# Session 40 — Block 026 pre-read teach (2026-08-09)

**Status:** Side-session, pre-read teach per 2026-07-27 S32-side convention. No discrete block entry; pre-reads loaded as substrate for Block 026 main session.

## What we did

**Macro-mentor skill recovered.** Restored `finance/macro-mentor/` (SKILL.md + 48 references) and `finance/macro-mentor-probe-discipline/` (SKILL.md + 4 references) from the 2026-07-31 curator backup. 54 files, 1:1 match with backup. Root cause: a curator run between 2026-07-31 and 2026-08-07 wiped the entire `finance/` skills directory (124 files → 1 file in tarball). New snapshot taken at 2026-08-09T15-12-01Z (133 files, 360 KB).

**Skills backup taken.** `/home/rpi/.hermes/skills/.curator_backups/2026-08-09T15-12-01Z/` — recovery point for current state.

**Source update (per user's pre-read list note "verify most current BIS WP on the topic").** Source 3 substituted: from unverified Sarno/Schmeling 2014 reference to **BIS Bulletin #90 (Aquilina, Lombardi, Schrimpf, Sushko, August 2024) + BIS Bulletin #124 (2025)**. The August 2024 yen carry unwind is the canonical post-COVID case study — on par with 2008 and March 2020 as the third historical anchor for Block 026 carry unwind regime classification.

**User feedback applied during teach.** First attempt on Source 1 led with arithmetic / cost-of-carry calculation probes. User flagged: "I am not planning to trade arbitrages so I think we can skip the calculation probes for now and rather test the conceptual understandings." Re-shaped to substrate-test probes (no arithmetic). Follow-up on Source 1: user raised wealth-client perspective ("I have traded forwards directly with custody banks for wealthy individuals..."). Honest answer delivered: forwards are operationally cheaper than futures for wealth clients; futures-specific advantage is daily settlement reflexivity, not execution cost. The framework substrate for Block 026 carries forward as: *the daily settlement / margin discipline mechanism applies to any cleared derivative, not just futures — the textbook "futures amplify" framing is a simplification.*

## Substrate delivered (the three sources)

### Source 1 — Hull Ch. 6 (currency futures)
- **Three distinguishing features vs forwards:** standardized, cleared, daily MTM
- **Pricing identity:** cost-of-carry / CIP in futures form
- **Load-bearing insight for Block 026:** daily settlement is what makes carry trades their own accelerant in stress via forced daily reckoning. Forwards (OTC, uncollateralized) don't have this; cleared swaps/ forwards do.
- **Wealth-client framing:** futures worse than forward on most wealth-client dimensions (tenor, notional, no daily margin calls). Futures' actual advantage is the basis as a clean read on the interest rate differential and as a regime signal.

### Source 2 — Lyons Ch. 1-2 (microstructure approach)
- **Customer order flow = proximate cause of FX price movement**
- **Three concepts:** customer order flow as information signal, hot-potato inventory management, macro information transmission via positioning changes
- **Operational use:** order flow as confirmation signal, customer-dealer breakdown matters for FX driver analysis, microstructure vs macro horizon separation, microstructure explanation for forward bias persistence
- **What Lyons doesn't do:** doesn't speak to macro drivers themselves; mostly about spot; "customer-dealer" frame is FX-specific

### Source 3 — BIS Bulletins #90 + #124
- **Load-bearing identity #1:** carry = vol-regime-dependent ("nickels in front of a steamroller" framing)
- **Load-bearing identity #2:** monetary policy transmission to FX is *state-dependent* on carry activity (25bp shock → 4-10% FX move when carry heavy, no significant move when carry light)
- **August 2024 case study:** yen carry unwind, ~¥40T / $250B visible, positioning unwind not funding crisis, basis moved but didn't reach crisis, Fed did nothing, self-corrected in days
- **Three historical anchors (canonical):**
  - 2008 GFC: USD-funded, Lehman trigger, -100bp+ basis, months to resolve
  - March 2020: USD-funded, COVID trigger, -100bp+ basis, weeks to resolve, Fed swap line expansion
  - August 2024: JPY-funded, weak NFP + BoJ trigger, elevated but not crisis basis, days to resolve, no Fed action
- **PM diagnostic layers:** carry-to-risk ratio (price-based), net speculative positioning (CFTC COT), option-implied vol regime. All three align = unwind imminent.

