← Analysis page  ·  Paul Sankey hub  ·  Research hub

Actionable insights — Global Diesel Shortage Gets Worse

The repeatable analysis behind the call. The point is not that Sankey is bullish oil and refiners but how he reads the market: products before crude, observed-history inventory floors, physical markers over screen prices, a demand-side collar, elasticity thresholds, and a seasonal calendar he's willing to flip.
2026-SEP-15 · David Lin Report · Paul Sankey with David Lin · ▶ Watch · full analysis · transcript
How to read this page: each insight is a method used or described in the conversation, written as steps you can rerun later in a different oil cycle. The boxed line shows how it played out here. Timestamps deep-link into the video.

13:00 1. Watch the product (diesel) inventory against its observed historical floor, not crude

The repeatable method
  1. Track weekly US distillate inventories against their full observed history, not only the 5-year average.
  2. When stocks reach the lowest level ever observed, treat it as "tank bottoms": the only way to learn how low inventory can go is observed history, and below it price, not volume, must clear the market.
  3. Check whether heating oil (distillate) is leading crude. If it is, the squeeze is in products and refiners capture it.
  4. Price the product in $/bbl alongside crude so the crack spread is visible.
Here: "US distillate inventory, I think, is at the bottom of the tank… We have never been lower. So, I think that's why diesel prices have gone exponential." Diesel "is $250 a barrel" against ~$110 Brent (31:43); "heating oil has been leading crude" (25:50). Conclusion: "very bullish for the likes of VLO."
Watch for

14:07 2. When utilization is near 100% into turnaround season, price the accident risk

The repeatable method
  1. Compare refinery utilization with the seasonal norm. Autumn is normally turnaround season, when runs fall.
  2. If runs are still at 97–98% with inventories at the floor, there is no buffer: a forced or unplanned outage becomes an asymmetric upside shock to product prices.
  3. Check the other buffers: hurricane season, winter weather outlook (El Niño), and the lead time on foreign product exports (crude purchase → refinery → export, ~50 days).
Here: "98% utilization and 17 million barrels a day… there's only really one way that number can go." Refiners "have to turn around. And if they don't, there'll be a risk of an accident"; "the cavalry… is going to be Chinese exports of products," still ~50 days away (15:01).
Watch for

11:20 3. Check physical markers against screen Brent to gauge real tightness

The repeatable method
  1. Pull dated Brent, Oman/Dubai and Shanghai crude alongside the Brent futures price on your screen.
  2. If the physical and regional grades trade well above futures, the physical market is tighter than the headline suggests.
  3. "The only good number in the oil market is the oil price": trust market action over official supply estimates or political reassurance.
  4. Cross-check official outage claims against independent cargo trackers.
Here: Oman priced a record $166 early in the crisis; "Shanghai oil is trading a good $10 above Brent"; dated Brent and Oman "are all quite a bit above where Brent's actually trading" (12:35). Wright's "few days" Saudi outage vs Kpler's "4 to 6 weeks" (20:49).
Watch for

5:34 4. Bracket the price with the marginal buyer's behaviour (the "China collar")

The repeatable method
  1. Identify the marginal buyer (here China) and estimate the price at which it steps up buying and restocks (floor) and where it backs off (cap).
  2. Estimate its inventory cushion, remembering underground strategic storage is invisible to satellite tank-lid counts.
  3. Map its marginal suppliers. If those are disrupted (Iran sanctioned, Saudi route shut), the cap is less reliable because the buyer fears for supply.
Here: "Once you get much below $80 a barrel, Brent, China starts buying more oil… the upper end of that band… around 100 and now you're not far off." China's marginal barrel "comes from Iran," and Saudi replacement barrels are now disrupted, so "I'm not sure they're going to stop buying oil at $107" (39:47).
Watch for

29:55 5. Find the demand-destruction level from a historical retail-price step, then translate to crude

The repeatable method
  1. Check actual price elasticity now: are gasoline, jet and diesel volumes falling despite high prices? If not, prices can keep rising.
  2. Look for non-discretionary demand (harvest, AI construction, travel as an "experience" staple) that won't respond to price.
  3. Pick the retail price at which demand historically stepped down (here ~$4.50 gasoline), compute the % move from today, and translate it into a crude price range as the practical ceiling.
Here: jet fuel prices doubled while "jet demand is up 2%" (27:28); "$4.50 gasoline and you see a step down in demand… another 20% higher… around $120, $130 a barrel Brent for a demand destruction number."
Watch for

15:42 6. Keep a seasonal oil calendar, and flip it when inventories must be rebuilt

The repeatable method
  1. Default rule: short oil around Labor Day (driving season over) and buy at the first New York snow (~December 5).
  2. Override it when inventories are at the floor going into the winter build season: go long through December 5 instead.
  3. Pair it with a relative call (oil equities vs the S&P) while oil's share of GDP is rising, and set a fixed review date (six or seven weeks).
  4. At the end date, check the weather: if winter doesn't show up, sell.
Here: "Normally, we would short oil Labor Day and buy it at the first snow in New York… This year, I think what you do is you long oil from here through to December the 5th"; "I expect the oils to continue outperforming versus the S&P for a good six or seven more weeks here and then we'll take another look" (16:04).
Watch for

16:48 7. Overlay the political response on the windfall winners: export bans and windfall grabs

The repeatable method
  1. When pump prices hit records before an election, list the interventions on the table (product export ban, windfall tax, emergency restarts via the Defense Production Act).
  2. Map who inside the administration opposes each one, and which companies have access.
  3. Watch how the most-exposed stock trades on each headline; an intraday sell-off on "export ban" wires is a live probability read.
Here: an export ban "would crater the price of gasoline and diesel but… cause havoc globally," "somewhat irresistible… into the midterms"; VLO "was trading off earlier on the idea that there would be an export ban"; "Burgum and Wright oppose it" and CVX "has the ear of the president" (18:44).
Watch for

21:19 8. Price a prediction market by reading the contract, then the counterparties' incentives

The repeatable method
  1. Read the resolution terms first. What exactly counts ("binding agreement… enforceable upstream participation right… before January")?
  2. For each named party, ask what signing would cost it: outstanding legal claims, strategic focus (gas vs oil), management's history of holding positions.
  3. Go long the party with the least to give up and short the one with a claim it won't abandon.
Here: Shell at 62% is "quite rightly priced," though "Shell's pretty much into the gas"; Exxon "don't back off their legal position… they'll just wait for the Democrats," and ConocoPhillips' award is bigger. "I would probably be long SHEL short XOM right there" (23:13).
Watch for

Methods distilled from the public YouTube video (transcript in transcript.html) for personal study. Not investment advice. © David Lin Report / Sankey Research for source material.