Actionable insights — The Bond Market's Biggest Contrarian Trade

3:17 1. The two-to-three-unconnected-PMs shift — his primary entry signal

1. Maintain a roster of specialists by vertical (commodities, high yield, financials, fixed income) — people deep in one thing, not generalists.
2. Track each one's direction of change, not their current position. A permabear turning bearish is noise; a two-year bull turning bearish is data.
3. Require independence: the managers must not know each other or share a source. Correlated opinions are one opinion.
4. Fire when two to three of them shift the same way inside a short window — "that's what gets us excited." Triangulate via ideas dinners, the live chat and bull-vs-bear cage matches, so both sides of every idea are stress-tested before it is traded.

• Long-standing bulls in any vertical capitulating, and unconnected specialists arriving at the same new position within weeks of each other.

32:28 2. The "everyone knows the bear case" screen — quantify how much is priced

1. Find an asset where the negative story is universally repeated and effortless to recite. Universality is the setup, not the verdict.
2. Measure the crowd instead of arguing with it: CFTC futures positioning plus any bulls/bears survey. Compare to the same asset's own extreme in the other direction (bonds now vs 2017–21, when "everyone was max long" and TLT was one of the most popular ETFs).
3. Put a number on how much is discounted — his estimate here is "80 to 90% priced in" — and accept that the last 10–20% can still hurt: "I think it could go a little bit worse."
4. Then hunt specifically for the bull case nobody is working on. "Everyone knows the bear case. Nobody's thinking about the bull case. And that's what great investors do."
5. Size and phrase accordingly — "start at least thinking about buying," i.e. begin the position, do not swing it.

• CFTC net positioning at multi-year extremes; a bear thesis so well-rehearsed that the marginal seller has already sold.

28:55 3. The supernova sequence — treat an inflation spike as the recession trigger

1. Reject the reflex that "inflation up = own less duration." Run the 1980s–90s sequence instead: hot economy → commodity-led inflation → long yields spike and bond portfolios are destroyed → the inflation wounds the consumer → recession arrives fast.
2. Check the consumer's shock absorber. If the bottom 70% is already "decimated with higher interest rates, higher inflation," the energy shock has nothing left to absorb it and the lag between spike and recession collapses.
3. Date the spike from physical inputs you can observe before they hit CPI: diesel breaking out, corn and wheat breaking out, a shipping strait effectively closed for ~200 days. He gives a lead time — "it's going to start playing out in the next couple of months," across "the next three CPI prints."
4. Position for the second leg while the first leg is still doing the damage: buy the asset the inflation spike is destroying, because that same spike causes the recession that then re-rates it.

• Diesel and agricultural breakouts leading CPI by ~2–3 months; consensus earnings expectations at multi-year highs as the complacency confirm.

30:11 4. The convexity math — price the recovery before you buy a discounted long bond

1. Find long-dated bonds trading far below par from good credits, and note the coupon (a Google issue at 88 on a 6⅛ coupon; an Apple bond near 49; an Oracle bond paying 8%).
2. Do the two-scenario arithmetic explicitly: if the 30-year goes to 3–3.5%, that 88 bond prints "maybe 120 or 130"; bonds in the 80s "could go back to 105 to 110." Then price the downside — one more punch takes 88 to ~85.
3. Take the trade only when the payoff is that lopsided — "the convexity of the situation is like a rubber band that's pulled." Asymmetry, not direction, is the entry condition.
4. Check currency before you assume the exposure: he flags the Google issue as a sterling bond paying in sterling. A great rate view expressed in the wrong currency is a different trade.

• Investment-grade long bonds in the 80s or lower; the gap between the downside (a few points) and the recession upside (20–40 points).

4:43 5. The cheap-hedge sizing rule — spend 1–3% of gains, not conviction

1. Trigger on the price of insurance, not on a forecast: check whether options/vol are cheap relative to the identifiable catalysts ahead. "You can buy equity volatility very cheap relative to that kind of risk."
2. Fund it from the year's gains: "if you're up 18% on the year, 16%, you take 1% of your gains and buy protection." Scale to 1–3% of the portfolio when the book is already long hard assets and growth.
3. Pick the instrument by holding period. Long-dated options (one-year puts on XLF or BAC) for the thesis; short-dated vol funds only as a rental.
4. Alternative if none of it appeals: simply raise cash by the same 1–3%.

• Sector vol at multi-year lows while positioning is one-sided; a dated catalyst (a CPI run, an election) the option can be sized to cover.

