Actionable insights — Markets after Nvidia and Warsh: Your Next Move

1. Use the catalyst-failure test: when the best available news lands and the stock doesn't move, the trade is a positioning problem, not a fundamental one (Nvidia's Impact on Momentum and Market Outlook)

1. Before the event, write down the outcome that would qualify as unambiguously good — a specific revenue number, a guidance raise, a policy tone. Do this in advance so you cannot rationalise afterwards.
2. Let the event happen. Score the outcome against what you wrote, ignoring the price.
3. Now look at the price. Good outcome + no rally is the signal; it means the marginal buyer is already fully invested and the marginal seller is selling for a reason unrelated to the news.
4. Diagnose the seller. If the fundamentals are intact, the seller is almost always mechanical — a factor unwind, risk-limit reduction, or index/flow effect. Identify which, because that tells you how long it runs.
5. Treat the failed catalyst as confirmation to reduce, not as a dip to buy — and note that everyone else running the same test gets the same answer, which is what makes the unwind self-reinforcing.
6. Set the invalidation: the trade is over when the same group of names starts rising on no news, which is the mirror image of this test.

• A pre-written definition of "good news" so the test is honest; whether the non-reaction repeats on the next scheduled catalyst (Broadcom Wednesday) — one failure is noise, two is a regime; and the reverse signal, the group rising on no news, as the all-clear.

2. Measure the factor, not the story — check whether "AI is breaking" is really "momentum is unwinding" (Nvidia's Impact on Momentum and Market Outlook)

1. When a group of unrelated-looking stocks falls together, do not reach for a thematic explanation first. Ask what they share as a factor — momentum, high beta, low profitability, crowded ownership.
2. Go and price the factor directly (the momentum index versus the S&P over the same window) rather than eyeballing individual charts.
3. Put the number in historical context — percentile, or "worst month in N years" — so you can tell an ordinary drawdown from a genuine regime break.
4. If the factor's move is historically extreme while the underlying businesses are fine, you are looking at a flow event with a fundamental cover story, and the right response is to manage exposure, not to change your view on the industry.
5. Then check where the money went. Factor money rarely leaves the market; it relocates. Find the receiving sector, because that is where the next trend starts.

• Momentum-index performance versus the S&P on a rolling quarter; the semis-versus-software spread since the end of June; whether the receiving sector's strength is broad (an ETF making highs) or two stocks carrying an average.

3. Before rotating, run the fundability test on the destination — can that sector actually meet the four-quarter earnings bar? (Political Noise and Positioning in the AI Trade)

1. Write down the rotation you are proposing as a pair: what you sell, and precisely which sectors receive the money.
2. For each destination, pull the consensus earnings expectations four quarters forward — not the current multiple, not the yield. The rotation only works if those numbers get met.
3. Ask the historical question: when this sector has previously been assigned high forward expectations, did it deliver? Look up the base rate rather than assuming.
4. Eliminate the destinations that are secretly the same trade. A sector that has been re-rated by the theme you are leaving is not a diversification (industrials carrying AI-infrastructure demand is still an AI position).
5. What survives is usually a much smaller set than the pitch implied — size accordingly, and consider intra-sector rotation (moving within a sector) instead of a cross-index move, which keeps you in the sectors that historically do meet high expectations.

• Four-quarter-forward consensus EPS revisions for the destination sector; that sector's historical hit rate against elevated expectations; index weight (can it absorb the flow?); and any destination whose recent re-rating traces back to the theme you are exiting.

4. Value software with the replicability test: "could a general-purpose model recreate this?" (Cybersecurity and Software Earnings: Palo Alto, DocuSign)

1. Take the product and describe, concretely, the job the customer hires it to do — not the category label.
2. Try to do that job with a general-purpose AI tool yourself. This is a real experiment, not a thought experiment; run it on something you actually need done.
3. Sort the result into one of two buckets. Replicable: the value was skill or content generation the model now supplies. Not replicable: the value is in things a model cannot conjure — counterparty networks, security and compliance posture, legal standing, and integrations into other companies' systems.
4. For the not-replicable names, check whether the market has applied the sector-wide AI-disruption discount anyway. That gap is the opportunity.
5. Confirm with cash, not narrative: free-cash-flow yield, earnings multiple, and whether the recent price move is company-specific or just the sector rotation carrying it.
6. For the replicable names, accept the honest conclusion — if you cannot forecast the revenue, you cannot value the stock, and standing aside beats a number you do not believe.

