Actionable insights — Stocks Rise as Yields Fall: The Committee's Investment Strategy

1. Treat an intervention as a revealed pain threshold — then trade the level, not the news (Committee Debates Falling Yields)

1. When an authority acts unexpectedly, work backwards to the trigger: what number was printing the day before it moved? That number is the level it is defending.
2. Confirm the level is about the specific instrument, not the asset class — establish which point on the curve actually caused the discomfort.
3. Assume the defence persists while the motive persists, and date the motive (an election, a funding calendar, a policy review) rather than treating it as open-ended.
4. Separate what the authority can control from what it is merely leaning on. A buyback influences the long end; it does not set it.
5. Ask what the gesture's timing bought: acting into saturated one-way positioning produces far more price effect than the same action into a neutral tape.
6. Set an expiry on your own conviction — measure the half-life of the move rather than assuming a permanent regime change.

• The exact yield/price that printed immediately before an intervention; whether positioning was one-sided when it landed; how long the move holds before the level is retested.

2. Date the intervention window off the political calendar, and position inside it (Committee Debates Falling Yields)

1. Ask who is politically exposed to the variable being managed, and through which transmission channel (here: long yields → mortgage rates → housing → voters).
2. Put dates on it. Count the weeks from the intervention to the political event; that is the window in which the authority has maximum incentive to keep acting.
3. List the other variables the same actor is watching, because the same incentive applies to them (oil prices into the autumn, for instance).
4. Then treat the managed variables as bounded rather than random for the duration — which lets you underwrite the assets that were being penalised by them.
5. Mark the window's end as a scheduled re-underwriting date, not a permanent regime.

• The size and cadence of the buyback operations; mortgage-rate prints; oil into October; and what happens to the same variables once the window closes.

3. Convert a macro shock into a flow map before picking names (Committee Debates Falling Yields)

1. After a macro trigger, don't start with valuation — start by naming where capital is leaving and where it is arriving. Both halves, explicitly.
2. Describe the exits by factor as well as by sector (momentum, memory, semis), because factor unwinds hit names that look unrelated on a sector screen.
3. Verify the rotation is real and not a one-day reflex by checking that it "extends" across sessions.
4. Cross-check the destination sectors independently — do they have their own fundamental story, or are they just the residual?
5. Only then look for the individual names inside the receiving sectors.

• Whether the rotation extends beyond a single session; factor-level (not just sector-level) exits; whether the receiving sectors have independent catalysts.

4. Test whether a rotation will stick: does it need the other side to fall? (Healthcare's AI-Driven Growth Potential)

1. When money moves into a laggard sector, check the prior instance of the same rotation and what happened to it.
2. Ask the decisive question: is the destination rising because the source is falling, or are both rising? A rotation funded purely by a selloff usually reverses when the selloff stops.
3. Look for a durable narrative in the destination sector rather than a flow artefact — a structural reason it deserves capital.
4. Accept a long dead period as the cost of the trade, and say so out loud, because the entry has to be made before the sector works.
5. Add the sector-specific timing constraint that determines when it works (here: the cost of the enabling technology has to fall first).

• Whether both the source and destination sectors are rising; whether an earlier version of the same rotation failed; the cost curve of the enabling input.

5. Strip the mega-cap skew out of any index earnings number before you use it (Broad Earnings Growth)

1. Take the headline index earnings-growth figure and ask how much of it comes from a handful of names. Back that portion out explicitly and restate the number.
2. Drill one level further inside the leading sector — usually two or three companies carry most of even that.
3. Then check breadth on the line that is hardest to engineer: revenue growth by sector, and how many sectors are participating.
4. Set the forward tell separately from the level. It is the rate of change of earnings growth that turns the narrative, not the absolute number.
5. Keep the multi-quarter path in view (this quarter vs the next two) rather than reacting to a single print.

• Ex-mega-cap index earnings growth; revenue (not EPS) breadth by sector; the second derivative — whether the growth rate is decelerating.