5:32 6. Rent, don't marry — the decay rule for levered and inverse ETFs

1. Before buying any inverse or levered ETF, ask who rebalances it and how often. "When you buy an ETF that is short something, the portfolio manager has to rebalance that every day."
2. Treat the daily reset as a known, recurring cost — "there's a real decay" — so the position must have an expiry date attached at purchase.
3. Set that window from the catalyst, not from the chart: here, September and October specifically.
4. If the view needs to be held longer than the window, switch instruments — long-dated puts have a fixed, up-front cost and no daily bleed.

• Any position in a daily-reset product outliving its catalyst window — that is when the decay quietly takes the gain.

11:47 7. Read the credit sandwich — stress at both ends while equities sit at highs

1. Build the sandwich and check all four layers at once: investment grade at the top (LQD vs the S&P), the loan market and CCCs at the bottom, and the private-credit vehicles in the middle (BDCs like Blue Owl, private equity like KKR).
2. Use relative charts, not absolute yields: LQD against the S&P 500, KKR against the financial sector. The divergence is the message — "if you pull up a chart of KKR versus the financials, it smells to high heaven."
3. Require confirmation from both ends. Top-tier stress from new-issue supply plus bottom-tier stress from deteriorating credit is a different (and worse) signal than either alone.
4. Calibrate the stage rather than calling the crash: "I'm not saying it's late 2007… it's definitely late 2006 type dynamic" — early enough to hedge cheaply, too early to expect the blow-up.

• CCC spreads widening while indexes make highs; IG underperforming equities on supply; BDC and private-equity charts decoupling from the banks.

8:18 8. The cash-burn turnaround date — the one number that prices the AI credit

1. For each hyperscaler, find the year the capex burn is projected to flip to positive free cash flow. That single date is what the off-balance-sheet loans are underwritten against.
2. Test the loan against it: banks lend against double-A-rated future free cash flow, so the credit is only as good as that date. "If that's in 2030, 29, okay. If that's in 2031, 32, 33, Houston, we have a problem."
3. Track the size and location of the financing, not just the spend: Meta from $30B off-balance-sheet "up to 600 billion" in 2–3 years; hyperscalers issuing "an extra 5, 600, 700 billion of debt over 12 to 24 months."
4. Read the lenders' own hedging as the verdict: banks "aggressively buying the credit default swaps" on the Mag 7 they just financed is an insider signal that the date is slipping.
5. Watch for style drift as the enabling condition — banks stretching outside their mandate to keep marquee clients (SpaceX, OpenAI, the Mag 7) is how the exposure got built (10:21).

• Any slippage in guided free-cash-flow crossover years; CDS volumes on Mag-7 names; new off-balance-sheet vehicles announced alongside capex raises.

23:41 9. Corporate issuance as a rates variable — count the paper competing with the Treasury

1. Stop treating the long end as a pure macro/inflation instrument. Add up the corporate long-dated supply and compare it with annual 10- and 30-year Treasury issuance: the hyperscalers are now "a big threat to long-term issuance."
2. Separate the two drivers of a yield move — inflation expectations versus term premium/supply. Here inflation expectations were falling while yields rose, which points at supply and politics, not at CPI.
3. Follow the issuer of last resort. Watch Treasury's response function: threatening to issue fewer 10s and 20s, hinting at buying them back (an operation twist), playing with the yen, backstopping the Bank of Japan.
4. Include the political supply premium — French and UK elections, DSA House/Senate seat counts — since "global yields breaking out" is transmitted straight into US bonds.

• Quarterly refunding announcements and buyback hints; hyperscaler bond calendars; DSA seat projections and French/Japanese long-yield breakouts.

38:52 10. Hedge the winners instead of selling them

1. Score the multi-year theme on realised results rather than narrative: coal +32%, copper names +82%, gold miners +50% over the year, against a Mag-7 ETF up 4–5%.
2. If the structural case is intact ("the picture the next five years is still phenomenal"), do not sell into it — the tax and re-entry cost of exiting a working theme is the real risk.
3. Instead, buy protection on the part of the market that would drag the winners down in a broad de-risking — here the crowded, record-priced financials — so the hedge is uncorrelated with the thesis.
4. Keep the hedge small (1–3%) and dated, so being wrong costs a rounding error against the gains it protects: "we've been really bullish on companies that own total hard assets… but now you want to protect those gains."

• A theme with a large embedded gain and an intact 5-year driver — the setup where hedging beats trimming.