--- deck: Macro Study::Phase 1 — Components::FX tag: block-028 session:2026-08-16 refactor:2026-08-16 note_type: Basic status: draft-revision-3 revision_basis: user pushback 2026-08-16 (Cards 6 reframe, historical-event scope cut) --- # Block 028 Anki card batch — REVISION 3 User pushbacks applied: - **Card 6 family:** "Q1/Q2/Q3..." recall has no anchor context. Reframe to *what information source answers the question*, not the Q-number. - **Historical event scope:** cards should test *concepts*, with historical events as scaffolding. Pure event-recall cards removed (Cards 4, 5, 5b). Concept cards that *use* events as illustration retain the event but the recall target is the principle. - **Tag string simplified:** `block-028 session:2026-08-16 refactor:2026-08-16` --- ## 1. Sterilized vs unsterilized intervention ### Card 1 — Sterilized intervention **Front:** What is *sterilized* FX intervention? **Back:** CB buys/sells FX + simultaneously offsets via domestic OMOs → monetary base unchanged, no rate effect. Only signaling/portfolio-balance channels work. ### Card 1b — Unsterilized intervention **Front:** What is *unsterilized* FX intervention? **Back:** CB buys/sells FX without offsetting OMOs → monetary base changes, rate effect non-zero. The FX intervention *is* the monetary policy move. ### Card 1c — DM vs EM pattern **Front:** Which type of FX intervention dominates modern DM (BOJ/SNB/ECB), and which dominated EM regimes (2000s Brazil, China pre-2015)? **Back:** **Sterilized** dominates DM (targeting FX expectations, not monetary conditions). **Unsterilized** dominated EM (managing FX as part of the monetary stance). --- ## 2. Why intervention works despite tiny volume ### Card 2 — Signaling channel **Front:** SNB's intervention was ~$500m/day ≈ 0.1% of daily JPY turnover. Why did it move the FX market? **Back:** **Signaling channel** (load-bearing): intervention changes *expectations* about future policy (rate path, balance sheet, capital controls), not flow. The market believes the CB has more info/commitment than the print reveals. *Announcement* often moves FX more than the actual trades. ### Card 2b — Portfolio-balance channel **Front:** When does the *portfolio-balance channel* of intervention do meaningful work? **Back:** When intervention is **unsterilized** — it changes relative supply of domestic vs foreign currency assets in private hands → risk-premium compression can move FX. Small relative to signaling in modern DM/EM. --- ## 3. FRB-NY Treasury-Fed split ### Card 3 — Who has policy authority **Front:** Who has *policy authority* for US FX intervention, and who *executes* it? **Back:** **Treasury (ESF, $200bn+)** = policy authority. **Fed (SOMA custody)** = operational execution. Fed doesn't formally have FX policy authority but executes Treasury directives. ### Card 3b — PM read on Treasury+Fed vs Fed-alone **Front:** You see a Fed-alone FX statement vs a Treasury+Fed joint statement. Which signals broader fiscal+monetary intent? **Back:** **Treasury+Fed joint** = broader fiscal+monetary intent. **Fed-alone** = narrower, typically dollar funding not dollar level. --- ## 4. SNB 2015 — concept cards (event-as-scaffolding, NOT event recall) ### Card 4 (was 4b) — Why long holds happen (concept) **Front:** What makes a CB hold an FX peg or floor for *years* despite the cost of defending it? **Back:** **Real-economy justification.** The peg serves a policy goal beyond the FX level (e.g., preventing deflation via import price stability). As long as the real-economy justification holds, the CB defends the level. Credibility is load-bearing. *(Scaffolding: SNB held EUR/CHF 1.20 from Sept 2011 to Jan 2015 — 3.3 years — on real-economy justification of preventing CHF deflation.)