Actionable insights — Weekly SSR: stripping the beta out of a cheap thematic name, buying a guided sequential dip, grading an arb against its unaffected price, reading the Kalshi-versus-banks split on the Fed, and selling a whole sector off a political calendar

1. Strip the theme out of a cheap thematic name — buy the company, short the ETF

1. Start from the objection, not the idea: when a name is cheap because its sector sold off, you are being asked to take two bets at once — that the company is mispriced, and that the sector stops falling. Separate them explicitly before sizing anything.
2. Establish that the company's own results moved up while the price moved down. Not "the stock is cheap" but "the stock fell on a quarter that beat." A beat plus a decline is a flow event; a miss plus a decline is information.
3. Identify the thematic index the name is being sold with, and check that a liquid ETF tracks it. The hedge only works if the instrument is tradable at reasonable cost and the correlation is the thing actually driving the drawdown.
4. Short the ETF against the long, sized to the beta you want removed, not to the dollar value of the position. What remains is the relative bet: does this company outperform its own theme?
5. Only run this where the company-specific case is genuinely idiosyncratic — a market-share position, a design win, a segment inflection. Hedging out the beta of a name whose only merit was the beta leaves you with nothing.
6. Prefer the under-followed name inside a crowded theme. The whole point is that the hedge removes the crowded part; what is left should be something the crowd has not priced.

• The financing and borrow cost of the ETF short (it can quietly eat the spread on a small-cap long); the correlation breaking down, which turns a hedge into two unrelated positions; the company reporting with the theme rather than against it, which is your signal the residual thesis has failed; and — the specific tell here — whether smart-glasses unit volumes and the automotive display driver share hold, since those are the only reasons to own the residual at ~30× forward earnings at all.

2. Distinguish a guided timing dip from a guided growth impairment — and buy the first

1. When a company guides a segment down sequentially and the stock breaks, read the stated reason word for word. "Customer spending timing," "order lumpiness" and "pull-forward" are calendar language; "pricing pressure," "share loss" and "demand softness" are not.
2. Check the reason against the segment's structure. A high-margin business selling to a small number of large clients on short contracts will be lumpy — that is a known property, not new news.
3. Find what analysts did with the guide. If consensus has been reset to a specific, much lower quarterly run-rate, the bar is now the thing to underwrite — you are buying against the revision, not against the peak.
4. Build your own full-year number from the segments rather than from the quarter, and value the company off that. Divide the enterprise value by your forward figure, not the trailing one — a fast-growing segment makes trailing multiples structurally misleading.
5. Bracket the multiple between the two businesses inside the company: what the growth segment would fetch standalone, and what the declining segment would fetch standalone. If the whole trades near the bad half's multiple, the good half is free.
6. Insist on a dated catalyst that arrives before the market has to believe your growth rate — something mechanical and calendared, not a re-rating you have to wait for.

• The next quarter confirming the reason — if the "timing" revenue does not reappear in the following period, it was demand; the renewal or loss of the affiliate contract, which is a binary on roughly half the company's identity; the 85% growth in cloud costs and sales commissions eating the incremental margin; and, generally, the screen this comes from: every high-growth segment inside a low-multiple parent where the guide-down language is calendar language.

3. Grade an arbitrage against the unaffected price, not the current one — and say when it is finished

1. Find the unaffected price: the last level before any leak, rumour or strategic-review disclosure. That is what the business is worth with no buyer, and it is the denominator for everything that follows.
2. Estimate the plausible deal price from the reporting, the comps and the rejected offers — not from where the shares are trading, which already contains the market's guess.
3. Write the ratio explicitly: dollars of upside to the expected deal price versus dollars of downside back to the unaffected price. A spread is only interesting when that ratio is favourable, regardless of how confident you are the deal closes.
4. Timestamp the rumour cycle. Money is made owning a candidate before a buyer is named; once the acquirer is in print, most of the premium has already been paid to whoever was early.
5. When the ratio flips, say so out loud and stop calling it an idea — a closing spread is a position doing its job, not a position to add to.
6. Keep a microcap sleeve for the spreads institutions structurally cannot trade, and route those alerts to a real-time channel, because their half-life is days.

• The antitrust calendar as the real driver of spread width — this week: LCII (Friday), ROKU 9/8, LNTH 9/17, TECH 9/18, KVUE 9/29 after South African clearance; political opposition as an under-priced regulatory risk (the deck flags "growing republican opposition" to the rail deal); rumours that go quiet for weeks, which usually means terms, not collapse; and the microcap screen specifically, where the spread is wide because of size rather than because of risk.

