Actionable insights — "I'm insanely bullish" on the long bond

3:29 1. The saturation fade — count the covers, then take the other side

1. Treat mainstream coverage as data, not background. Count the concrete artefacts: front-page headlines, magazine covers, op-eds by famous investors, podcast episodes on the theme.
2. Set a bar in units per week, not vibes. Four magazine covers in one week on the same theme, or a wire service running the same story on page one "every single day," is the reading he acts on.
3. Check who is making the argument. When people who are expert in the subject but not in markets start forecasting market outcomes — an AI podcast predicting "double digits or maybe even triple digits" bond yields — that is the sentiment extreme, not the analysis.
4. Take the other side of the saturated view, then look for the data that supports the trade rather than the other way round.
5. Do not convert a sentiment extreme into a timing call. Sentiment sizes the trade; the chart still governs the timing.

• A theme appearing on more than one general-interest cover in a week; domain experts from outside finance making price forecasts; a famous investor publishing in a non-financial outlet.

2:08 2. Answer a supply panic with the demand side of the ledger

1. When a consensus bear case rests on supply, notice that supply is the observable half — "it's very easy to measure the supply of bonds, but nobody ever talks about the demand for bonds." The unobservable half is where the mispricing lives.
2. Normalise the scary absolute number. A $2 trillion deficit is frightening; 6% of GDP is not, especially against 12% in 2010.
3. Find the historical period with worse supply and check whether the market actually failed to clear. In 2010 "people showed up at the auctions. The auctions had bid to covers of three or higher."
4. Ask what state of the world creates the missing demand, and how likely it is. Here: "if stocks were down 20%, trust me, interest rates would be much lower… People would show up and buy bonds." The asset is its own hedge, which is the demand nobody models.
5. Do the same for the new competing supply rather than ignoring it — private-sector issuance now crowds alongside the government, so size the trade for a slower path, not a straight line.

• Deficit-to-GDP versus prior periods that cleared; auction bid-to-cover ratios; any risk-off catalyst that would summon reflexive demand.

1:18 3. The dashboard mismatch — price one variable, check all the others

1. Write down what the market is currently pricing as a probability, not as a narrative (here: "a 66% chance of a rate hike").
2. Pull the full macro dashboard for the same week — payrolls, PMI, JOLTS, ISM, the inflation trend — and grade each print against expectations rather than against its own history.
3. If the priced variable points one way and every other series points the other, the price is the anomaly. "I don't understand the obsession currently when the rest of the economic data is actually terrible."
4. Weight direction over level on inflation: above target for five years matters less than "it continues to come down."
5. Name the single dated print that would resolve the mismatch, and pre-commit to what it means — do not re-decide on the day.

• Rate-path pricing that contradicts the growth dashboard; a single scheduled release that would force the repricing.

7:51 4. The manual chart sweep — sort every chart into topping or basing, then count

1. Pull the charts for a defined universe — the top 50 names in the index, or all 500 if you have the time. Do it by hand, one at a time; this is a Lehman-era ritual he ran with "a glass of scotch."
2. Classify each chart into exactly two buckets: topping (a long advance flattening and rolling over) and bottoming/basing (a long decline flattening out).
3. Roll the classifications up to sector level first — the sector verdict is more reliable than any single name.
4. Count the ratio. The output that matters is not the best chart but the balance: "I'm seeing a lot more charts that are rolling over than charts that are basing."
5. Keep individual exceptions as exceptions. A base inside a topping sector is a stock that already had its crash, not evidence the sector has turned.
6. Cross-check against flow: a sector that is receiving rotation money while the leadership is sold is a late-cycle destination, not an early one.

• The rolling-over-to-basing ratio across the universe each month; sectors that catch a bid immediately after a leadership sell-off.

10:58 5. The second-derivative-of-growth top signal — wait for deceleration, not valuation

1. Accept that in a momentum leader the multiple carries no information — "I don't really think of things in terms of fundamentals."
2. Track the growth rate itself over successive prints and watch the change in it: 70% → 60% → 50%. The level of growth is not the signal; the deceleration is.
3. Treat the first decelerating print as the trigger, not the first disappointing one. Beats can continue all the way through a rolling second derivative.
4. Note the base-rate context: the leader already decelerated once (200% → 50–60% a few years ago) without topping, so the signal is the next leg down in the growth rate, off the currently guided level.
5. Be patient with it. He has been waiting for this specific signal "for the last six months" without acting on price.

• Guided growth rates stepping down between quarters in the AI leadership; the gap between guided and realised growth narrowing.

9:41 6. Map the owner, not just the position — and let the data correct you

1. Before judging crowding, split the holder base: retail, hedge funds, institutions. "You have to distinguish between people like you and me… and people who work at hedge funds versus your average retail people."
2. Sample the holder base directly where you can. He teaches college students, so he asks to see their portfolios — one was 50% Nvidia, 50% Broadcom and nothing else.
3. Ask the behavioural question, not the valuation one: how will this holder act into a drawdown? "They're probably not going to sell at the highs."
4. Then check the anecdote against real positioning data before trading on it — and change the answer if the data disagrees.
5. Re-assign the crowding to whichever cohort the data actually implicates, and re-derive the risk from there.

• Retail vs institutional positioning series diverging; concentration in the portfolios of people you can actually survey.