* ### Card 4b (was 4c) — What triggers capitulation despite long holds (concept) **Front:** A CB has held an FX peg/floor for years. What three things can trigger capitulation? **Back:** (1) Cycle position reverses — the cycle the CB was fighting strengthens. (2) Real-economy justification erodes (e.g., a major policy shift abroad removes the deflation/inflation pressure that motivated the peg). (3) Capacity cost crosses threshold — reserves accumulate to a level that creates its own balance-sheet risk. *(Scaffolding: SNB Jan 2015 — anti-EUR cycle strengthened, ECB Q€ eroded the CHF-deflation justification, $200bn+ EUR reserves = ~30% of SNB balance sheet.)* **Audit note on Card 4 / 4b:** the events (Jan 15, 2015, $200bn+ reserves, ECB Q€) are scaffolding — they illustrate the concepts but are not the recall target. If the user remembers the principle but forgets the SNB-specific year, the concept still fires. Card 4b keeps the 3-item back per exception (the items are co-equal triggers, the question asks "what three things?"). Acceptable per back-side-list exception. --- ## 5. BOJ 2022 — concept cards (event-as-scaffolding, NOT event recall) ### Card 5 (was 5c) — Why capped intervention works as a pause, not a reversal (concept) **Front:** A CB commits a *capped* amount of intervention (e.g., ¥5-10T per episode then pauses). What does this commitment size tell you about the intervention's likely time horizon and reversibility? **Back:** **Pause mechanism, not reversal.** Weeks-to-months hold at most. Cannot reverse a global cycle — only slows the move. Intervention can pause, intervention can slow, intervention cannot reverse a global cycle. *(Scaffolding: BOJ Sept-Oct 2022 — capped at ~¥5-10T per episode, USD/JPY held below 150 for weeks-to-months, then drifted back to 150+ on cycle when intervention stopped.)* ### Card 5b — Unlimited vs capped commitment distinction **Front:** Unlimited intervention (e.g., SNB 2011-2015 on EUR/CHF) vs capped intervention (e.g., BOJ Sept 2022 on USD/JPY) — what time-horizon difference does this commitment-size distinction create? **Back:** **Unlimited** = hold measured in years (capitulation-bet territory). **Capped** = hold measured in weeks-to-months (pause, not reversal). **Audit note on Card 5 / 5b:** both cards test *conceptual* recall (what commitment size implies about time horizon). The SNB/BOJ scaffolding is in parentheses so the user can anchor to a known case but doesn't have to recall the specific date/magnitude. Pure event cards (original Cards 4, 5, 5b — what happened, what was deployed) are dropped entirely. If a "historical market events" module is built later, those go there. --- ## 6. Intervention decision tree — REFRAMED User pushback: "you can't ask for e.g. 'Q2' in one of the cards, then it doesn't have sufficient context to answer the question." Reframed: instead of testing recall of the Q-number, test recall of the *information source* needed to answer that decision step. ### Card 6 — Information source: rate channel **Front:** To determine whether an intervention transmits through the *rate channel* (vs only the FX level), what do you check? **Back:** Whether offsetting OMOs were announced the same week. Sterilized = no rate channel. Unsterilized = FX intervention *is* the monetary policy move. ### Card 6b — Information source: time horizon **Front:** To determine the likely *time horizon* of an intervention, what do you check? **Back:** The commitment-size language — capped (¥5-10T per episode, "as needed") or unlimited ("we will use all tools necessary," "no limit"). Capped = weeks-to-months pause. Unlimited = years hold. ### Card 6c — Information source: positioning consequence (the reframed Q3) **Front:** To determine whether existing positions will *profit or lose* from the intervention, what do you assess? **Back:** Whether the CB is pushing back against the cycle (positions lose) or letting the cycle work (positions profit). Pushing back = real-economy cost to the CB from the cycle's direction. Letting it work = cycle is doing what the CB wants, no reason to intervene. ### Card 6d — Information source: Trade 3 viability **Front:** To determine whether a *capitulation bet* (long-horizon