Actionable insights — The AI Trade, Rising Rates, and Why the Market Is Still Standing

5:13 1. Judge a correction by the internals, not by the leaders' drawdown

1. When the market's most-watched stocks fall hard, resist grading the event by their decline. Measure instead what the rest of the index did while they fell.
2. Use one simple breadth series and compare it at two index highs, not two dates: the percentage of members above their 200-day moving average. "You only had about 50% of the S&P above the 200-day average on June 2nd. Today… 75% of the S&P above the 200-day."
3. Read the direction as the verdict. Breadth that improves through a leadership drawdown means money is rotating, not leaving: "even as the market has churned for the last 8 weeks, the internals have gotten better, not worse."
4. Sanity-check it against the historical version of the same drawdown, so you know what the leaders' fall would have cost in a genuine risk-off: "go back to '99 or 2000. If I told you the leading stocks, Cisco or Sun Micro, were going to be down 40 or 50% over an 8-week period, what do you think the S&P or the Nasdaq would be down? A lot."
5. Convert the reading into a position statement rather than a forecast: "Money doesn't want to leave the asset class of equities… we're in this remarkably rotational tape."
6. The corollary to keep honest: this test tells you whether the selling is broad, and nothing about whether the thesis is right. It buys time; it does not settle the argument.

• Breadth deteriorating with the leaders (the genuine risk-off signature); a new index high on falling participation; the reverse setup where breadth improves but the equal-weight index still can't make a high.

6:14 2. Find the rate that actually competes with equities — and treat it as a moving threshold

1. Ask the only question that dates the end of an equity cycle: at what yield does the marginal owner prefer bonds? "I don't think we have yet to find the rate of interest that gets money to leave equities."
2. Write your own number down, then hold it accountable to the tape. Strategas' was 4.50% — "in 23, 24, 25, it was largely 450… but now it seems higher."
3. Distinguish a level that causes a correction from one that ends the cycle: "you got corrective periods from when we hit 450 in the past, but if we're really going to end the cycle, it's a level much higher than people think. It's not 470."
4. Calibrate with the bubble-era analogues rather than the recent average, because the recent average is what you already priced: Japan '89 (JGBs "four to eight that year as the Nikkei was melting up"), Nasdaq '99 ("US 10s went four to seven"), '87 ("long rates went from six to nine percent while the stock market was up 30%").
5. Anchor it to nominal growth, not to the last cycle's comfort zone: "with nominal at six and a half, maybe it's a level higher than people think."
6. Then invert the comfort. A market that ignores rising long rates is not strong, it is late: "that's the worst sign, when stocks are going straight up, and they don't care at the same time long-term interest rates are going up, and the stock market seems impervious. That's the point at which you have the most risk."

• Your own threshold being revised upward twice without evidence (the tell that it has become a rationalisation); nominal GDP growth rolling over while yields stay up; long rates rising on an equity up day, repeatedly — the '87 signature.

3:30 3. The buggy-whip audit — a correct thesis expressed as a levered pair is one position, not two

1. Take any long/short book and ask what would have to be true for both legs to lose at once. If the answer is "a change in sentiment about the theme," you own one position, not a hedged pair.
2. Run the deliberately extreme version to expose it: "imagine it's 1900 and you're very bullish on autos taking over the world. So, you buy every auto and auto parts company that's public and you short every buggy whip company that's public."
3. Grant that the thesis is right — that is the point of the exercise. "Given what we know about history, obviously that trade is going to be correct."
4. Then add the two ingredients that kill it: leverage and one bad headline. "But if you're four times levered, and let's imagine there's a bad auto accident… the problem is you're long X and you're short Y, but it's the same thing."
5. Size to the drawdown you can finance, not the outcome you expect. Being eventually right does not survive a margin call — "as South Park said in the classic Margaritaville episode, and it's gone."
6. Diagnose a blowup by the same test before accepting the market's explanation. A large fund dying in a calm tape — "the VIX below 20 and… a 15, 20 basis point increase in long-term interest rates" — is evidence of a levered same-thing pair, not of hidden macro stress.