## Substrate-Recall substrate for Block 026 main session

**Five sentences the user should have firing as cold-recall:**

1. Daily settlement = forced daily reckoning = reflexivity in stress (Hull, Source 1)
2. In FX, customer order flow is the proximate cause of price movement; macro variables are inputs, not the mechanism (Lyons, Source 2)
3. Carry trades are interface trades between rate-differential carry and the volatility regime — gains in calm, losses in turbulence (BIS #90, Source 3)
4. August 2024 yen carry unwind was a positioning unwind, not a funding crisis — basis moved but didn't reach crisis levels, Fed did nothing (BIS #90, Source 3)
5. Monetary policy transmission to FX is state-dependent on carry activity — 25bp shock → 4-10% FX move when carry heavy (BIS #124, Source 3)

## Pre-block probe queue (carried to Block 026 main session)

The conceptual probes Block 026 will open with:

1. **Hull — daily settlement reflexivity.** Why does the daily settlement feature of futures make them behave differently from forwards in a carry unwind, even when the underlying exposure is the same?
2. **Lyons — order flow as proximate cause.** Why does customer order flow have predictive power in FX? What does it tell you that price alone doesn't?
3. **BIS #90 — why carry persists despite CIP.** What's the empirical evidence that carry trades accumulate in low-vol regimes and unwind in vol regime shifts? What's the most-cited explanation?
4. **BIS #124 — state-dependent monetary policy transmission.** Why does the FX response to a 25bp policy shock vary by an order of magnitude depending on carry trade activity? What is the mechanism?

These are substrate-test probes (no arithmetic), per user feedback during this session.

## Process notes

- **Pattern #11 (mechanism cold-recall) applies throughout Block 026 main session.**
- **Pattern #17 (confirmation-loop check)** — if user's pushbacks start teaching the mentor on Block 026 substrate, close out, don't grind locked substrate.
- **Three-attempt probe discipline** — if probe muddled twice, restart with cleaner mental model (rule filed Session 39, Block 025 Target 4).

## Anki card reminders (push at Block 026 main session open, NOT now)

1. Daily settlement = forced daily reckoning = reflexivity in stress (Source 1)
2. Carry trade = interface trade between rate-differential and vol regime (Source 3)
3. August 2024 yen carry unwind was positioning, not funding (Source 3)
4. State-dependent monetary policy transmission (Source 3)

## Source URLs (verified)

- BIS Bulletin #90: https://www.bis.org/publ/bisbull90.htm (Aquilina, Lombardi, Schrimpf, Sushko, August 2024 yen carry unwind)
- BIS Bulletin #124: https://www.bis.org/publ/bisbull124.htm (2025 state-dependent monetary policy transmission)
- Hull 10e Ch. 6: textbook (currency futures section)
- Lyons "The Microstructure Approach to FX" Ch. 1-2: book
- Menkhoff, Sarno, Schmeling, Schrimpf (2012) "Carry trades and global FX volatility" — Journal of Finance — foundational academic reference

## Directive for Session 41

1. Open with pre-block probes (4 conceptual, one at a time per Pattern #11). User feedback during this session: probes test substrate, not arithmetic.
2. If user fails a probe, re-anchor the substrate (don't bundle corrections in a single response — restart with the load-bearing sentence that should fire).
3. Targets as scoped: FX drivers framework → carry trade mechanics → carry unwind regime (2008 / 2015 CHF / August 2024 / March 2020) → PM implications.
4. Three-attempt probe discipline applies (Pattern #11 + 3-attempt rule from Session 39).
5. Block 026 `## Plan` section to be filled at session close (per 2026-06-23 convention).
6. Anki cards + competence map + coverage tracker updates at block close.

## State for next session

- Macro-mentor skill + macro-mentor-probe-discipline skill: present (restored from 2026-07-31 backup, fresh snapshot taken at 2026-08-09T15-12-01Z)
- Current-block pointer: Block 026 marked as ● current in current-block.md (Session 40 update only — no Block 025 closure log entry, that's done in Sessions 38-39)
- Pre-reads: taught as substrate per user directive. No discrete session-log entry per 2026-07-27 S32-side convention.
- Resume file: this document.