Actionable insights — Called The Tech Wipeout, Now Reveals The Next Great Rotation

17:07 1. Read positioning before you read fundamentals

1. Before forming a view on a hot sector, ask who already owns it and how they own it. The measurable version: retail call-buying volume, and specifically concentrated call buying in a single theme.
2. When that hits record all-time highs, treat the marginal buyer as exhausted — the demand that would have arrived later has been pulled forward into options.
3. Confirm with a crowding survey ("most crowded trade in the world") and euphoria in commentary. Crowding plus euphoria is the rotation trigger; it does not need a fundamental catalyst.
4. Rerun the same read on the bounce afterwards: if the identical pattern reappears (leveraged ETFs, one-day options, calls on the same names), the punishment repeats.

• Record single-sector call volume; leveraged-ETF exposure to one theme as a share of assets; "most crowded trade" survey readings.

12:19 2. Ask "is it known?" before you react to good news

1. When a bullish print lands, don't ask whether it's good — ask whether it's new. The market is a discounting mechanism; only surprise moves price.
2. Check the expectations bar, not the result: where do revenue-growth and earnings expectations sit relative to their own history? A result that merely meets a record bar is not bullish information.
3. Find the last time the bar was that high and look at what happened next — that's your base rate, not the headline.

• Consensus revenue-growth expectations at multi-year highs; strong prints that fail to make new price highs.

13:37 3. Diagnose a rip: dead-cat bounce or new uptrend?

1. After a waterfall decline, test the bounce on structure, not size. Run the charts of the whole complex (he scanned ~50) and count how many have made a new high. A violent rally with no new highs is a counter-trend move.
2. Identify who is doing the buying. Short covering plus retail chasing with short-dated calls (which forces dealers to buy the underlying to hedge) manufactures a rally with no durable owner behind it.
3. Assume duration, not a day: after real institutional damage, waterfalls "persist for more than just a few weeks" and each dead-cat bounce is sold for weeks or months. Fade rips repeatedly rather than calling one bottom.
4. Overlay the calendar — seasonality (late August, pre-election) tells you roughly how long the bounce can run.

• Breadth of new highs within the bouncing group; short-dated option volume as a share of the rally; whether the prior lows hold on the next test.

18:07 4. Trace the redemption chain — turn fund losses into a selling calendar

1. Collect the monthly returns of the big levered funds in the crowded theme. A cluster of double-digit-to-catastrophic monthly losses is the input, not the conclusion.
2. Walk the plumbing forward: month-end loss → redemption requests on the first of the next month → the manager must raise cash over the following weeks/months.
3. Predict how they raise it: professionals do not sell in the hole and do not chase 30% rallies. They wait for a bounce and sell strength — so the supply arrives precisely when the tape looks best.
4. Size the problem: the publicly known casualties are the canaries; assume the actual list is much longer.
5. Position the other side — be a seller of the rally they need, and keep dry powder for the lower levels they're selling toward.

• Monthly hedge-fund performance tables after a drawdown; quarterly redemption notice dates; strength that arrives on falling volume in the most-owned names.

21:59 5. Screen on relative lows, and buy the barbell

1. Don't screen on absolute price. Chart each sector relative to the index and rank by how many years back you have to go to find a lower ratio. That isolates what has been abandoned to fund the crowded trade.
2. Set a bar high enough to matter: 25-year relative lows = generational; 10-year relative lows = still "backing up the truck" territory.
3. Buy a barbell rather than picking one macro outcome — a defensive leg (staples, healthcare) that works if the tape breaks, and an offensive leg (discretionary) that works if the consumer holds. Both legs qualify only because both are washed out.
4. Exclude the contaminated defensive: any "defensive" sector that has been re-rated as part of the crowded trade will fall with it. Screen out anything that traded up on the theme.
5. Cross-check with a demand test you can state plainly — e.g. "never bet against the US consumer when they have a job" at 4% unemployment.

• Sector-vs-index ratio charts at multi-decade lows; a defensive sector whose correlation has flipped to the growth theme (disqualify it).

27:06 6. Watch what they do — the buyback-in-the-hole confirmation

1. Ignore management's tone on the call; score their capital allocation. "Watch what people say, but watch what they do is more important."
2. Apply the test where it carries real information: a turnaround normally hoards every dollar for the fix, so a turnaround repurchasing shares while the stock is still down is management paying to express a view.
3. Use windfalls as the natural experiment. When an unexpected cash inflow arrives (a tariff refund, an asset sale, a legal settlement), watch where it goes — reinvestment, dividend, or buyback in the hole.
4. Run the mirror test on the crowded theme: insider selling vs buying, and whether historically dominant repurchasers have stopped. Companies that can no longer buy their own stock have lost the flow that supported it.

• Repurchase activity at beaten-down turnarounds after a cash windfall; buyback authorizations quietly lapsing in the crowded sector; insider transaction ratios.