6. Run your own crowding discipline against the call you are already winning (Market Momentum in Energy)

1. Once a recommendation has worked, re-score it on positioning and sentiment as if you had no position — the same test you would apply to a name you don't own.
2. Look for the tell that everyone already knows the story: your own thesis being repeated back to you, uniform bullishness, volume concentrated in the obvious names.
3. Find the closest recent analogue — a trade that looked exactly this crowded shortly before it stopped working — and use it as the reference point.
4. Do not confuse crowding with a broken thesis. If the fundamental driver is intact, the correct action is a rotation within the theme, not an exit from it.
5. Name the less-crowded expressions of the same driver explicitly so the capital has somewhere to go.

• Your own thesis being quoted back at you; a recent analogue trade that topped out on the same sentiment reading; less-crowded names with the same fundamental driver.

7. Split the momentum score from the revenue score before chasing a breakthrough (Merck, Moderna and Healthcare)

1. When a stock you exited jumps on good news, state which score actually failed — price momentum, or the underlying business.
2. If momentum was fine and the business was not, the exit reason is unchanged by the news; a one-day event does not fix a three-year revenue trend.
3. Quantify the trend rather than describing it: multi-year revenue growth rate, not the latest quarter.
4. Put the multiple next to that growth rate in one sentence. A rich multiple attached to low-single-digit growth is the whole argument.
5. Let the other side answer on valuation explicitly — if the holder concedes the price and is holding for an unproven future driver, you have located the real disagreement.

• Three-year revenue CAGR versus the current multiple; whether the good news changes the revenue trajectory or only the sentiment; the holder's actual (often unstated) thesis.

8. In a headline partnership, count what is contingent before you price it (Marvell's Custom Chip Deal)

1. Take the headline dollar figure and find the share count behind it, then split that count into vested and contingent.
2. Read what the contingency is tied to — usually future purchases or delivery milestones. That converts the headline from a payment into an option on execution.
3. Ask whether the arrangement is novel or the latest instance of a template the industry has already established; a template implies repeatability, not a one-off windfall.
4. Check what the deal fixes for the recipient beyond money — credibility, a design-in, a restarted price trend.
5. Then price the second-order effect on the incumbent supplier separately (next insight).

• Vested vs contingent share counts; what the contingency is tied to; whether the same structure has been used elsewhere in the industry; actual purchase orders following the announcement.

9. When an incumbent falls on a rival's win, ask which order of risk is being priced (Marvell's Custom Chip Deal)

1. Separate first-order risk (the rival takes share in the existing product) from second-order risk (the rival, once inside, wins adjacent business later).
2. Decide which one the announcement actually supports. Incumbent relationships in complex, co-designed products rarely flip immediately.
3. If it is second-order, the correct time horizon for the damage is years, and the existing revenue stream is not at risk this cycle.
4. Check the customer's own incentive: a buyer diversifying suppliers is usually strengthening its negotiating position, not preparing to fire one.
5. Read the customer's diversification as a positive on the customer, which is a separate, often better trade than shorting the incumbent.

• Whether the incumbent's current programme is actually at risk; the customer's supplier count trend; adjacent product lines where the rival could expand next.

10. Slow-walk an exit — and say out loud that you are doing it (From IBM to Alphabet's Future)

1. When a thesis breaks on an earnings report, don't liquidate into the gap. Start selling and let the recovery bounce do part of the work for you.
2. Date the decision. Knowing when the selling started tells you how much has already been distributed.
3. Name the destination for the proceeds, so the sale is a switch with a comparative case rather than a pure exit.
4. Say plainly that more selling is coming and how much is left. Disclosing the remaining size removes the guesswork about future supply from your own hand.
5. Keep the residual bull case separate and specific (the one asset that could bring you back), so re-entry has a defined trigger rather than nostalgia.

• The date selling began versus the price since; the residual position size; whether the stated re-entry catalyst (here, quantum) shows commercial traction.

11. Hedge half a position into a scheduled event with a tight covered call (TJX Earnings)

1. Identify a holding with a known binary date (earnings) where you want to stay long but reduce the downside.
2. Sell calls against part of the position — half, not all — so a good result still participates on the unhedged half.
3. Place the strike tight (roughly 1% out of the money) and the expiry short (days, not months): event premium is concentrated in the nearest expiry.
4. Execute at the close the day before the print, when implied volatility peaks and none of it has decayed.
5. Be honest about annualising the premium — a huge annualised figure on a three-day option is arithmetic, not an achievable run rate.
6. Separate the trade result from the thesis: judge the print on the operating detail (guide, division mix), not on the option's outcome.