• A hedge whose two legs share a single narrative driver; leverage rising while realised volatility falls; a blowup that the market explains with a macro move too small to have caused it (always the sign of leverage, not information).

23:57 4. Keep tracing the concentration until you reach whoever writes the cheque

1. Start with the index concentration everyone quotes, then refuse to stop there. Top-ten weight and sector weight are the first layer, not the answer.
2. Ask which companies produce the growth inside those weights, and count them. "The hyperscaler index — UBS has one — it's five companies… It's an index. It's five stocks."
3. Then ask who pays those five, and get a percentage of the specific revenue line: "something like 70% of the AI hyperscaler revenue is from just those two companies."
4. Assess the payer's solvency separately from the payee's stock price. Here both labs are "totally negative cash flow" and funded by capital raises, so their demand is a function of market conditions, not of their own earnings.
5. Add the competitive vector that could compress the payer before it fails outright — "the Chinese are competing with them."
6. State the conclusion as a conditional with an explicit unknown timing, which is what keeps it usable: "if this is the big if, if Open AI and Anthropic ever get in trouble, the ecosystem is in trouble."
7. Separate the layers when you rate the names: franchises with real (if capital-hungry) businesses at one end, pure derivatives of the two payers at the other.

• A funding round for either lab at a flat or lower valuation; token price cuts; enterprise migration to cheap open-weight models; any hyperscaler disclosing customer concentration in a filing; the moment either lab's financials become public.

25:29 5. Give a corrected group the benefit of the doubt — with a written exit condition

1. Identify the group that already took the damage. Not the theme's leaders, but the second-order names that fell hardest: "AI-adjacent or build-out type names… these probably down 30% from the highs."
2. Require a response, not just a bottom. The evidence is that they "have all responded over the last couple weeks" — the drawdown has been partly repaired on real buying.
3. Then write the falsifying condition down before you need it, in price terms you cannot argue with: "you have to give these the benefit of the doubt until they attempt to rally, don't make new highs, and fail. And we haven't seen that yet."
4. Note the structure of that rule: it is a three-part sequence (rally → failure to exceed the prior high → reversal), so a single down week does not trigger it and a genuine loss of leadership does.
5. Prefer the group where the earnings are contracted rather than priced — construction backlogs and equipment orders over commodity output — so the fundamental resolution arrives on a schedule.
6. Confirm with one hard print rather than the sector chart alone: an individual name up double digits on results is evidence the earnings, not the narrative, are moving.

• The exact failure pattern — a rally that stalls beneath the prior high and rolls; backlog growth decelerating; hyperscaler capex guidance trimmed rather than raised (the input the whole group's order book depends on).

18:32 6. For any capital-intensive growth story, score the cash flow and ask who funds the gap

1. Stop reading revenue growth as the health metric once a business turns capital-intensive. Put the whole peer group's free cash flow on one line and compare it to a year ago.
2. Read the direction, not the absolute: the best in the group being down 25% year over year is a more useful fact than any single company's level.
3. Ask the funding question explicitly, because it determines who bears the cost: "I don't see how you do it without hitting the bond and equity markets… it's not going to come out of cash flow."
4. Treat an equity raise by a historically self-funding company as a regime marker, not a transaction: raising $85bn "is shocking given the fact that the last time they actually raised equity capital was in the 2000s."
5. Track the depreciation ramp separately from the capex, because it is the part that hits reported earnings for years after the spending stops: "it went from like 4 billion to 6 billion, it'll go to 8, then 10, then 12. It's like a weight on your shoulders."
6. Compare expense growth against revenue growth as a single ratio and say plainly which way you want it: revenue +28% against expenses +55% — "we want the reverse."
7. Keep the conclusion proportionate. Capital intensity plus dilution is a reason to be market-weight and to prefer the suppliers; it is not, on its own, a short.