13:32 7. Deleveraging marks a bottom — the sequencing rule that stops you calling the bottom

1. Treat a forced liquidation as a bullish event in the short run: ownership transfers from a seller who has no choice to buyers who do. Farley's formulation — "the owners go from weaker hands to less weak hands."
2. Confirm the transfer actually cleared: who took the block, and have they since worked out of it? A cleanup print that is still sitting on a dealer's book has not cleared.
3. Then apply the sequencing rule. "The most leverage player gets taken out first" — the first casualty is a ranking, not a conclusion about total leverage.
4. Test it against the 2008 template: Bear Stearns blew up March 17th 2008, the S&P rallied 17% over three months, "everybody thought the coast was clear," and then Lehman happened.
5. So trade the bounce, but keep the position and the horizon consistent with another blow-up: "my guess is there's another Situational Awareness coming in the months down the line."

• Block-trade announcements and who absorbed them; the second fund in trouble, which is the one that tests the low.

21:53 8. Invest, then investigate — buy the unfamiliar theme before you understand it

1. When a genuinely new theme crosses your desk for the first time, take a starter position before doing the work: "the first time you hear about something, you should buy the stock and then you do your research."
2. The rule exists to beat a specific failure mode, not because research is worthless: "people say, 'Oh, it's a baldness drug. All right, I'm going to research that. I'll get to that later.' And then they never do. And then the stock's up 200% and they miss the whole trade."
3. Size it as a research option, not a conviction position — small enough that being wrong on an unresearched idea is survivable.
4. Be explicit about the epistemic status when you pass it on: "I don't know anything about it. Literally I just saw a tweet and I put it in the newsletter… Research this and maybe it turns into something."
5. Prefer themes with a mechanism you can state in one sentence — a cancer drug that uses the immune system; a known compound at a dose previously blocked by a side effect.

• A theme that sounds faintly ridiculous but has a one-sentence mechanism; the moment you catch yourself deferring the research.

23:49 9. Read the capital structure of a boom, not its multiple

1. Ask how the build-out is being financed — equity or debt. That single question separates a bubble that deflates from one that breaks: "the leverage is what gets people into trouble."
2. Match the funding tenor against the asset's useful life. AI compute is "going to be obsolete in a couple years," so ten- or thirty-year debt commits the borrower long after the asset stops earning. Equity carries no such schedule.
3. Compute the actual coupon, not the spread. A tight 60–80bp spread over Treasuries still means "essentially paying a 6% coupon" when the underlying yield is high — versus "2% or 2 and a half %" in 2021.
4. Compare against the prior cycle: "the dot-com bubble was all equity. Nobody was issuing debt and this time we have a lot of debt."
5. Do not expect the cost of capital to act as a brake — "are the CEOs going to say, oh my god, the 30-year Treasury just hit 6%, stop capex? It's not going to happen." The constraint shows up in credit, later.
6. Follow the linkage outward. If the boom is debt-financed, everything else levered to the same credit is one trade: "AI, private credit, something else — I think it's all connected."

• Hyperscaler and neocloud issuance calendars; the effective coupon rather than the spread; debt maturities long relative to asset life.

25:36 10. Design the portfolio around the drawdown you can actually sit through

1. Start from the behavioural constraint, not the return target. A drawdown is not a number — "drawdowns affect your psychology," you stay miserable "until you get back up to the high water mark," and there is "a decent chance that you're just going to tap out and sell."
2. Recognise the real cost of tapping out: "that's the worst thing you can possibly do because then you stop compounding." The strategy that fails is the one you abandon.
3. Measure your own tolerance honestly. Dillian's is explicit — he could not work at a pod shop "because I sustain drawdowns larger than 5%" — and he will not hold a life savings in something "moving around 9% a day."
4. Diversify across asset classes, not across stocks: stocks, bonds, gold, cash and real estate, equal-weighted.
5. Price the trade-off before you accept it: roughly 1–2 percentage points of annual return given up, volatility cut in half, worst drawdown 12% against about 40% for the index in a calendar year.
6. State the honest counterfactual rather than overselling: "if you buy the S&P 500, you will have more money when you retire… That's if you can hang on."

• Your own realised behaviour in the last drawdown, not your stated tolerance; daily-move volatility of the portfolio versus what you check your balance for.

29:39 11. Public markets reprice in weeks; private markets take years — set the clock accordingly

1. When bearish on a private asset class, separate the direction from the timetable. Both were right here, but only one was fast.
2. The mechanism is liquidity, not sentiment: "in the public markets, when something unwinds, there is liquidity. You can sell… and the market will reprice very quickly. In the private markets, that doesn't happen."
3. Look for the symptom of a market that cannot clear — assets held rather than sold: "portfolio companies not being sold for a really long time."
4. Do not treat headline negativity as capitulation. Daily "private credit doom" articles and a wave of listed alternative-asset stocks trading below their level two years ago are not the same as marks resetting.
5. Express the view where it can actually reprice — the listed proxies — and hold the private-market call open for years: "a bear market in the privates is just going to take a much longer time to play out, but we have not found the bottom yet."

• Exit and secondary-sale volumes rather than headlines; the gap between listed alt-manager prices and reported private marks; any AI credit event, since he treats the two as one trade.