4. Trade a collapsed takeover as a bounded range with both ends anchored

1. After a bid dies, stop valuing the company as an acquisition and value it as a business. That number is your floor and it must be defensible without any buyer at all.
2. Take the ceiling from the board, not from your model. A publicly rejected offer plus the price the board signalled it would accept is a disclosed, negotiated range that the directors are on record defending.
3. Buy toward the floor, sell toward the middle, and be willing to do it repeatedly. The trade is the range, not the outcome.
4. Enumerate the plausible bidders and, for each, name the specific obstacle. A bidder list without obstacles is a wish; a bidder list with obstacles is a probability estimate.
5. Identify what the company can do for itself if nobody bids — buybacks, cost cuts, divesting a discrete asset. That self-help is what keeps the floor from eroding while you wait.
6. State plainly when the price is in the middle of your range and there is nothing to do.

• The floor moving — a deterioration in the underlying business is the one thing that breaks a range trade, because it lowers the level you keep buying at; the board changing its stated number, which resets the ceiling; a bidder's obstacle clearing (a rate cut for the PE case, an AI capex pause for the big-tech case); and rumour sources of varying quality — a specialist M&A blog is a lower-grade signal than a wire report, and should be sized accordingly.

5. When the prediction market and the sell-side disagree about the Fed, take the side with the political constraint — and price the path, not the level

1. Read the prediction market as the market-clearing probability, not as truth. It is money-weighted, so it tells you what is priced; that is the level you are being paid to disagree with.
2. Collect the sell-side distribution rather than one house's call, and note why each changed its view. A forecast revised on one speech and one payroll print is a weaker input than one revised on a data trend.
3. Look for a constraint the economists' models do not carry — an election calendar, committee composition, an institutional incentive to appear unified. Central bankers optimise for more than inflation.
4. Separate the chairman's signal from his intention where the committee is contested: hawkish language can be the price of internal consensus rather than a commitment to act.
5. Name the single data release that will settle it, and refuse to conclude before it prints.
6. Then price the path, because that is what equities respond to: one hike and a hold is a different asset-price regime from a sequence of hikes.

• The August CPI print itself (headline expected +0.4% m/m on war-driven energy, core flat at 0.2%, y/y falling to 2.4% on the housing weight) and PPI the day before (headline 0% → 0.4%, with PPI ex food, energy and trade expected to fall — the tell that "a lot of this is just due to the war"); the composition of any strong payroll print; Truflation, "the closest real-time inflation indicator we have," now at the highest point of the year; and the second-order political trigger — Trump threatening tariffs to force cuts, when "if we do charge new tariffs, the Fed would actually have to hike, so it's very counter-intuitive."

6. Buy operating leverage at a discount to private-market value — then put a date on the trigger and admit if you are early

1. Quantify the operating leverage as a unit sensitivity, not a narrative: how many dollars of annual EBITDA does a one-unit move in the underlying commodity add? A business where a small price change swings earnings materially is an option, and should be underwritten as one.
2. Value the assets against private transactions, not against listed comparables. Where private buyers routinely pay more per unit than the public market implies, the gap is the margin of safety.
3. Check that the asset base is being monetised in ways that do not require the cycle to turn — leases, alternative land use, adjacent revenue. That is what pays you to wait.
4. Read the payout policy as part of the option. A variable dividend formula tied to free cash flow converts an earnings inflection directly into cash to shareholders, which is a second source of re-rating.
5. Name the trigger and date it. An option with no expiry is a good thing; an option with no identified trigger is a hope.
6. Then test the trigger against your own macro work — and if your own analysis says the trigger is further away, say you are early and size accordingly rather than talking yourself into it.

• Single-family housing starts specifically (not total starts — the mix is the whole thesis); the 10-year Treasury yield, since mortgage rates track it and it is currently going the wrong way at 4.8%; lumber and OSB spot prices as the direct EBITDA input; private timberland transaction prices, which are what validate the NAV discount; and the variable dividend declaration, which is the first place an earnings inflection becomes visible in cash.

7. Size a credit scare before you act on it — and separate the liquidity problem from the solvency one

1. When a credit metric breaks out, immediately compute what it is worth in dollars: what share of the market it represents, and what a plausible default rate on that share actually costs.
2. Compare the size of the potential loss to the size of the asset class it is supposed to threaten. The same absolute number can be systemic in one era and irrelevant in another.
3. Distinguish the spread widening (a price signal about risk appetite) from the gating of redemptions (a structural signal about liquidity). Gates say nothing about credit quality and everything about the mismatch between illiquid assets and liquid promises.
4. Trace the gating back to its cause rather than treating it as a shock. Illiquidity in a fund complex usually has an upstream explanation somewhere in the capital cycle.
5. Keep a separate list of household-level metrics — delinquencies, not spreads — since those move independently of institutional risk appetite and reach the real economy first.
6. Then set the level at which the credit story starts to matter to equities, and track the distance to it.

