Actionable insights — BoJ, Oil, Warsh & the Credit Comment

1. Watch the primary bond market, not the credit index — new-issue premiums and oversubscription as a regime detector

1. Understand why the signal lives in primary. Fixed income is more passive than equities — even nominally active IG managers stay "within 5% of the duration of the index" (IG index duration 6.6; managers 6.3-7.0yr) "because anything outside of that is too much career risk." Their behaviour tells you nothing.
2. Note the calendar mechanic: new issues only enter the index at month end, so passive money cannot buy them at launch. The marginal buyer of a new deal is therefore "active guys and fast money hedge funds" — a genuinely discretionary bid, which is what makes primary a clean read on real risk appetite.
3. Track the two primary statistics weekly, not the spread level:
   • New-issue premium (concession) — how many extra basis points an issuer must pay above its own secondary curve to get done. Compare this week to the July average, the YTD average, and last year's YTD.
   • Oversubscription ratio — order book divided by deal size. This is book depth, and it deteriorates before spreads do.
4. Apply the directional rule that defines the regime: "in bull markets in credit, new issue spreads tighten towards secondary, while in bear markets, secondary widen towards new issue levels." So a rising concession is not just a worse deal for one issuer — it is the level the whole secondary market will be dragged to.
5. Cross-check planned versus realised supply. Issuance running far below the month's initial estimate, while fund inflows stay healthy, isolates the problem to price rather than demand — issuers are choosing not to pay.

• Concessions staying elevated for consecutive weeks (one week is a calendar artifact); oversubscription slipping under ~3×; and the confirmation trade — secondary spreads widening out to meet new-issue levels rather than new issues grinding tighter. The reverse (premiums collapsing back toward 3bps, books rebuilding above 4×) is the all-clear.

2. Deals pulled and one repricing print — the ground truth that arrives before the index moves

1. Ask syndicate/credit contacts one question: what got pulled this week, and why? A deal withdrawn "because the all-in yields are just not economical" is a hard economic fact — the project's IRR no longer clears its cost of capital — where a spread chart is just a mark.
2. Find the marginal print: the one deal in a sector that had to clear regardless. Its pricing is the true clearing level, and it drags the whole comparable set with it. Measure the sector-wide reset in basis points, not the single deal's headline.
3. Compare the print to its rating. When a supposedly A+ credit has to pay a spread the market associates with a much lower rating, the ratings are the thing that is stale — trade the market's number.
4. Convert the observation into the sector question: can this business model still be financed at the new level? If the answer is no for the marginal borrower, the growth rate of the whole sector's capex is capped — irrespective of demand.
5. Weight the messenger. A source "who tends to be skeptical" but "almost never waves his arms" changing tone is itself a datapoint; so is the tell that colleagues "want to keep dancing."

• Whether pulled deals return within a few weeks at the new level (a clearing market) or stay shelved (a closing one); whether the repricing spreads from the one sector into adjacent ones; and the equity-market lag — the shares of the capex beneficiaries usually have not discounted a financing cap yet.

3. For any funding-dependent asset class, track the fundraising run-rate, not the returns

1. Identify businesses whose marks and liquidity both depend on fresh inflows — private credit, non-traded vehicles, anything where redemptions are paid from new money rather than by selling assets.
2. Pull the quarterly gross capital raised from the filings, and — critically — the segment breakdown. A modest decline at the group level can conceal a collapse in the one segment that matters.
3. Compare year-over-year, not quarter-over-quarter (fundraising is lumpy and seasonal), and locate the reading in its own multi-year range ("lowest in three years" is the phrase to look for).
4. Then size the queue on the other side: how much debt is waiting to be refinanced through this exact channel — project finance parked on bank balance sheets, maturing loans, warehoused paper. Supply of paper rising while the supply of money falls sets the price of concession.

• The next quarter's segment number (a second down-quarter turns a stumble into a trend); redemption/gating headlines at any peer; and discounts to NAV on the listed proxies, which price the private marks the managers won't.

4. Read the credit market's verdict on an equity story — CDS as the balance-sheet lie detector

1. When a company announces a large financing commitment — vendor financing, an equity stake in a customer, an offtake guarantee — do not score it on the revenue it implies. Score it as a liability.
2. Go to the company's CDS, not the share price. Equity holders are paid to be optimistic; protection buyers are paid to be right about solvency. A CDS widening on a "growth" announcement is the market reclassifying the news.
3. Check whether the widening happens against the tape — i.e. on a risk-on day when credit generally is bouncing. Divergence from the beta is what makes it a signal rather than noise.
4. Trace where the funding actually lands: if the announcement lets a customer buy the announcer's product with the announcer's money, the receivable and the revenue are the same dollar. That is the circularity to price.

• The CDS holding wider after the risk bounce fades; the same pattern appearing at the counterparties being financed; and the equity finally acknowledging it — the credit market normally leads by weeks.

