Actionable insights — Macro Update #19

1. Turn a rare momentum signal into a conditional base rate before you lean on it

1. Define the signal mechanically so it can be counted: here, the index going from a 21-day closing low to a 21-day closing high in 4 trading days — a fixed lookback, a fixed direction, a fixed elapsed time. No discretion in the definition.
2. Count every historical occurrence and check the unconditional outcome first. Eighteen instances on record, "higher on some occasions, lower on others" — accept the honest verdict: on its own it is a coin toss, "nothing to lean on."
3. Now add one orthogonal condition that changes the regime the signal fires in — location relative to the trend, not the thrust itself: only the occurrences that landed within 2% of a 52-week high.
4. Re-read the conditional sample: 9 of the 18, higher a year later every time, mean gain 13.28%. A thrust from a depressed tape and a thrust into new highs are two different events wearing the same statistic.
5. State the sample size out loud and refuse to over-trust it — "even if it is a relatively small sample size" — and only act on it where it corroborates a framework you already hold, not as a standalone reason.
6. Pre-name the conditions that void the record, so the signal can be falsified rather than defended: here, the Fed and another run in the oil price.

• The two named voiders: a Fed that stops pinning the front end, or a fresh spike in crude.
• Whether the tape is still within 2% of its 52-week high when the signal fires — outside that band the base rate does not apply.

2. Cross-check a headline data print against the places its story must also show up

1. Decompose the headline before reacting to it: a falling unemployment rate is not the same event depending on the numerator and the denominator. Here 4.1% arrived "only because people walked out of the labor force rather than into jobs," with participation at 61.4% — outside the pandemic the weakest since the 1970s.
2. Check the revisions alongside the fresh number: −23,000 for the month is one thing; a combined −103,000 revision to the prior two months changes the trend, not just the print.
3. Ask the corroboration question: if this were real, where else would it already be visible? For a cracking labour market the answer is spending figures and what companies are actually saying on their calls. "And mostly it isn't."
4. Hold both readings at once rather than picking a side — flag the print as possibly reflecting one-time circumstances, but don't build a position on your suspicion.
5. Then separate your read from the market's read and trade the second: "let's take the numbers as the market takes them." Skepticism informs your risk, not your direction.

• Spending data and company commentary catching down to the labour print (confirms it) — or continuing not to (keeps the one-time explanation alive).
• Wage growth: 3.2% and the slowest in five-plus years is the piece that most directly hands the Fed its inflation-is-behind-us line.

3. Track a policymaker's script against their evidence to read the reaction function

1. Log the stated position at each public appearance and date it — confirmation testimony, first press conference, subsequent conference appearances.
2. Flag any shift, then ask the diagnostic question: what evidence was cited for the change? A pivot argued from named indicators is a data-driven Fed; a pivot with "not a single indicator" named is a political or preference-driven one.
3. Look for the parallel channel that would explain the shift — private meetings, who is being blamed, who is being investigated, whether the board is being reshaped around the incumbent.
4. Translate the reaction function into the curve, not just the level: a chair who wants cover to keep the front end pinned into a soft print means rate-hike odds pushed out of pricing, lower yields and a bull-steepener — that is the trade, not "dovish therefore buy."
5. Check whether the sell-side is already loaded into that same setup before you size it — a consensus setup pays less and unwinds harder.

• Whether the next CPI print is soft enough to convert the script into actual pricing (hike odds out, yields lower, curve steeper).
• Evidence finally being named — a genuinely data-anchored pivot behaves differently from a preference-driven one when the data turns.

4. Map a geopolitical stalemate to a sequence, then locate yourself on it

1. Measure negotiating distance, not headlines: put the two sides' stated demands next to each other and ask whether the gap widened or narrowed week over week. A list of six concessions on one side against a counter-demand for compensation on the other is a widening gap.
2. Discount the partial agreements that don't move the binding constraint — new shipping-lane coordinates agreed with a third party "does not reopen the waterway." Ask what each announcement actually unlocks.
3. Write the transmission chain out explicitly so you know which link you are watching: no deal → increased strain on the UST market → weakness for risk assets and the general market → eventual fiscal intervention.
4. Note where the intervention sits in the chain — it comes last. "Eventual fiscal intervention, but not before more selling" means the rescue is not a reason to be early.
5. Separately track the market's sensitivity to the same headlines: when the tape starts "caring less and less" about the chokepoint, the geopolitics has become a headline traded around rather than a wall run into — and the chain is dormant until something re-arms it.

• The floated 30-to-60-day ceasefire actually landing — that is the link that would break the chain at step one.
• Strain showing up in the Treasury market: that is the first link with a price, and the one that would confirm the sequence is running rather than dormant.

5. Watch for the moment a capex boom turns into a credit product — that is the sensitivity tell

1. Track how the buildout is being funded, not just how large it is. Equity-funded capex answers to shareholders; debt-funded capex has to be refinanced on whatever terms the credit market offers later.
2. Look for the four markers of the transition: (a) an asset being pledged as collateral that was previously just equipment (compute itself), (b) the vendor backstopping a slice of its customers' borrowing, (c) a bank taking the bookrunner seat for public debt to come, and (d) the language changing — a supplier calling its product "an investable infrastructure asset."
3. Add the regulatory dimension: check whether the rules governing that paper are being loosened while issuance accelerates — the post-2008 risk-retention regime being waived for data-center securitizations is the classic late-cycle tell.
4. Size the issuance trend against the rule change: $2.4bn in 2020 → $15.5bn last year → on pace for another record, with the brake removed for "the fastest-growing corner of new credit."
5. Note the infrastructure being bought around the flow (an exchange paying $6bn for a corporate-bond venue) — capital committing to the plumbing is a vote on how much paper is coming.
6. Conclude in terms of sensitivity, not timing: this doesn't say when it breaks, it says the complex now breaks on credit conditions rather than on AI demand alone, and every circular deal adds a layer.

• The first public bond deals out of this consortium — pricing and demand there is the read-through on the whole structure.
• Data-center ABS issuance pace versus spreads: accelerating supply into widening spreads is the point the sensitivity becomes visible.

6. Hold the residual cash through a scheduled catalyst — and judge the reaction, not the print

1. Deploy on weakness when it comes (he bought more in July), then stop — do not chase the rest in on the way back up.
2. Identify the next dated catalyst and make it the decision point for the remaining cash, so the pause has an expiry rather than being open-ended indecision.
3. Judge two things at that catalyst, in order: how the data comes in, and — "more importantly" — how the market reacts to it. A market that rallies on a bad print, or sells a good one, is telling you what is already positioned.
4. Keep the reason for the residual cash explicit ("there is still plenty of risk out there") so you know what has to resolve before it goes to work.
5. Hold the two conclusions in balance: the momentum work says path of least resistance up, the macro work says the backdrop is uncertain — "so keep that balance in check" rather than resolving it prematurely in either direction.

• The reaction function at the print: a soft CPI that fails to lift the tape would contradict the whole bullish setup even though the number "cooperated."
• The named momentum-breakers (the Fed, an oil run) arriving before the cash is deployed — that flips the residual cash from opportunity cost to dry powder.