# Adjustments — Macro Study Program

Rules the mentor follows to adapt pace based on the competence map.

**Expertise over speed (user-set 2026-06-25, restructured 2026-06-26 to block-based architecture).** No content is skipped to meet a calendar deadline. Topic blocks extend until all sub-topics land at Working+ per rule 11. Program length is determined by content coverage, not a calendar target. **Block-based architecture (added 2026-06-26):** the program is a sequence of blocks (`blocks/` directory) rather than weeks. The user elects how many blocks to cover per session based on time/energy. Verification blocks (every 5-7 content blocks) consolidate prior batch.

## Competence status definitions

| Status | Meaning | Implication |
|---|---|---|
| **Mastered** | Answer cold, no hesitation, correct through mechanism | Skip in future blocks; maintain via Anki only |
| **Strong** | Correct with minor hesitation | Standard coverage; don't dwell on basics |
| **Working** | Correct with help or partial answer | Standard coverage |
| **Weak** | Wrong, or can't articulate | Targeted drilling in next 2 blocks |
| **Untested** | Not yet probed | Probe at next opportunity |

## Adjustment rules

### Topic-level

1. **3+ "Mastered" on a topic area** → remove from block planning. Maintain via Anki.
2. **"Working" for 2+ blocks with no improvement** → add Anki reps + 1 dedicated block on it.
3. **"Weak" on a topic** → next 2 blocks include targeted drilling on that topic before any new material.
4. **All topics in a phase area are "Mastered" or "Strong"** → compress that phase, advance to next phase.

### Verification gate (added 2026-06-25, user-set standard)

The user has stated the verification standard: **no Weak competence-map item is left unverified, and no item is assumed to improve on its own without verification.** Three rules codify this:

14. **Drill-first block template when Weak items are pending.** If a block opens with one or more Weak items on the competence map that are scheduled for the current or next phase, the block structure is drill-first (per the macro-mentor skill "Drill-first session template"). Drill block expands to fill most of the time; new material is deferred to the next block. Synthesis blocks are skipped until drill items land.

15. **Verification before block advance.** No advance from one block to the next until all competence-map items scheduled for the current block are at Working or higher. The check happens at verification blocks (every 5-7 content blocks) — if any item is still Weak at end of verification block, the next block opens with a drill prefix until the item moves. **Silent carry-forward of Weak items across a block boundary is disallowed.** If a Weak item must be carried forward, the user explicitly accepts it with documented reasoning in `weakness-register.md`.

16. **Backlog review (Weak items from prior calibration).** Weak items logged at calibration but not addressed in subsequent blocks are addressed either:
    - (a) as pre-work at the start of the block that needs them (probed at the start of the relevant block), OR
    - (b) on a dedicated review cadence (verification blocks include a backlog slot if the block's own items are clean).
    The user does NOT assume self-improvement on these items. Verification via cold-recall probe is required before the item moves to Working. Backlog items that haven't been re-probed in 4+ blocks get flagged at the Sunday planning cron.

### Block-level

5. **Block ends with user demonstrating 80%+ of probes correct** → reduce next block's planned material by 20%.
6. **Block ends with user demonstrating <50% of probes correct** → add a review block, extend next block by 15min, slow phase advance.
7. **Block material consistently below demonstrated level by 2+ blocks of competence** → skip ahead. Note in `session-log.md`. This is the "wrong-vehicle" guard — curriculum addiction is real.
8. **Phase 6 takes over** → when user is doing live trades that exercise the material, weekly synthesis replaces structured blocks for that topic area.

### Drift detection (NEW — per Q18)

The user has a stated end goal: world-class expert-level macro hedge fund trader. Drift away from topics that build toward that goal is the failure mode. Two rules:

9. **Drift warning** — 3+ consecutive blocks focused on a direction not on the curriculum path:
   - Mentor reads `coverage-tracker.md` and identifies what topics have been bypassed.
   - Mentor notifies the user at block end: "We've spent 3 blocks on X. Topics in the original plan that haven't been touched: [list]. Want to redirect?"
   - Does NOT redirect automatically. Gives the user the choice.

10. **Goal-alignment reset** — sustained drift AND the new direction is not on the path to "world-class macro hedge fund trader":
    - "On path" = anything that improves (a) component knowledge of traded/monitored instruments, (b) cross-asset linkage understanding, (c) macro framework depth, (d) process discipline.
    - "Off path" = pure academic theory without market application, backtesting/systematic strategy work, trading psychology pop-science without process backing, asset classes outside the universe, anything that doesn't improve a, b, c, or d.
    - Mentor recommends returning to the original plan. Asks before redirecting.
    - Tracked in `coverage-tracker.md` "Drift summary" section.

11. **Coverage balance check** — at every verification block:
    - Compare `coverage-tracker.md` (what's been covered) to `curriculum.md` (what should be covered).
    - If a phase has been entirely bypassed for 4+ blocks, raise as a planning concern in the next block's plan.
    - If 3+ phases have bypassed topics, recommend phase-skipping review: should the user deprioritize those phases or accelerate them?

### Cadence-level

12. **3+ consecutive missed blocks** → trigger calibration: mentor reads `competence-map.md` and `session-log.md`, summarizes state, asks if user wants to reset calibration or continue.
13. **Weekly Anki review queue > 100 cards pending** → flag in next block; suggest focused 2-day catch-up before new material.

## Status transitions

A topic moves through statuses as:
- Untested → (probe) → Mastered / Strong / Working / Weak
- Working → (more reps) → Strong or Mastered
- Weak → (drilling) → Working or Strong
- Strong → (over time + Anki) → Mastered

The mentor updates the competence map at the END of every block based on what was demonstrated during the block.

## Block architecture (added 2026-06-26)

The program is a sequence of blocks (not weeks). Each block is roughly day-sized and covers one topic area. Blocks are numbered sequentially (`blocks/001-091` in the current program) and live in the `blocks/` directory. Verification blocks are interspersed every 5-7 content blocks.

**Block types:**
- **Content block** (most blocks): new material + drill as needed
- **Verification block** (007, 015, 022, 030, 037, 043, 049, 051, 072, 079, 085, 091): re-probe flags from prior batch, write synthesis note, update state files, regenerate next-block plan

**Per-block structure (60 min default for content blocks; 90-120 min for verification blocks):**
- 00-05: Pre-block probes (Anki retention + cold-recall on prior block's flagged items)
- 05-45: Main material (drill or new content depending on flags)
- 45-55: Dialogue / synthesis (conditional on drill landing)
- 55-60: Anki + log + carry-forward to next block

**Probe-format rule (added 2026-06-26 Session 7):** Probe prompts contain the question only — pass criteria are the mentor's evaluation rubric and are NOT included in the prompt. Including pass criteria in the prompt reveals the answer (user-flagged 2026-06-26 Re-probe 1). The mentor evaluates the user's answer against the pass criterion silently, then surfaces feedback (pass/fail + what landed vs what missed) only after the user has answered.

**Carry-forward rule:** if any planned material from a block wasn't completed, it goes at the TOP of the next block's main material block — not appended as "if time permits." Silent skipping is disallowed. Explicit skip (with reasoning from user or mentor) is the only exception.

**The active block** is tracked in `_session-state/<DATE>-resume.md` frontmatter (`current_block` + `block_status` fields). Updated at each block's session close. (Session 41 triage 2026-08-09: pointer moved from deleted `_meta/current-block.md` to resume file.)

**Schedule doctrine (added 2026-06-26):** block numbers are orientation only. Topic blocks advance per verification gate. No silent calendar advance. Backlog items unaddressed 4+ blocks get flagged at verification blocks per rule 16.