Actionable insights — "You're making a mistake" if you own no commodities

1:54 1. Read the positioning, not the headline — a top-decile CFTC short is fuel

1. When a market refuses to respond to obviously bullish news, stop arguing with the news and go look at who is already positioned. Pull the CFTC Commitments of Traders report for the commodity.
2. Rank the speculative net short against its own history, not against zero. A reading "probably in the top 10% than it's ever been" is the signal; a merely large number is not.
3. Convert it to a physical quantity so the size is intuitive — here ~480 million barrels — and ask what has to be bought back if the crowd is wrong.
4. Identify the narrative holding the position together ("it's going right back to 50 or 60"). The trade is against the narrative, not against the price.
5. State it as an asymmetric setup, not a forecast: "I'm not making a prediction on it, by the way. I'm just saying that's the setup that you have right here."
6. Cross-check the physical side for whether the shorts have a cushion: strategic reserve levels (US and Chinese), import behaviour, refinery crude-slate compatibility. "Nothing happens until it's too late, usually."

• Weekly CFTC net-spec positioning at multi-year extremes in any commodity where the physical story contradicts the paper positioning; SPR refill/drawdown data; a resumption of Chinese crude imports.

5:54 2. Own the whole spectrum instead of picking the winning segment

1. When asked which sub-sector benefits most from a macro shift, refuse the question: "I think you have to own the whole spectrum."
2. Map the value chain end to end and take a position in each link — upstream large-cap, upstream mid-cap, midstream/transport, the commodity variant (gas vs oil), and services/equipment.
3. Underweight, don't exclude, the link whose economics you like least ("we're not as big on the refiners, but you can probably own those as well").
4. Size each by the clarity of its earnings, not by its beta to the underlying commodity — you are buying cash flows the market has mispriced, not a leveraged commodity proxy.

• Any sector where you can name a macro driver but not the specific beneficiary — that's the signal to buy the chain rather than guess the link.

12:15 3. Earnings-visibility screen — pay 8× for profits you can see six-to-eight quarters out

1. Ask first whether you can genuinely forecast the company's earnings for the next six to eight quarters. If you can't see them, the multiple doesn't matter.
2. Compare that visible earnings stream's multiple with the index's: 8–9× against a 25× S&P and a 42 CAPE. The conclusion is a re-rating claim — "this multiple ought to be higher" — not a growth claim.
3. Apply the same screen across sectors, not just within one: the test produced energy producers, gold miners, and tungsten/copper/iron names on identical arithmetic.
4. Explicitly reject the beta framing. Asked whether he owns resource stocks as leverage to the commodity, the answer is no — "we like them because of the earnings look."
5. Discount reported index earnings before using them as the comparator (see insight 6) — the denominator on the expensive side is probably worse than it looks.

• Single-digit-multiple businesses with contracted or commodity-linked revenue you can model; the spread between that multiple and the index multiple widening rather than closing.

30:15 4. Sell into the parabola — the momentum-buyer tell

1. Set a magnitude trigger, not a price target: when a commodity position roughly doubles or better in twelve months (silver +215%), the size of the move itself obliges you to act. "When you have something move like that, you have to take it, because that's a commodity."
2. Identify who the marginal buyer has become. Momentum funds "always come in late and they're real hot money, and when they leave, they just sell at any price" — a late influx of that money is the exit signal.
3. Size the downside you're protecting against, then act on it: "if you think they're going to correct 40 or 50%… as a money manager, you've got to take some of that off" (32:52).
4. Differentiate by asset type: sell all of the pure commodity (silver), sell only portions of the operating businesses (gold miners) that still generate earnings.
5. Plan the re-entry before you sell. The position is not closed, it is parked — re-establish the same portfolio weight after the correction.

• A 12-month commodity move above ~100%; retail/momentum inflow data arriving in the final months of a run; the same names correcting 35–40% as the re-entry window.

32:52 5. "Pay the tax" — the partial-exit framework for low-basis winners

1. Treat the tax bill as a cost of doing business, not a veto: "You need to pay some tax, believe me… if you're making money, you're going to pay some tax." Accept that clients (or you) will be irritated.
2. Don't binary the decision. The standard action is a partial exit — "take your cost out and take some profit out" — leaving the position alive but the original capital recovered.
3. Weigh the tax against the time value of money, not against the gain: the loss from a decade dead in the water, adjusted for inflation, dwarfs a capital-gains rate.
4. Run the historical stress test on the specific holding: the late-'80s-to-late-'90s drug run (Merck, Pfizer et al.) where holders who wouldn't trim "had to go another 23 or 24 years before you ever got back to those prices again" (34:35).
5. Allow one exemption: an investor at 80–85 whose heirs receive a stepped-up cost basis. At 50 or 60, "need to take a look at it."
6. Baseline holding period stays long — 3 to 10 years — and only "a massive move that actually made the stock overvalued" shortens it (36:15).

• Any position where the reason you're still holding is the tax bill rather than the forward return; concentration created by a single decade-long winner.

