Actionable insights — a generational bear market ahead, gold in "the early innings"

3:23 1. Locate the cycle with Templeton's four stages — and check the stages in order

1. Start from the frame: bull markets "are born on pessimism, grow on skepticism, mature on optimism, and die on euphoria." The question is never "is this a bubble" but "which stage is complete."
2. Check the stages as a sequence, because they arrive in a fixed order. (a) Did the move start from a genuinely cheap valuation? (b) Have retail investors and individuals come in? (c) Has the supply side turned on — "a lot of IPOs and a lot of money starts being raised through debt offerings and stock offerings"?
3. Only when all three are ticked do you call the last stage: "Now it's all in place and so you certainly got all the euphoria there."
4. Convert the stage read into a duration expectation, not a timing call. Fourth year up plus euphoria means "I'd have to be surprised that you would do a fifth year" — while still allowing the melt-up to run ("a decent chance at 8,000 on the S&P").
5. Do not let the stage read drive the buying. He is explicit that Oxbow buys "mostly stocks from the bottom up"; the cycle read sets the cash weight, not the stock list.

• The offering wave specifically — IPO count, secondary issuance, convertible and high-yield issuance funding the theme. It is the last box to tick, and it is observable weekly.

12:55 2. Sell when the valuation no longer holds the price — staged, and tax-aware

1. Run the valuation numbers forward on the position you own and ask one question: do these multiples come back in? "If they do, that stock's going to be cheaper" — regardless of how good the business is.
2. Separate the company from the security. He calls Microsoft's business fine and sells it anyway: "for us it looks like the valuation doesn't hold the price. That's why we sell it."
3. Sell in tranches on strength, not in one decision. Oxbow had already sold "all the way up two or three or four times, just little bits at a time" before deciding to exit completely — each sale into a rally, never into a break.
4. Use a run-up as the execution window: the final decision landed "on this little runup we've had in the last month."
5. Treat the tax basis as a real constraint to be worked down, not an excuse to hold forever. Where accounts are so low-basis that "we can't sell all of it for taxes," the position is reduced over time — "we're kind of working through that."
6. Accept being early. The Intel case is his own evidence that a 100% further rise after your sale still beats a 26-year round trip (11:50).

• A long-held core position whose forward multiple now depends on the multiple staying put rather than on earnings growth; a strong month that gives you a clean tranche to sell into.

10:28 3. The earnings-quality test — back out the debt and the depreciation before you believe the multiple

1. Take the reported earnings of the companies carrying the theme, and refuse to use them as-is.
2. Back out the debt that has been "taken in" to fund the build — financing an asset base with borrowed money flatters current returns and defers the cost.
3. Back out the depreciation that has not yet landed: capacity built this year shows up as an expense over the next two or three. Ask what the P&L looks like once it does.
4. State the conclusion as a forward comparison, not a valuation opinion: "I don't think what you look at 24 months from now will be the same as it is today."
5. Apply the same discount to the index earnings you use as your comparator — an inflated denominator makes the whole market look cheaper than it is.

• Rising capex funded by new debt issuance while depreciation schedules lengthen; the gap between reported earnings and free cash flow widening across a whole sector at once.

10:53 4. The Cisco-1999 overbuild analogue — separate "the technology matters" from "these shares are worth it"

1. When a capex boom is justified by a genuinely transformative technology, stop arguing about the technology. Grant it entirely.
2. Ask instead the two questions that actually determine returns: how much capacity is being laid down relative to near-term usable demand, and how long does the resulting asset take to depreciate.
3. Use the 1999–2002 sequence as the template: routers and fibre were built for a paradigm that did arrive — "all that stuff got overbuilt and so you looked up in 2002 and all of a sudden… it's not happening like we thought it was."
4. Distinguish the winners from the thesis. "It ended up in different ways, not what everybody thought it was in '98 and '99" (46:40) — being right about AI does not tell you which names capture it.
5. If you cannot name the eventual beneficiary with confidence, decline the trade and buy the physical inputs the build-out consumes instead.

• Utilisation rates on newly built capacity; the first quarter in which capex guidance is cut rather than raised; contract lengths shortening on new data-centre leases.

22:28 5. Two index diagnostics: a rising CAPE and the top-50 share of value

1. Read the CAPE (the cyclically-adjusted P/E, which smooths ten years of earnings) as a level and a direction. 42 is "about as expensive as you ever get" — and it has "gone up every month for four months in a row."
2. Refuse to treat the level as a timing tool: "that doesn't say you couldn't get more expensive. You certainly could." It governs what you'll pay, not when you'll sell.
3. Measure real diversification separately from nominal diversification: "the top 50 stocks in the S&P are like almost 80% of the value. Well, you're not really as diversified as you think you are."
4. Apply a hard filter to yourself regardless of what the crowd pays: "everybody can be paying anything they want for anything, but if we look at it and we think it's stupid, we're not going to do it… if I can't buy the valuation, we're just not going to buy it."
5. Accept the tracking error that comes with it, and measure it honestly — "we haven't really been left that far behind."
6. Cross-check the same test on any crowded group by pulling every constituent's chart: if all of them are at new highs at once and the basket is up 20%, "usually that's a sign that you probably don't have a lot more to go" (9:42).

• The CAPE's monthly direction more than its level; the top-50 concentration ratio; the point at which a basket's advance stops needing individual earnings to explain it.

