Actionable insights — Trading Ahead of the iPhone Event & the CPI Report

1. When you fear a macro variable, name the corporate action it actually stops (Navigating Inflation, Fed Hikes, Oil, and Earnings)

1. List the macro risks the desk is worried about, then refuse to hold them all at equal weight. Force a ranking.
2. For the one you rank first, write the transmission channel as a sentence with a company in it: "if X rises, then company Y stops doing Z." A risk you cannot finish that sentence for is a mood, not a risk.
3. Prefer channels that end in a financing decision over channels that end in a valuation multiple — financing decisions are observable, dated and reported; multiples are opinions.
4. Then find the same channel one layer down, at a single company you can actually monitor, and use its credit market as the early gauge rather than its share price.
5. Set the monitoring item, not the forecast: what would you have to see printed to know the channel has opened?

• Hyperscaler and AI-adjacent bond issuance calendars — a cancelled or downsized deal is the first observable. At the single-name level, the 5-year CDS on the most levered builder (Oracle here) crossing further above 200bp, and any capex guide raised while the credit is already widening.

2. Price the input that reaches the index, not the one on the screen (Navigating Inflation, Fed Hikes, Oil, and Earnings)

1. When a headline commodity is blamed for an inflation problem, ask which refined or downstream form of it actually enters the cost base of the goods in the index.
2. Check the three things that decide whether that form can mean-revert: inventories to draw on, a marginal buyer who can cut demand, and spare production capacity. Missing all three means the price is not a spike, it is a level.
3. Trace it to specific cost lines rather than to "the economy" — freight, agriculture, construction — so the CPI components you should watch are named in advance.
4. Only then form a view on the policy response, because the policy response is downstream of the print, not of the crude quote.
5. Cross-check that the trade implied by your input view is not the same trade as the headline commodity, which may behave differently.

• Distillate inventories and the diesel crack rather than the WTI quote; the freight and food components of CPI/PPI as the confirming print; and whether refiner equities (VLO, MPC) keep making highs while crude itself does not.

3. When a whole group is weak, stop explaining the members one at a time (Shake Shack, Starbucks, Airbnb, and GLP-1 Drug Impact)

1. Put up the multi-year charts of every member of the group, including the well-run ones. If they are all weak, a per-company explanation is doing too much work.
2. Ask what changed for the group's customers rather than for its managers — demand-side changes are the only kind that can hit every member at once.
3. Extend your candidate cause to an adjacent group that should also be affected. If the adjacent group has de-rated too, you have corroboration, not a reductio.
4. Find the beneficiary. A demand shift moves money somewhere; the chart of the beneficiary should look like the mirror image.
5. Say explicitly whether you have tested it. An untested hypothesis held openly is more useful than a company anecdote held confidently.

• Same-store traffic (not ticket) across the QSR group; calorie-weighted volume in packaged food; GLP-1 prescription growth as the driver series; and whether LLY's relative chart continues to invert the group's.

4. Rotate into a sector with a screen attached, and let it exclude good companies (Novartis Setbacks, AstraZeneca Success, and Biotech Growth)

1. Decide the sector on portfolio grounds — what it contributes that the rest of the book does not.
2. Then write one numeric screen that any candidate must pass, chosen for what this market is paying for rather than what the sector is famous for.
3. Apply it to the whole sector and publish the count: how many names you own, and how many fail the screen. The count keeps you honest.
4. Accept the sub-industries the screen deletes, even quality franchises. A rule with exceptions is not a rule.
5. Re-run the screen when the market's preference changes, not when a name you like fails it.

• Whether the market's paid-for factor rotates from revenue growth to free cash flow or dividend cover — the screen, not the sector, is what has to change. And watch names sitting at the boundary (AMGN "right there at around 10%"), which are the first to drop out.

5. Score a pipeline failure by what it was bought to replace (Novartis Setbacks, AstraZeneca Success, and Biotech Growth)

1. For a drug trial miss, separate the biological result from the commercial one: did it fail to work, or did it work on a marker without moving the outcome payers buy?
2. Trace the asset back to how it was acquired. A failure inside a large acquisition destroys the price paid, not just the future revenue.
3. Set it against the patent calendar. Failures matter in proportion to how soon the existing revenue rolls off.
4. Check the sympathy move in competitors — a rival falling on your failure means the market has downgraded the mechanism, which is a bigger event than one company's trial.
5. Look for the same day's winner in the same sector to calibrate: is capital leaving healthcare, or moving inside it?

• Whether the sympathy selling reverses (mechanism intact) or persists (mechanism doubted); the next asset in the acquirer's pipeline; and whether the day's winner keeps the sector's capital rather than exporting it.

6. Date the systematic signal against the sell-side note (Shake Shack, Starbucks, Airbnb, and GLP-1 Drug Impact)

1. When an upgrade lands on a name you already own, write down the date your own process bought it, and the reason.
2. Check whether the analyst's fundamental case restates facts that were public before your buy. If so, the signal led the narrative and the upgrade is confirmation, not information.
3. Test the fundamentals independently anyway — a raised company forecast, an unusual post-earnings price reaction, a demand cohort that has not cut back.
4. Do not add on the upgrade. The edge was the earlier date; buying the note converts your lead into someone else's exit.
5. Keep the record when the process is wrong too, or the record is worthless.

• The gap in days between the momentum add and the sell-side call, tracked over several names — it is the only honest measure of whether a systematic sleeve is leading or lagging the street.