intervention hold) is viable, what two conditions must both be present? **Back:** (1) Capacity is *finite* (CB will eventually exhaust). (2) Real-economy justification is *present* AND *eroding* (the policy reason is breaking down, not absent). ### Card 6e — Information source: Trade 2 viability **Front:** To determine whether *positioning reset* (Trade 2) is viable, what do you check? **Back:** Whether positioning is crowded in the direction the intervention pushes against. Crowded = unwind forced → Trade 2 fires. Uncrowded = no unwind, no Trade 2. **Audit note on Card 6 family:** now each card asks a *concrete operational question* (what do you check?) and the back is the *information source + the interpretation rule*. The user can answer each card without recalling the framework's Q-numbering — they recall the principle (rate channel = OMO check, time horizon = commitment language, etc.). Interference prevention: each front's discriminator ("rate channel," "time horizon," "profit or lose," "capitulation bet," "positioning reset") anchors the recall to a distinct operational question. --- ## 7. Sterilized/unsterilized PM diagnostic **Front:** When you see "BOJ/MOF/SNB/Treasury announces FX intervention," what do you check in the OMO schedule to determine sterilized vs unsterilized? **Back:** **Sterilized** = offsetting OMO announced same day → monetary base unchanged → CB signaling FX view only. **Unsterilized** = no offsetting OMO → monetary base changes → FX intervention *is* monetary policy. --- ## 8. Trade mappings (the user-flagged card, atomic split) ### Card 9 — Trade 1 mapping **Front:** Intervention where existing positions lose AND positioning is uncrowded. Which trade structure fires? **Back:** **Trade 1 only.** Fade the cycle, slow unwind. Defined-horizon (weeks), not capitulation bet. ### Card 9b — Trade 1+2 mapping **Front:** Intervention where existing positions lose AND positioning is crowded. Which trade structure fires, and what's the canonical case? **Back:** **Trade 1 + Trade 2 combined.** Crowded positioning unwind + intervention forces unwind. Canonical case: **BOJ Sept 2022**. ### Card 9c — Trade 3 mapping **Front:** Intervention where existing positions lose AND commitment is unlimited AND real-economy justification is eroding. Which trade structure fires, and what's the canonical case? **Back:** **Trade 3.** Long-horizon capitulation bet. Canonical case: **SNB 2011-2015** (held 3.3 years despite fighting global anti-EUR cycle, real-economy erosion in early 2015). ### Card 9d — Rare cases dropped **Front:** What two cases does the intervention framework deliberately drop, and why? **Back:** Positions profit + crowded (Plaza-style cycle inflection) and positions profit + uncrowded (CB helping, no edge). Both uncommon; complicate framework without adding load-bearing value. --- ## 9. Capitulation bet cost/payoff ### Card 10 — Cost/payoff structure **Front:** A capitulation bet "costs small most of the time, pays big at capitulation." What is the *cost* during the wait, and what is the *payout* at capitulation? **Back:** **Cost:** slow bleed from carry + opportunity cost + time decay, compounds during wait. **Payout:** capitulation move large enough to compensate for the entire bleed. ### Card 10b — Why positive expectancy **Front:** A capitulation bet has positive expectancy because the cost and payout are bounded differently. Why? **Back:** **Bleed is bounded** (carry, time decay — slow but finite). **Payout is not bounded** (capitulation can overshoot). Insurance-selling / vol-selling structure, not symmetric directional. --- ## 10. Trade 2 requires positioning loss ### Card 11 — Two-condition rule **Front:** Trade 2 (positioning reset) requires what two conditions to fire? **Back:** (1) Positioning is crowded. (2) Intervention creates *positioning loss*. ### Card 11b — "With cycle, no inflection" exception **Front:** "Intervention with cycle, no inflection" — does Trade 2 fire? **Back:** **No.