• Debt issuance replacing equity issuance (or vice versa) at the same company; depreciation growth outrunning revenue growth; a hyperscaler moving capex off balance sheet into leases or securitizations; the first raise that prices badly.

17:03 7. Read the reaction to your own research as a positioning indicator

1. When you publish a claim you consider uncontroversial, log the intensity of the pushback separately from its content.
2. Grade the claim first, honestly. Trennert's was deliberately mild — "I thought I had a pretty anodyne sentence… it's going to be hard to continue to grow AI investments without further dilution of shareholders."
3. Then grade the response. Losing a client over an anodyne sentence — "we got fired by a client… He said I don't want to read the work anymore" — is information about how one-sided the audience's book is, not about the analysis.
4. Confirm the direction of that book before drawing a conclusion: "the guy was long, clearly."
5. Use it as a crowding gauge with a known weakness: it tells you the consensus is tightly held, and says nothing about when it breaks. Pair it with a price-based trigger, never trade it alone.
6. Apply the same lens to the surrounding noise — Trennert extends it to the forward-guidance complaints, noting that the loudest objectors "have a big self-interest in the Fed giving forward guidance": media that "now [are] going to have to work a little harder," and hedge funds that "want a green light all the time."

• Cancellations, unsubscribes or abuse spiking on a mild note; sell-side analysts dropping coverage of a name after a downgrade; the opposite tell — a bearish note that draws no reaction at all, meaning the position has already been sold.

16:20 8. Attribute demand with the commodity complex — split the metals that answer to different buyers

1. When a macro debate has two candidate causes, look for a market where the two causes buy different things, and read the spread between them.
2. Split the industrial metals by end-buyer: iron ore is overwhelmingly Chinese construction and steel; copper, zinc and tin carry electrification, grid and electronics demand.
3. Take the readings at the same moment: "copper new high, zinc new high, tin new high, Chinese iron ore is collapsing right here." And a level check on the weak one: iron ore "made 52 week lows this week."
4. Then state the attribution the divergence forces: "the strength in the base metals that you would traditionally look to for China for some signal is not a China message here at all. It's AI and CapEx… this seems to be a China demand problem, not an AI CapEx problem."
5. Cross-check with a second complex that should confirm or refute it. Oil failing to rally on renewed hostilities — "the move… in crude was pretty tepid… here we are back under 80" — is consistent with weak Chinese demand rather than with a resolved conflict.
6. Carry the attribution into the equity call. If AI capex is the buyer, own the capex chain; if China were the buyer, the same tape would mean something entirely different.

• The divergence closing from the wrong end (copper joining iron ore's decline = AI capex slowing); iron ore turning up without copper (a China-only recovery); crude rallying on demand rather than on headlines.

39:34 9. Rank defensives name by name — and ask whether politics can take the profit

1. Refuse to buy "defensives" as a bloc. In a market with one dominant theme, the theme runs through every sector: "everywhere there's this AI theme that runs through so many of these."
2. Rank the sub-groups explicitly, worst to best, so the ranking is falsifiable later. Here: utilities avoid, staples bottoming-but-last, health care first.
3. Apply the anomaly test — the single most valuable step. Find the group that should be thriving in the current environment and check whether it is. If it isn't, something outside the fundamentals is setting the price.
4. Name that something before you act on it. Trennert's answer is political: as data centres bid up power, bills become visible and "the politics are starting to consume the utility sector."
5. Generalise it into a screen with a plain question: can this profit be taken away by a vote? "The politics in this country is increasingly getting populist… there is a reaction against big anything. Big media, big academia, big pharma, big banks."
6. Where the answer is yes, prefer the layer that gets paid on contracts rather than on a visible consumer price — equipment and construction over generation, suppliers over operators.
7. Distinguish "bottoming" from "leading" in your language and in your sizing: staples are "worth a look" but "I would hardly call the sector leader in the market."