5. The "cleanup print" — trade the end of a forced liquidation, not the middle of it

1. When a factor sees a violent unwind, ask who was carrying it and in what size. A fund that "can go from up double/triple digits to down on the year in a single month" is the signature of concentrated, levered exposure to the unwinding factor.
2. Watch for the liquidation to become discrete rather than continuous: an approach to LPs to take positions directly, a block crossing, "dumped all their listed investments through one enormous trade." The old NYSE name for it is a cleanup print.
3. Size the overhang before the print and confirm it is gone after: what fraction of AUM was in listed equities, and which sectors did it sit in? That names the beneficiaries of the relief.
4. Take the bounce as a supply event, not a thesis change — the forced seller is finished; nothing else has been fixed.

• Whether the bounce holds beyond a few sessions (a real clearing) or fails at the pre-liquidation level; and for the next holder of the same factor — one visible blow-up usually means others are still de-grossing quietly.

6. Scan the cross-asset volatility panel for the laggard — the last quiet complex is the next domino

1. Put implied volatility for every major asset class on one panel — equities, rates, credit, commodities, FX — and normalise each to its own history. Find the one that is not like the others.
2. Treat a persistently suppressed vol complex as stored energy, not as stability: it is where leverage has accumulated precisely because it looked safe (carry trades live in low FX vol).
3. Model the feedback, not the level: once the laggard wakes, it "reflexively feeds back into the already rising volatility in other major asset classes" — vol targeting, margin and VaR budgets are computed across the whole book, so a shock in the sleepy leg forces de-grossing in the others.
4. Weight for the liquidity calendar. The same shock in "one of the most illiquid months on the calendar" produces a far larger price move than in a full-staffed month.
5. Identify what would wake it, and the actor who controls the timing — here, a central bank waiting for another central bank before spending its reserves.

• FX implied vol failing to fall back after the intervention (intervention that raises vol permanently has changed the regime); carry-sensitive assets underperforming on yen strength; and whether the other complexes' vol rises with it, confirming the feedback rather than a one-off.

7. A leveraged blow-up near the highs is a leverage signal — check the base rate before calling it a top

1. When a fund fails only a few percent below all-time highs, resist the instinctive conclusion that the market has topped. The information is about how much leverage was required to be profitable, not about the market's direction.
2. Go to the historical precedent and ask what actually happened next, not what it felt like at the time.
3. Separate the two horizons: a blow-up removes a forced seller (short-term supportive) while revealing systemic leverage (medium-term dangerous). Trade the first, position for the second.

• The interval between blow-ups shortening — one is an accident, a sequence is the regime. And for the 2007 rhyme: new highs made on narrower breadth after the first casualty.

8. "Doing nothing is dovish" — grade a central bank through an emerging-market operator's eyes

1. Write down the expected action and the reasons for it before the meeting — the rhetoric, the committee composition, the historical precedent (e.g. "pretty much every new Fed chair in modern history except Arthur Burns" uses the first meetings to shake the tree). Then score the deviation.
2. Apply the EM rule: inaction plus silence is easing, whatever the accompanying language claims. A chair who wants to "keep the market guessing and let it do the work" has, in a country with an inflation credibility problem, simply delegated policy — "you know that doing nothing will be read as dovish no matter what you say about the future, especially if you say nothing about the future."
3. Grade the reaction function by the cross-asset response, not the equity close. The diagnostic combination is USD down + equities down + bonds down — a "Triple Yasu" — which says the market is repricing credibility, not growth. Within it, the bond is the tell: falling stocks with falling bonds removes the safe-haven bid that a pure growth scare would create.
4. Measure the catch-up gap: how far market pricing has run ahead of the official path, and how many meetings it would take to close it. A wide gap with few meetings left is the setup for a forced, larger move later.
5. Follow the delegation to its conclusion: if the central bank has told the market to do the tightening, expect the long end to tighten conditions on its behalf — and to overshoot.

• Long yields continuing to rise on soft data (the delegation working); the USD failing to rally on higher yields (the credibility read); and any attempt to retake the narrative with intermeeting guidance — the tell that the "let the market do the work" stance has become uncomfortable.

9. When the news flow is censored, let the curve report — front-of-board versus back-of-board

1. When you suspect an information blackout — restricted reporting, confiscated devices, an absent mainstream account — stop trying to source the story and read the market's own microstructure instead.
2. Split the commodity board in two: the prompt/physical leg (immediate-delivery grades, nearby timespreads) and the deferred leg. Prompt prices and spreads reflect what physical participants can actually see today; deferred prices reflect narrative and hedging.
3. A divergence — front and timespreads exploding while the back end is sold — is not a contradiction to average away. It means the physical market is reacting to something the paper market is discounting as temporary (or being told to discount).
4. Assemble the unreported supply events into a list and check whether any single one is priced. If the flat price sits still while the list grows, treat the suppression itself as the setup: a "beach ball under water," i.e. an accumulating, mechanically unstable position rather than an equilibrium.
5. Trade it as a volatility proposition rather than a directional one when the timing is unknowable — "mechanically this thing looks like a volatility rupture in the making."

• The back end catching up to the front (the market conceding the supply loss is structural, not transient) versus the front collapsing back (the events were absorbed). And the generalisation: any market where a visible stack of unpriced news meets a suspiciously pinned flat price is a candidate for the same treatment.