24:10 6. Hold liquidity as a paid option on lower prices — at the front end only

1. Set the cash weight from the cycle read, not from a forecast of the next move: 45–50% of the equity strategies in Treasuries.
2. Keep it at the very front of the curve — "short-term treasuries, less than two-year maturities" — so the money is genuinely available and carries no duration risk while you wait.
3. Require the liquidity to pay: "they've done well this year, actually." A cash position that yields nothing is a cost; one that yields is a financed option.
4. Invert the emotional framing of a decline. "We carry a lot of liquidity. So if we get a lot of cheap prices, we look at it as a positive because we're trying to buy things cheap" (6:33).
5. Deploy against your own valuation filter, not against a level in the index — the trigger is a name passing the test, not the S&P falling a set percentage.

• Front-end yields relative to expected equity returns; the moment your own screen starts producing more qualifying names than you have room for — that's the signal to spend the option.

26:49 7. Read a bottom by whether the hot money has been rung out — not by the chart

1. Count the retests. A level that has been hit "four or five times" over months is doing work: each visit removes another tranche of weak holders.
2. Survey what the commentators are saying, and look for unanimity rather than direction. When everyone writing about the asset says the same thing — "it'll have one more tick down," to $3,500–3,600 — the remaining sellers are already positioned.
3. Ask specifically whether the fast money has left, and over what window: "all hot money that had been in that stuff got rung out between really February and about six or eight weeks ago."
4. Buy on cheapness, not on the low. "I don't know if it's going to go $200 $300 more down or not, I just know it's cheap now."
5. Pre-commit to being wrong about the exact level: "maybe we go down make a new low. It's okay with us if we do because we think we got all this stuff at the right price… if we did we would probably buy some more of it."

• Consensus notes converging on one more leg down; repeated defended retests of the same level; momentum/CTA positioning flat-to-short while the physical story is unchanged.

30:41 8. Sell the blow-off in a commodity — completely — then plan the re-entry

1. Set the trigger on the magnitude of the move, not a price target. A commodity up 212% in a year has, by that fact alone, "gotten too much at one time. But it was time to go."
2. Sell all of a pure commodity position — unlike an operating business, there are no earnings to grow into the price.
3. Expect and ignore the criticism. "We got a lot of heat" is stated twice; the discipline is worthless if consensus approval is a precondition.
4. Re-enter on price, not on a bottom call, and size the re-entry below the original: silver is back in the book with only "three or four bucks, maybe five" of profit, and remains "not near as big in that as we are gold."
5. Keep adding while it's cheap — "added some more actually yesterday" — rather than waiting for confirmation.
6. Run the same loop on the equities, but in portions rather than in full: HL at $9 → sold above $30 → repurchased at $14.50–15 is the completed round trip.

• A twelve-month gain of 100%+ in a physical commodity; a parabolic final leg accompanied by mainstream coverage; your own reluctance to sell because the trend is working.

41:38 9. Judge a resource equity on profitability at the settling price, not on the headline commodity

1. Form a view of the settling range rather than a point forecast: "we felt all along that the price would basically settle in between 65 and 85."
2. Test company economics at the middle of that range, not at the peak: "the energy companies can make a lot of money at $75. A lot of money… they don't need a $100 oil to do that."
3. Identify the market's error explicitly — investors extrapolate the direction of the commodity rather than the level of profit. "They think, well, if the price has gone from 105 back down to 82… that just means it's all over. Well, not really."
4. Note that a falling commodity with unchanged company profits makes the equity cheaper, not worse: "that just makes the companies cheaper when you get right down to it."
5. Own the whole chain so the call doesn't depend on one link: producers, gas, midstream, services.
6. Name the one thing that actually breaks it — demand, not price. "If you go into a major recession, you don't use as much. That's just bottom line, that's the risk" (44:49).

• Breakeven and free-cash-flow disclosures at mid-range strip prices; a dividend yield that falls only because the share price rose; recession indicators in physical demand rather than in the futures curve.

36:34 10. The supply-deficit screen — domestic consumption divided by domestic production

1. For any critical material, put two numbers side by side: annual domestic consumption and annual domestic production. Uranium: ~50 million pounds used, "somewhere between two and a half and three" produced.
2. Look at the ratio, not the gap. A 15–20× shortfall is a structural condition, not a cycle — "you think about a demand supply curve that's out of balance."
3. Run the screen across the whole critical-minerals list rather than stopping at the first hit: "there's numerous things like that, by the way… a lot of the critical minerals are like that."
4. Overlay the geopolitical filter: in a multipolar world "everybody's hoarding their assets," and "some of the assets you want to own are probably other countries have more of them than we do" (35:10).
5. Apply the same test to what fails it. US farmland is rejected on price — Brazilian acreage costs 20% of the US equivalent — so "I don't look at that as a hard asset you want to own" (35:45).

• Any material where domestic use is a double-digit multiple of domestic output; export restrictions or stockpiling announcements from the producing countries; new domestic production that would close the ratio.

51:02 11. Balance the portfolio to buy optionality, not to hedge a forecast

1. Measure concentration at the asset-class level first, before looking at individual holdings: the problem he names is investors "90 95% in the market. They don't own anything else."
2. Reject the usual defence of a balanced portfolio (smoother returns) and use the real one: "your assets need to be balanced because then it gives you some options. If things were to go bad, you've got some options."
3. Note the asymmetry: "if you've got all of it in one thing, you don't have many options if that one thing goes bad." Balance is bought in advance because it cannot be bought during the event.
4. Understand what "balanced" means here — different asset classes (short Treasuries, metals, energy, hard assets), not different equities.
5. Set the horizon to match: hard assets are "a good mix to hold the next seven or eight years" (36:09), against 8–10 years of stagflation.

• Your own asset-class concentration measured honestly (an index fund plus target-date fund is one bet, not two); the moment a decline would force you to sell rather than buy — that is the missing optionality.