28:53 7. Name the single indicator that decides the thesis — then ignore the rest

1. For any capex supercycle, write down the one measurable outcome the spending must eventually produce. Everything else — guidance, backlog, "insatiable demand" — is noise until that number moves.
2. For AI, that number is aggregate productivity growth. Track it monthly. If it isn't rising, there is no return on invested capital yet, regardless of what any company reports.
3. Frame the race explicitly: does the return arrive before the bill comes due? That single question converts a vague valuation debate into a timing question you can monitor.
4. Separate consumer benefit from investor return. In 2000 the internet delivered everything promised, but the payers (Global Crossing, MCI WorldCom) went broke while society kept the fibre. A great technology and a bad investment are fully compatible.
5. Sanity-check yourself, not the vendors: are you measurably more productive per minute, and has it raised your bottom line relative to what you've spent?

• Monthly productivity releases; enterprise token spend migrating to open-weight models; the free-cash-flow inflection date slipping further out.

35:08 8. The collateral test — "what exactly do you repossess?"

1. Whenever a new securitization appears, skip the structure and ask one question: if the borrower stops paying, what asset do you take back, and what is it worth then?
2. Classify the collateral. Productive and appreciating (real estate) survives a bad underwriting decision — you can still liquidate at a profit years later. Depreciating and technologically obsoleting (GPUs, five years used) does not.
3. Find the load-bearing assumption the sponsor must defend publicly. If the CEO keeps insisting the asset's useful life exceeds N years, the ratings — and therefore the whole deal — hinge on that claim.
4. Follow the incentives of every party: arrangers earn fees for brokering and distributing, rating agencies earn fees for blessing a new asset class, and the credit risk ends up with insurance companies and retail. Nobody at the table is holding the bag.
5. Note the urgency tells. Principals convening at an implausible time (mid-August, out of the Hamptons) signals the existing financing channel has run dry.
6. Set the horizon honestly: this is a "mañana problem" — useless as a 3-month trade, decisive as a 3-to-5-year risk. Log it and revisit rather than trading it.

• Rating-agency decisions on the first GPU-backed tranches; public arguments about asset useful life; insurance-company allocations to the new asset class.

15:18 9. Count the paper — issuance is a supply shock to the stock

1. Track the sector's net share count, not just its earnings. When free cash flow is exhausted, funding shifts from internal cash to debt, then to equity — a visible three-step deterioration.
2. Treat record equity issuance as a supply problem in its own right: "at some point when you keep issuing equity, the supply overwhelms demand," independent of business quality.
3. Judge each raise on outcome, not intent — if the short-term results already show a poor return, more capital raised for the same purpose is more value destroyed.

• Share-count growth across the theme; the buyback-to-issuance flip; each new mega-raise being upsized rather than trimmed.

46:40 10. The turnaround checklist — cash intact, new jockey, margin of safety

1. Survivability first: is the business still generating substantial free cash flow while the stock trades "like it's going out of business"? Cash generation is what lets you wait.
2. Return on invested capital: a real ROIC number (not a promise) means you get your money back without needing the story to work — "we don't have to worry about getting our money back like we do with some of the other trades."
3. Diagnose the damage: split temporary from permanent. Separate a demand pull-forward (COVID drinking), a scare with decaying evidence (85% of GLP-1 users quit within two years) and one fixable segment (US tequila) from a broken franchise.
4. New jockey: "the new jockey is critical in a turnaround" — require a CEO with a track record of the specific skill needed (cost discipline, brand management) demonstrated elsewhere, and from the highest-return part of the business where possible.
5. Margin of safety on a lower baseline: underwrite the stock assuming the new normal is permanently below the old peak. If it still works, size it.
6. Structural leadership: prefer the clear category leader (revenue multiples ahead of the runner-up) so the recovery accrues to your name.

• A CEO hire from a proven turnaround; capital reallocation toward the highest-ROIC segment; the loss-making segment inflecting to free-cash-flow positive.

45:20 11. The disintermediation filter and the "too hard" box

1. Sort every candidate by one question: is this business a beneficiary of AI or a target of it? A cheap price is irrelevant if the business can be disintermediated.
2. Where you cannot be confident either way, refuse the trade regardless of how attractive the free-cash-flow multiple looks — "I'm an investor, not a psychic." Capital impairment risk outranks apparent value.
3. Keep a formal "too hard" box so the decision is recorded rather than relitigated every time the price falls further. You may well be leaving a generational buy on the table; that is the price of certainty.
4. Positive test for the other side: the business survives because the thing it sells is human and physical (a drink with a friend, a day at a park), and AI plausibly lifts its margins instead of its competition.

• Evidence a software category's seat count or pricing is being replaced rather than augmented; conversely, AI showing up in a consumer name's margin line.

42:37 12. Pre-commit the re-entry to sentiment, not to a price target

1. Accept that being right about the technology says nothing about the path. Count the drawdowns inside the last comparable boom before setting expectations.
2. Define the buy trigger in sentiment terms you'll recognise: the consensus flips to "the AI trade is over, there's no return on investment," and the group is down another leg. Despondency, not a level.
3. Expect one more melt-up first — index targets rising, retail flushed out then coaxed back in — and treat that as the setup for the trapdoor, not a reason to chase.
4. Deploy into a pre-built long-term list so you're choosing from research done in calm conditions rather than shopping in a panic.

• Headlines declaring the theme dead; capitulation in the most-crowded names; your own long-term list being ready before the flush.