20:39 6. Discount the reported number — three ways this cycle's earnings are flattered

1. Before using index earnings as your valuation denominator, list the non-recurring and non-cash items inflating them.
2. Private-investment marks: large companies "can have these private investments and they don't have to mark to market, and they carry them as a profit."
3. Legislative accounting changes: "there's three or four things that went into that one big business bill that came out that adjust those earnings."
4. One-time cash windfalls: tariff refunds now being distributed after the Supreme Court ruling get booked as income — "it's a one-time throw. You say that won't happen again."
5. Deliberately do not trade the windfall itself ("we don't go in and look at companies that are going to get the most money"); use it only to mark down the quality of the aggregate number.
6. Conclusion to carry into every multiple you compute: "when you shake it all out, they won't be near as good as what people think they are by just looking at raw numbers."

• Level-3 asset marks in megacap filings; the earnings impact of the tariff-refund distribution over the next two quarters; any legislated change to depreciation or expensing.

18:50 7. The CAPE>35 base rate as a portfolio governor

1. Track the Shiller CAPE (10-year inflation-adjusted P/E) and its direction — "it's been up 4 months in a row. It's 42 now."
2. Apply the historical base rate rather than a market call: "the history of that when that ratio is over 35 — you don't make any money the next 5 years."
3. Stack it with independent sentiment gauges (BofA's indicator, Market Vane, record margin debt) so you're not relying on valuation alone.
4. Read record margin debt structurally, as a percentage of gross market value, not in dollar terms — "it's indicative of where you are in a bull market at the end of it" (9:45).
5. Treat a V-shaped, reflexive dip-buy rally as confirmation of late-cycle behaviour rather than a momentum entry: "it just feeds on itself… someday you buy the dip and it doesn't work."
6. Don't short it — shift the mix. The base rate governs how much you own of the expensive index, not whether you're bearish.

• CAPE printing consecutive monthly gains above 35; margin debt as a share of market cap making new highs; the first dip that doesn't get bought.

25:07 8. The dark-fibre test — diagnosing an un-investable capex boom

1. When a build-out is described as a "new paradigm" that "everybody had to get into," reach for the closest capex analogue — for him, the 1999–2000 fibre-optic build-out.
2. Ask the utilisation question, not the demand question: two-and-a-half to three years after the build, will buyers say "we'll never use all of this"? Fibre demand did eventually arrive; the shareholders of the builders still lost.
3. If you cannot name where the returns land, refuse the theme outright: "Not for us because we don't know where it's headed… there's not enough for us to want to go in and really play that game."
4. Instead, buy the physical inputs the build-out must consume regardless of which operator wins — copper, aluminium, iron, uranium, natural gas.
5. Find the real bottleneck rather than the advertised one. In Texas the constraint is not power but water: "we have the energy. We don't really have the water, honestly… in West Texas and South Texas, you don't have enough water" (23:20).
6. Price the political layer: a construction moratorium standing two and a half months before a state election "probably all changes after the election."

• Announced capacity versus plausible utilisation 3 years out; water rights and grid interconnect queues as the real gating item; post-election reversals of local construction moratoria.

28:42 9. The domestic supply-deficit screen — buy what the state will eventually have to stockpile

1. For each strategic material, compute the raw ratio of domestic consumption to domestic production. Uranium: ~50 million pounds used against ~2.5 million produced.
2. Rank by how badly the gap must be closed and how long closing it takes. "We're so far behind on uranium, it's incredible… we could really get in a mess with uranium."
3. Check who holds the surplus. If it's the strategic rival — China and Russia hold much of the critical-minerals supply — the gap is a national-security problem and gets policy money.
4. Front-run the government rather than follow it: the federal "vault" for critical minerals is new; "we've been earlier than they are."
5. Note the exception that proves the screen: natural gas is "the only one we are not behind" on — which is why gas producers are held for earnings, while the deficit metals are held for scarcity.
6. Hold through the boring years. These were bought long before the theme was popular — "it's not something new to us really."

• USGS/EIA consumption-vs-production ratios for the listed critical minerals; announcements of federal stockpile purchases as confirmation you were early rather than as an entry signal.

11:17 10. Leverage only when the cash yield swamps the borrowing cost

1. Default to none. Oxbow has used margin once in decades — at the '08/early-'09 low, about 20%, and not for every client.
2. The single condition: current cash flow — "dividends or interest," not expected appreciation — must be far above the financing rate. There, 15–17% yields financed at 6–7%: "then you just made a deal."
3. Because the test is cash-on-cash, it only fires in genuine dislocations, which is the point — it is a crisis-entry tool disguised as a leverage rule.
4. Invert it as a risk gauge for the market: when aggregate margin debt is at a record and the borrowing is against appreciation rather than yield, that's late-cycle behaviour, not opportunity (9:45).
5. Same discipline on the other side of the balance sheet: never lend long to a borrower you don't trust. "I don't think anybody should be looking at 30-year paper" for the next five years — stay at the short end, where a two-year Treasury at 4.25% also avoids state tax (14:59).

• Dislocations where quality yields exceed financing costs by 8–10 points; the 30-year Treasury approaching 5.5%; Treasury's issuance mix drifting ever shorter as a sign the long end can't absorb supply.