00:29 1. Grade your annual themes mid-cycle — with an "incomplete" grade
The repeatable method
- Pull the original written forecast (not your memory of it) and walk every theme in its original order.
- Grade each one: right, partial credit, wrong — and incomplete when you never did the work or the event hasn't resolved. "Incomplete" keeps an unexamined call from quietly being counted as right.
- For each partial/incomplete, write the specific follow-up owed (a deep dive, a revisit date).
Here: mega themes and "deaths exaggerated" graded right; geopolitics and environment partial credit; nuclear ("we still haven't") and gas M&A ("hasn't happened yet") incomplete; industrial AI "directionally correct but a little bit of an incomplete."
Watch for
- Themes you mentioned but never analyzed — they are the ones you'll later claim credit for without having earned it.
04:35 2. Separate "right for the reason" from "right by luck"
The repeatable method
- When the outcome matches the call, ask what caused it. If an unforecast event (a war) delivered the outcome, the call is only credited to the extent the stated reasoning was independent of that event.
- Restate the original reasoning (e.g. sentiment too bearish = value opportunity) and check it would have held without the shock.
- Then roll the same reasoning forward to the next consensus extreme.
Here: no 2026 oil glut — "the Iran War happened. We didn't expect it" — but he credits the sentiment call ("people are way too bearish on oil"), and applies it forward to the new 4–6 mb/d 2027 glut forecasts.
Watch for
- Agency/bank oversupply forecasts extrapolating the next resolution of a supply shock — the next consensus extreme to test.
12:02 3. Check for asymmetric skepticism across parallel narratives
The repeatable method
- List the consensus narratives you pushed back on, then list the structurally identical ones in adjacent markets.
- If you fought one and let its twin pass, that's a blind spot — the same flawed logic was likely wrong in both places.
Here: he fought the oil glut but "did not push back as hard on LNG glut as we did on oil glut"; likewise "the peak global gas demand call has been as bad as the peak oil demand call. It's even, frankly, more ridiculous."
Watch for
- A "peak demand" or "glut" narrative in one commodity when you've already rejected it for a sibling commodity.
17:05 4. Find the through-theme — the factor that shows up in every mega theme
The repeatable method
- Write your mega themes side by side and ask which single input is required by all of them.
- If it isn't itself a named theme, promote it — it's likely the most under-emphasized exposure in your framework.
- Test with a field signal: how often does it come up in real client/industry meetings?
Here: natural gas sits under Power Surge, energy-rich, AI/energy convergence and Geopolitical Super Vol — "it just comes up in every meeting we have" — yet was never a named theme.
Watch for
- An input repeatedly cited as the enabler across theses that has no dedicated coverage of its own.
18:01 5. Pick the value-chain link, not the obvious commodity producer
The repeatable method
- When a commodity theme is right, map the chain: producers, midstream/pipelines, downstream/LNG, services.
- Ask where abundant, low-cost supply leaves pricing power: an abundant resource tends to cap producer returns and reward the bottleneck/infrastructure links.
- Look for adjacent business models incumbents can extend into (e.g. services companies into distributed power).
Here: shale gas producers have been "one of the least exciting ways to play natural gas"; midstream and downstream (LNG, pipelines) were the huge opportunity; oil services led the move into distributed power generation, more than upstream.
Watch for
- A commodity bull case where the resource is abundant and cheap — look past the producers to the transport/processing links.