• Implied volatility into the print; the fraction of the position hedged; whether the operating detail behind the miss is in the core division or a peripheral one.

12. Weight a retail miss by the division's share of the business (TJX Earnings)

1. Before judging a retail print, get the revenue split by banner or division.
2. Locate the weakness. A miss in the division carrying the majority of revenue is a different event from a miss in a smaller one, however good the rest looks.
3. Check the miss against the specific metric that has been the company's story — for a discounter, comparable-store sales, not total revenue.
4. Note whether full-year guidance was maintained; a cut quarter with an intact year is a timing issue, a cut year is a thesis issue.
5. Treat store-expansion announcements sceptically when comps are soft — growth by square footage masks weakening per-store economics.

• Revenue share by banner; comparable-store sales in the largest division; full-year guide maintained or cut; store-count growth outrunning comps.

13. Use the options tape as a dissent check on your own fundamental view (Walmart's Upcoming Earnings)

1. Ahead of an event, read the implied move first — it tells you what the market has already agreed to be surprised by.
2. Then read the skew: not just volume but where the premium dollars are going, since size in premium is a better conviction signal than contract count.
3. Rank the largest trades by dollar amount and count how many are bullish, bearish or neutral.
4. Check whether the prior event's actual move exceeded the implied move; a company that has recently gapped harder than options predicted attracts extra hedging.
5. Hold your fundamental view against it explicitly: if you disagree, name the concrete, checkable reasons rather than restating the thesis.

• Implied move vs the last realised move; premium (not volume) skew; the dollar ranking of the top ten trades; a specific, datable fundamental item your view rests on.

14. Judge a pre-IPO private on margin profile, and discount leaked numbers for lag (OpenAI vs. Anthropic Profitability)

1. For a private company, treat any reported figure as a dated snapshot and check which period it covers before reacting.
2. Compare the loss trajectory against the revenue trajectory. Losses growing faster than revenue is the pattern that gets penalised; the reverse is tolerated indefinitely.
3. Prefer margin profile to headline revenue when ranking two competitors — revenue can be bought, margin cannot.
4. Look for any profitability marker at all (even EBITDA-level) as the qualitative dividing line between the two.
5. Apply the comparability discount: private accounting is not standardised, so treat gaps of a few percentage points as noise and only large structural differences as signal.
6. Note when the settlement date arrives — the S-1 or prospectus is when the numbers become auditable and the comparison becomes real.

• The reporting period behind any leaked figure; loss growth vs revenue growth; any EBITDA-positive marker; the S-1 filing that makes the comparison auditable.

15. Size patience in biotech by the payoff shape, not the current chart (Merck, Moderna and Healthcare)

1. Accept that the pre-catalyst chart is uninformative: a flat or falling line for months is the normal cost of the position, not evidence against it.
2. Identify the binary readout that changes the trajectory, and confirm it is genuinely binary (a first phase-3, an approval decision), not incremental.
3. Once it hits, do not treat the one-day jump as the trade. Ask whether adoption converts it into compounding revenue growth over subsequent quarters.
4. Use a completed prior example of the same shape to calibrate the multi-year size of the move.
5. Check whether the sector's recent growth came from science or from acquisitions — bought growth carries a limited number of years.

• The approval filing and full data release after a positive readout; quarter-on-quarter revenue build (not the announcement pop); how much of a pharma's growth is acquired versus discovered.

16. Test a textbook sector relationship against this year's evidence before trading it (Committee Debates Falling Yields)

1. Name the classical relationship you are about to rely on (flatter curve → weaker banks) and state its mechanism.
2. Check the recent record for the same conditions. If the relationship has already failed for months, the mechanism is being overwhelmed by something else.
3. Identify what that something else is before overriding the textbook — otherwise you are just extrapolating price.
4. Then hold the contrarian version of the risk separately: what policy surprise would restore the classical relationship?
5. Define the observable that would tell you the surprise is happening.

• Jackson Hole tone; short-end yields rising faster than the long end (a bear flattener); bank performance versus the curve shape over the last several months.