• The 10-year against the 5% line; whether private-credit gates are lifted next quarter or extended (an extension is the escalation, not the initial cap); Q2's $99 billion of new private-credit fundraising, which he calls "quite alarming" precisely because it arrives while redemptions are gated; global PE deal count and distributions as the upstream cause; and subprime auto delinquencies as the cleanest read on the borrowing half of the consumer.

8. When the input-cost regime reverses, sell the sector — and date the reversal off the political calendar

1. Ask what a cyclical position was actually a bet on. A "cyclical bottom" trade is a bet on the spread between input cost and output price widening — not on volumes, and not on the company.
2. Check whether the input side has moved against you since entry. If the raw material is repricing faster than the finished product, the margin thesis is invalidated regardless of how well the company is run.
3. Identify what would restore the spread, and ask whether there is a dated, plausible catalyst for it. "Eventually" is not a catalyst.
4. Where the input price is set by geopolitics rather than by supply and demand, use the political calendar as your timeline — elections, ceasefire windows, sanctions review dates.
5. Then apply the conclusion at sector level rather than at name level. If the driver is an input price, every company that buys that input is affected, and picking the "good one" does not help.
6. Say the action word. "Trim or sell" is a decision; "less constructive" is a mood.

• Any credible de-escalation path before the midterms (the one thing that would reverse the call); European TTF gas and US diesel crack spreads as the live input-cost gauges; the same test applied in the other direction — refiners are the beneficiary of exactly the spread that is hurting chemicals, "which is very good for refiners right now"; and, structurally, every prior cyclical-bottom position in the book, since the invalidating condition is shared across all of them.

9. Read insider transactions as a two-sided screen — and treat a conflict as a research task, not an average

1. Screen Form 4 filings systematically rather than anecdotally. Open-market purchases and sales are the only disclosures where the person with the best information puts their own money behind a view.
2. Weight a purchase by the buyer's role and the size relative to their own history — the chief executive's eighth-largest purchase out of thirteen means something; a routine grant does not.
3. Weight a sale by the proportion of the stake, not the dollar amount. Executives sell for many innocent reasons, but halving a position is a different act from trimming one.
4. Flag unusual plan structures explicitly (a purchase routed through a 10b5-1 arrangement, for instance) — the mechanism is itself information.
5. When a name shows a bullish sell-side signal and a bearish insider signal at once, do not net them. Put it on the work list and say you have not concluded.
6. Build the tool: turn the screen into a ranked, repeatable output so the sorting is not done by memory each week.

• The October release of the Form 4 screener, which is where this becomes repeatable rather than anecdotal; sales clustered across multiple executives at one company rather than a single seller; purchases that coincide with a drawdown (the "buy the dip" pattern) versus purchases into strength; the size of the sale as a percentage of the holder's total stake, which requires reading the filing rather than the headline; and whether the corroborating macro agrees — a cyclical CEO selling into a decelerating cycle is a very different signal from one selling into an expansion.

10. Read a big assembler's earnings call as a survey of everybody else's supply chain

1. Identify the companies that buy from everyone — assemblers and integrators sitting downstream of the whole component chain. Their management commentary is a census of the industry, not a statement about one company.
2. Extract the constraint list verbatim and count it. One shortage is a supplier problem; a shortage across nine unrelated components is a systemic capacity problem that no single factory build-out will fix.
3. Turn each named constraint into the supplier that benefits from it, and check whether the scarcity is contracted or spot. Scarcity that has already been locked into long-term agreements accrues to the supplier; scarcity in the spot market accrues to whoever holds inventory.
4. Cross-check with the pricing evidence — the ratio of spot price to contract price is the cleanest measure of how tight a component actually is.
5. Corroborate with a second assembler's print before concluding demand is industry-wide rather than share gain.
6. Use the same list as a timing warning on the demand side: a build-out constrained across nine components will take longer than announced, which stretches the revenue schedule of everyone in it.

• The spot-to-contract ratio narrowing, which is the first sign a shortage is ending; new capacity announcements with real dates (Samsung converting its Pyeongtaek S5 foundry line to memory, Micron adding 60k wafers/month); a competitor removing itself from the market (China's CXMT stalling HBM3E production, "positive for Micron and SK Hynix and Samsung"); the next assembler's call repeating or shortening the constraint list; and power specifically, since it is the constraint that no chip fab can relieve.