• Rate-case filings, windfall-tax proposals, data-centre tariff riders and state moratoriums; a sector making new lows on unambiguously good fundamental news; the equipment/contractor layer starting to weaken too, which would mean the demand itself, not the politics, has turned.

41:07 10. Retire a view the tape has already discounted — and say out loud that you missed it

1. Separate "my reasons for disliking this were correct" from "the price still reflects them." The first can stay true while the second stops being.
2. Check whether the low is in before re-litigating the fundamentals: "these put major lows in… late last year early this year."
3. If it is, name the failure mode in yourself rather than in the market: "I got too dogmatic on those and I missed it. We try not to do that."
4. Say it publicly. An analyst who logs the miss keeps the process auditable; one who quietly drops the name preserves the habit that caused it.
5. Then re-rank rather than reversing wholesale. He is constructive on managed care and says the actual leadership is elsewhere — "the pharma and the biotechs I think are where the leadership is right now."
6. Keep the surviving structural objection attached to the new stance, so the position stays informed: government's expanding role "worked for them for many years [and] it's obviously working against them now."

• A name you have disliked for years that stops making new lows on bad news; your own arguments getting longer rather than shorter over time; a group where every reason to avoid it is now common knowledge.

37:02 11. Decompose an index's new high before treating it as diversification

1. Never take a foreign index's record as evidence of a foreign economy. Ask which groups produced it and how many names they represent: "the Euro Stoxx is at a new high right now… but it's much more focused into a few groups or names."
2. List what is not participating, because that is where the local economy shows up: "the European auto industry is in a multi-year bear market. European luxury remains in a multi-year bear market" — the two sectors "that rely on Chinese demand."
3. Then name the drivers and check whether they duplicate the exposure you already own: Schneider Electric ("AI buildout"), Siemens Energy ("turbines"), "plus banks."
4. Draw the conclusion for allocation: an index high built on the same theme as your home market is not diversification, it is the same trade in another currency.
5. Apply the cheapness test separately, and demand a reason for it: "they're cheap for a reason… they're not dynamic economies. They don't have AI… they basically are just trying to stay in the game." Corroborate with a hard series — "German GDP has basically not grown for the last four five years."
6. Contrast with the market that carries the themes and a structural change: Japan breaking out of deflation and enforcing return on equity by "delisting companies that don't make certain minimums in terms of price to book or return on equity."
7. Where a bond-market move is being read as a crisis, ask whether it is reflation or rupture, and set an equity tripwire rather than a yield level: "until the Japanese insurance stocks or… the Japanese bank stocks start to really weaken here, I'm not concerned about higher JGB yields."

• Japanese bank and insurer share prices (the tripwire that would convert higher JGB yields from reflation to stress); European autos/luxury turning, which would mean the high has actually broadened; the Tokyo delisting rules being softened, which would remove the governance catalyst.

42:44 12. Put a scary market move in its distribution before accepting the narrative

1. When a move is described as "explosive," stop and measure it against its own history rather than against how it feels.
2. Choose a fixed window with a neutral start date — here, a political one that nobody chose for its convenience: "the 400 days since Trump was inaugurated."
3. Measure range, not level: "this is the narrowest range in 10-year yields that we've ever seen. The range of the 10 years 85 basis points over the last 400 days."
4. State the gap between the data and the perception explicitly, because that gap is the tradable part: "it's been a remarkably narrow range. It doesn't feel like it for some reason."
5. Convert it into a requirement rather than a prediction: "I just don't think this move in yield is as explosive yet as it's going to need to be if it's going to disrupt the equity market."
6. Hold that alongside the opposite risk from insight 2 — a narrow range is not safety, it is a compressed spring, and the '87 analogue is what its release looks like.

• The 400-day range starting to widen (the regime change, before the level matters); realised bond volatility rising while the level stays flat; equity multiples compressing on a yield move small enough that the narrative can't explain it.