** Intervention reinforces (positions expand), doesn't unwind. No Trade 2 even if positioning is crowded. --- ## 11. Aligned-with-cycle intervention is rare **Front:** Why is intervention *aligned with* the global cycle rare? **Back:** CBs don't intervene to *help* the cycle. They intervene to push back when the cycle hurts domestic policy. If cycle is doing what the CB wants, let it work — no reason to intervene. --- ## 12. Why "cycle position" was the wrong question **Front:** The intervention framework originally asked "with cycle or against cycle?" Why was this reframed? **Back:** CBs have multi-dimensional goals (inflation, employment, financial stability, currency, reserves, political). The same intervention can be with-cycle on one dimension and against-cycle on another. The operational question is **positioning consequence**, not cycle label. --- ## Summary of revision 3 | Original | Revision 3 | Reason | |---|---|---| | Cards 1 / 1b / 1c | Cards 1 / 1b / 1c | unchanged from rev 2 | | Cards 2 / 2b | Cards 2 / 2b | unchanged | | Cards 3 / 3b | Cards 3 / 3b | unchanged | | **Card 4** (SNB event) | **DROPPED** | historical event recall, not concept | | Card 4b → Card 4 | Card 4 (reframed) | event becomes scaffolding, recall target = concept | | Card 4c → Card 4b | Card 4b (reframed) | event becomes scaffolding | | **Card 5** (BOJ event) | **DROPPED** | historical event recall | | **Card 5b** (BOJ magnitude) | **DROPPED** | historical event recall | | Card 5c → Card 5 / 5b | Card 5 / 5b (reframed) | event becomes scaffolding, recall target = concept | | Cards 6 (5 Q-cards) | Cards 6 / 6b / 6c / 6d / 6e (5 info-source cards) | reframed per user pushback | | Card 7 | Card 7 | unchanged | | Card 8 | Card 8 | unchanged | | Cards 9 / 9b / 9c / 9d | Cards 9 / 9b / 9c / 9d | unchanged | | Cards 10 / 10b | Cards 10 / 10b | unchanged | | Cards 11 / 11b | Cards 11 / 11b | unchanged | | Card 12 | Card 12 | unchanged | | Card 13 | Card 13 | unchanged | **Net cards: 27** (rev 3 final). Card count breakdown by section: - Section 1 (sterilized/unsterilized): 3 - Section 2 (channels): 2 - Section 3 (Treasury-Fed): 2 - Section 4 (SNB 2015 concepts): 2 - Section 5 (BOJ 2022 concepts): 2 - Section 6 (decision tree info-source): 5 - Section 7 (PM diagnostic): 1 - Section 8 (trade mappings): 4 - Section 9 (capitulation bet): 2 - Section 10 (Trade 2 conditions): 2 - Section 11 (aligned-cycle rare): 1 - Section 12 (cycle position reframed): 1 - **Total: 27** --- ## What's NOT in this batch (deferred) - Pure historical event recall (when SNB removed floor, what BOJ deployed on Sept 22 vs Oct 21, exact magnitudes) — deferred to a future "historical market events" module. Substrate stays in the teach file. - 3 SNB / BOJ specific magnitudes ($200bn, ¥9.2T, ¥5-10T) — scaffold only, not recall target. If you want them as recall later (separate module), flag and I'll add. --- ## Open questions for user review 1. **Scaffolding format.** I put historical context in *(parentheses italics)* on the backs of Cards 4, 4b, 5. This is meant to be *reference, not recall* — the user can ignore it for the recall test. If the parenthetical is still a leak source (user reads it and pattern-matches), I should cut. Confirm the parenthetical scaffolding is the right format vs cutting entirely. 2. **Card 6d back has 2-item list.** Two co-equal conditions (capacity finite + real-economy present-eroding). Per the back-side-list rule, this is borderline. The exception is "items are the recall target and the question asks for them." "What two conditions must both be present?" → yes, the 2 items ARE the recall target. Acceptable per exception. Confirm. 3. **Card 5b removed SNB date but kept SNB reference.** "SNB 2011-2015" is in the back as scaffolding. If this still tests recall of the specific case (not the principle), I should change to a generic reference. Confirm whether the case names (SNB, BOJ) as scaffolding is acceptable or whether the cards should be case-agnostic.