whether the pure-play converter is outperforming the integrated producer (the tell that the margin, not the barrel, is the driver).

2. Stack every supply withdrawal in one ledger before judging how long a shortage lasts

1. List each independent hit to supply separately — physical destruction, chokepoint closure, export policy, voluntary domestic-first diversion — instead of collapsing them into "the war."
2. Tag each with a restoration timeline: physical damage (years), chokepoint (a political event), export bans (a stated expiry), domestic diversion (reverses when the domestic market loosens).
3. Count how many must reverse simultaneously for the shortage to clear. A shortage caused by four independent withdrawals is far more durable than one caused by a single event, even if each individual item looks reversible.
4. Then ask who absorbs the residual demand — the supplier of last resort — and whether that supplier has physical headroom left.

• Russian refinery repair/restart announcements and the export-ban expiry date; Chinese product-export quota releases; Hormuz transit counts; weekly U.S. product export volumes against the record — headroom exhausted is what keeps the spread bid.

3. Ask which side of the chokepoint the disruption actually lands on

1. For any geopolitical supply shock, do not stop at the headline commodity — trace whether the disruption removes raw supply, conversion capacity, or both, because they have different substitutes and different repair times.
2. Raw supply usually has spare capacity somewhere and strategic reserves; conversion capacity typically has neither.
3. Whichever side has fewer substitutes takes the larger price move — position there, even when the news flow is entirely about the other side.
4. Sanity-check by asking what a release of the constrained input would actually accomplish: if freed raw material can't be processed, the bottleneck is confirmed downstream.

• Whether product inventories fall faster than crude inventories; SPR-style reserves existing for the raw material but not the product; refinery outage trackers vs production shut-in estimates.

4. Date the relief — put a calendar on when new capacity actually arrives

1. For a capacity-driven margin story, find the industry's net capacity trajectory (additions minus closures), not just the announced additions.
2. Take the first year net capacity turns positive as the horizon of the trade — long-lead assets have public construction schedules, so this is knowable rather than guessed.
3. Set that against the demand trend on the same timeline. Slowing-but-positive demand against flat-to-shrinking capacity is a structurally widening gap, not a spike.
4. Convert the answer into a holding period, and stop treating the position as open-ended: the trade has a stated expiry.

• Refinery closure announcements (a closure extends the trade); FIDs and construction starts on 2028–2030 projects slipping or accelerating; the same analyst's fleet-capacity table updated next year.

5. Test the "structural decline" story against the actual substitution rate

1. Whenever an industry is priced for terminal decline, separate direction from rate: the substitute may be winning while still growing too slowly to balance the market.
2. Compare the annual demand erosion from the substitute against the annual supply erosion from closures. If supply is leaving faster than demand, the "declining" industry has expanding margins.
3. Treat the market's willingness to fund new capacity as the key variable — a peak-demand narrative starves the industry of capital, which is exactly what creates the shortage.
4. Re-run the comparison annually; the trade ends when the substitution rate finally overtakes the closure rate.

• EV share of new sales vs total fleet turnover; diesel (not gasoline) demand specifically, since freight and agriculture electrify slowest; whether capital finally returns to greenfield refining — the signal the shortage is being solved.

6. Check the buffer before sizing the shock — inventories decide whether it becomes a price event

1. Before assuming a supply disruption moves price, look at days-of-cover in storage: a well-stocked market absorbs the hit as a drawdown, a lean one prices it immediately.
2. Prefer product inventories over headline crude inventories when the shortage is downstream, and check whether a strategic reserve even exists for that product.
3. Add the seasonal demand calendar on top — a shock landing just before a seasonal peak compounds rather than averages out.
4. Low inventories plus an imminent demand peak is the setup that produces the outlier print; treat it as a timing signal, not just a fundamental one.

• Weekly distillate inventories vs the five-year range; days-of-supply cover; the harvest and winter-heating demand windows; any restocking that would signal the squeeze easing.

7. Choose pure-play versus integrated according to which end of the spread you want

1. When a margin widens, list the candidate ways to own it and mark each one's internal hedge: a pure converter has none, an integrated owns both the input and the conversion.
2. An integrated captures less of the widening (its upstream sells the cheap input to its own downstream) — but it is also protected when the spread normalizes.
3. Check whether the same event that widens the margin also damages the integrated's other segment; if so, the dilution is worse than the simple mix suggests.
4. Pick deliberately: maximum torque to the spread (pure-play) or a smoothed participation with downside protection (integrated) — do not treat them as interchangeable sector exposure.

• Refining as a share of each company's earnings; distillate-weighted yield by refiner (the more diesel per barrel, the more torque); shut-in disclosures at the integrateds; the pure/integrated relative performance as a live read on where the market thinks the spread goes next.

8. Name the single event that ends the trade, and monitor it explicitly

1. For any position that exists because of a disruption, write down the specific resolution that would remove it — not "things normalize," but the named event.
2. If there are multiple independent causes, require all of them to resolve; a partial resolution usually produces a headline selloff, not a fundamental one, and can be the better entry.
3. Separate the structural residue from the event: capacity permanently destroyed or never built does not come back with a ceasefire, so decide what the margin looks like on the other side of peace.
4. Keep the trigger list short and observable, and check it on a schedule rather than reacting to each headline.

• Hormuz reopening and Iran negotiation milestones; a Russia–Ukraine settlement plus the lifting of Russia's export restrictions; the gap between the headline reaction and the unchanged capacity math — that gap is where the structural leg of the trade survives.