7. Before a scheduled product event, ask what in it is genuinely unpriced (Debating Apple's Stock Ahead of the iPhone Launch)

1. Split the event into what is known (date, category, rough specification, rough price) and what is not (units, mix, the margin at that price).
2. If everything you can list is known, the event cannot re-rate the stock; the next earnings report is the first unpriced item, and that is what you should be positioning around.
3. Separately, and on a different clock, ask whether the product opens a new category. Category options are multi-year and cannot be settled by the month's price action — say so explicitly rather than defending them on a September chart.
4. Pull the seasonal base rate for the event, but treat it as context: it does not tell you which year you are in.
5. Check the street's posture. A stock the sell side has not "fallen all over itself" to upgrade into a marquee event has less positioning risk than the price alone suggests.

• Ship dates and lead times announced at the event (the first real unit signal), the mix toward the highest price point, and then the holiday-quarter guide — the first item on the list that was genuinely unknown.

8. State the re-entry price after tax, not after conviction (Qualcomm's Amazon Deal and Diversification Beyond Smartphones)

1. When you trimmed a position at a gain, compute the price at which repurchasing leaves the taxable holder whole — that number, not your target, is the buy level.
2. Keep the analysis and the execution separate, and say both out loud: liking the setup and buying it today are different decisions.
3. Let calendar risk defer, not cancel. "I will add, just not this month" is a position, and it is falsifiable next month.
4. Give the news its correct weight — a new customer in a new end market is a change in the shape of the business, not a quarter.
5. Note the counter-argument from the flow side: if everyone is signing similar deals, the deal is necessary but not differentiating.

• Whether the announced data-centre wins ever show up in the share price — Lebenthal's own flag is that this is "the third announcement this year" and the stock "hasn't really gained traction." Two more of those without a re-rating is the invalidation.

9. Write the print's pass/fail list before the print, in the company's own units (Oracle's Earnings, Funding, and AI Infrastructure Growth)

1. Pick the two or three line items whose values you would accept as settling the argument, and order them by importance.
2. Include at least one item that is a non-event — something the company should NOT do. Guidance raises are usually read as bullish; for a debt-funded builder, a capex raise is the opposite.
3. Include one credit-market gauge alongside the P&L items, because the equity will not tell you about funding stress until later.
4. Defer the second-order questions explicitly ("if they hit that, then we can worry about margins") so a mixed print does not become an argument about everything at once.
5. Name the physical asset whose ramp supplies the next data point, so the following quarter already has a question attached.

• A raised FY capex number alongside widening CDS — the combination, not either alone; and whether revenue growth lands at or below the 33% bar that makes the margin question premature.

10. Look for the beat in the segment nobody asked about last quarter (Navigating Inflation, Fed Hikes, Oil, and Earnings)

1. Read the previous quarter's call for what management did not spend time on, especially a line that normally gets airtime.
2. Ask why it was crowded out. A louder line item (deal flow, a one-off) can hide a cycle that is still running underneath.
3. Check whether the quiet segment has an independent driver — market levels for wealth management, loan growth for commercial banking — that you can observe without the company.
4. Take the position before the reporting season that will surface it, and accept that recent price action may be against you.
5. Express it in a name where the quiet segment is a large share of revenue, not in the index.

• Fee income and assets-under-management disclosures in the October bank prints rather than the headline EPS; loan growth as the commercial-cycle confirm; and whether the regional banks lead the money-centres, which is what a wealth/lending — rather than a trading — cycle would look like.

11. If a basket is the wrong shape for the theme, build the right one and give it a level (Mike Santoli on AI, Semis, and the Q Top Index)

1. Ask what the popular basket actually contains versus what the theme is about. If the two diverge, the basket's chart is answering a different question.
2. Construct or borrow a basket whose constituents include the theme's real cyclical exposures — for AI, the memory and merchant-silicon names, not only the platform owners.
3. Confirm it trades like the theme, not like the index, before you rely on it.
4. Attach one number to it — distance from its record high and time since the last one — so it functions as a tell rather than a narrative.
5. Use it to arbitrate between "the theme is back" and "one day it broadens, one day it is mega-cap."

• The Q Top making a new high (theme confirmed) versus the Mag 7 making one without it (re-concentration, not a theme). Santoli's own framing question also stands as a checklist: software rallied, healthcare got crowded — "have [semis] been de-risked enough to actually go back?"

12. Before buying the pure play, check whether you already own the exposure (Bearish Bets, Bullish Stocks, and Strong Commodity Trends)

1. When a sub-industry is working, ask whether an existing holding already contains it at scale — integrated companies often hide a large business inside a diversified label.
2. Size it: "third largest in the world" is a different answer from "has some exposure."
3. State the commodity floor your economics need, and where along the value chain the money is made above that floor.
4. Prefer the version with the fewer extra drivers only if you are being paid for the diversification; otherwise be honest that the pure play is the cleaner expression.
5. Cross-check with the options tape — heavy short-dated call buying on the diversified name suggests the market is expressing the same view there.

• Crude holding above $70 as the floor the economics assume; the diesel crack as the actual margin; and divergence between the pure refiners (VLO, MPC, COP at 52-week highs) and the integrated name, which would mean the substitution is costing you.

13. Read the options tape as a separate witness, and let it disagree (Bearish Bets, Bullish Stocks, and Strong Commodity Trends)

1. For a sector making highs, check the derivative market before assuming conviction: relative volume against the 30-day average, and the put/call balance by volume and by premium.
2. Treat "elevated volume, two-way trade, puts leading on count but calls leading on premium" as hedging, not doubt — and say which it is.
3. Separate the ETF flow from the single names. A quiet sector ETF with one violently active constituent is a stock story wearing a sector's clothes.
4. Scale the premium against something you know. Total options dollars traded is comparable across names of very different sizes and exposes crowding.
5. Never convert a flow observation into a recommendation. It tells you about positioning, not about the business.

• Whether Bloom's options premium persists past the day (crowding) or collapses (a single flow); and whether XLE volume finally picks up, which would mark the point institutional money joins a rally it has so far only watched.