• Whether the moat is a network/integration/compliance moat or a skill moat; whether the multiple already embeds an AI-disruption discount on a name that does not deserve one; and, in the other direction, forecast dispersion widening on the replicable names — the tell that nobody can value them either.

5. Stress-test a hyper-grower on cash conversion, not on the growth rate (Political Noise and Positioning in the AI Trade)

1. For any company posting spectacular revenue growth, go past the income statement to the balance sheet and pull accounts receivable — sales billed but not yet collected in cash.
2. Express it as a share of the quarter's revenue and track the trend across several quarters. The level matters less than the direction.
3. If receivables are growing faster than revenue, ask why. Longer payment terms to win orders, customer financing, and customers who are themselves cash-constrained all produce the same signature.
4. Look at who the customers are. A rising receivable balance owed by the richest companies on earth demands a different explanation than one owed by start-ups — and if the obvious explanation ("they can easily pay") does not fit, that is the flag.
5. Do not convert the flag into a thesis on its own. Use it to set the question you re-check next quarter, and to size the position more conservatively while the answer is unknown.

• Receivables as a share of revenue quarter over quarter; days sales outstanding; any vendor financing or customer-funding arrangements disclosed in the filings; and whether cash flow from operations is keeping pace with reported net income.

6. Check a demand narrative against the physical pipeline, not the order book (Political Noise and Positioning in the AI Trade)

1. Take the forward demand claim underpinning a theme — announced capacity, approved projects, signed contracts — and find the number that describes how much of it has physically started.
2. The gap between "approved" and "under construction" is the honest measure of how much of the story can still slip. Announcements are free; ground-breaking is not.
3. Decide which way the gap cuts for your position. A big unbuilt backlog means the build-out still has years to run (bullish for suppliers), and simultaneously that the delivery date is soft (bearish for anyone paid on this year's revenue).
4. Layer the constraint that actually binds — permits, grid interconnection, turbines, transformers, local politics — rather than the one in the headlines.
5. Re-check the same figure each quarter: the conversion rate from approved to under-construction is the cleanest single indicator of whether a capex theme is real.

• The approved-versus-broken-ground conversion rate each quarter; utility rate cases and ratepayer-rebate announcements (the political pressure valve); interconnection queues and turbine lead times; and whether the backlash is coming from voters or from amplified bot activity.

7. Trade the gap between priced policy and expected policy, and name the sector that carries it (Political Noise and Positioning in the AI Trade)

1. Read the number of rate moves the market has priced for the next few meetings. This is data, not opinion — start there.
2. Write your own expectation, and be explicit that it differs. If it does not differ, there is no trade.
3. Identify which sectors are currently being pressured because of the priced path — the rate-sensitive ones: banks, small caps, anything with floating-rate debt or a duration-sensitive multiple.
4. If you think the priced path is too aggressive, the pressure in those sectors is your entry, not a warning: the rotation into them is the catalyst you were told does not exist.
5. Confirm the mechanism is live by watching the pressured index in real time (a small-cap index leading the downside proves the market is trading the priced path, not your view).
6. Define the invalidation up front: the meeting, and the actions that would prove the priced path right.

• The number of hikes priced across the next two or three meetings; small-cap and regional-bank relative performance as the real-time read on that pricing; the mid-month Fed meeting as the resolution date; and whether long yields fall with a growth scare rather than without one.

8. In a deal, price the seller — and time the M&A wave off the political calendar (Four Big Deals and Investment Committee's Final Picks)

1. When a takeover is announced, resist reacting to the buyer first. Write down both sides: who receives cash, and who takes on price, debt and integration risk.
2. For the seller, work out the realised gain against the original purchase — an owner crystallising a large profit is a clean, immediate, already-banked event.
3. For the buyer, size the deal against its own balance sheet. A very large all-cash purchase is "a big bite": the cost is certain and now, the benefit is uncertain and later.
4. Read the set of the day's deals thematically rather than one by one. Where deals cluster tells you which end-markets corporates believe in with their own capital.
5. Overlay the political calendar. Deal-making needs a permissive regulatory backdrop, so an approaching election that could produce divided government creates a deadline — and pulls transactions forward.
6. Express the wave through the fee/carry earners (alternative asset managers) rather than by guessing individual targets.

• Seller-side realised gains versus buyer-side deal-to-market-cap ratios; the sectors the deals cluster in (data-centre cooling, Permian/LNG, pharma diversification); announced-deal volume between now and the midterms; and alternative-manager share prices as the wave's direct expression.