--- type: methodology status: active version: 1.0 created: 2026-06-15 source: Process.pdf §4 — manual discretionary decisions --- # Decision Process — Manual Discretionary **The book is run on judgment, not on a systematic signal.** Every entry decision is the result of the trader (you) reading news, scanning the upcoming calendar, and assessing the relevant securities against a checklist — then writing a pre-trade thesis that captures the reasoning before entry. This is the opposite of a rules-based system. The cost of that approach is that decisions are not backtestable in aggregate and the only audit trail is the pre-trade thesis. The benefit is that you respond to information the rules can't codify. ## When a trade consideration starts A trade is considered when **at least two of the following are simultaneously true**: 1. The macro calendar (see `07-calendar/macro-book.ics`) has a relevant Tier 1/2 event inside the next 1-10 sessions, OR a structural release (COT, GLD NAV) is due. 2. An instrument on the watched list ([[../06-universe/Coverage Universe]]) is at or approaching a level flagged in the per-instrument note (`06-universe/traded/.md`). 3. Something in news flow (Reuters, Bloomberg, FT, central bank communications, X/Twitter macro accounts) materially shifts the regime or pricing for an instrument you already watch. 4. Cross-asset confirmation — a move in one instrument is consistent with moves in correlated instruments per the bucket definitions in [[risk-framework]]. A single trigger is not enough. Two is the floor. Three is preferred. ## The assessment checklist (per instrument) When considering a specific instrument, work through this in the per-instrument note (`06-universe/traded/.md`) before writing a thesis. The instrument note has a dedicated "Assessment checklist" section that gets filled in. | Question | Where the answer lives | |---|---| | Where is price vs the technical levels I watch? | Instrument note → Levels section | | What's the bucket (rates / FX / commodity / equity-beta) and am I at the bucket cap? | [[risk-framework]] §3 | | What's the COT / positioning extreme for this instrument? | COT history + instrument note | | What Tier 1/2 events are inside the next 1-10 sessions that affect this instrument? | `07-calendar/macro-book.ics` | | What's the consensus / market pricing for those events? | Pre-event research in `03-research/event/` | | Is the trade inside the Tier 1 event blackout window (no new entries inside 60 min of release)? | Calendar + thesis §9 | | What's the invalidation — a fact, not a price? | You write it, in the thesis | | Does this risk 25-75bp of NAV at stop, with portfolio open-risk ≤ 250bp? | Thesis §4, §9 | | Am I duplicating an existing view? | Book check in thesis | If you can't answer the questions above with current data, the assessment is incomplete — do not enter. Read more, wait for the next event, or skip. ## What a "trade idea" looks like in this process A trade idea is a 3-5 sentence note in `01-journal/.md` that says: *what instrument, what direction, what level, what catalyst, what's the invalidation fact.* This is the cheap pre-stage. Most trade ideas die here. A trade idea becomes a thesis when the pre-event research file (`03-research/event/.md`) is filled in and the instrument note checklist is complete. ## From trade idea → thesis → execution ``` trade idea in 01-journal/.md (cheap, one-liner, often) ↓ pre-event research in 03-research/event/ (Tier 1 events only — required) ↓ per-instrument note checklist filled in (06-universe/traded/.md) ↓ pre-trade thesis in 02-trades//thesis (1 page, written BEFORE entry, timestamped) ↓ execution logged in 02-trades//execution-log (order type, fill, slippage) ↓ in-trade management in 02-trades//management-log (every adjustment + reason) ↓ post-mortem in 02-trades//postmortem (within 48h, A-F grade, lesson) ``` The thesis is the contract. Deviating from it without recording the deviation in the management log is a process violation. The post-mortem is the audit. The grade (A-F) is for process quality, not P&L. ## Anti-patterns (the things that will burn this process) 1. **The "I just feel it" trade.** No checklist items checked, no thesis written, no invalidation. Grade F in the post-mortem. 2. **The "I read the news and went long" trade.** Single-trigger. Needs a second trigger. 3. **The "I watched CNBC and they said…" trade.** News as confirmation is fine. News as the only trigger is not. 4. **Trading inside a Tier 1 event blackout.** Calendar is the source of truth. Hard rule, not a soft rule. 5. **Sizing up because the prior trade worked.** Process grade is independent of P&L. Halve sizing after a -5% peak-to-trough drawdown, full stop. 6. **Adding to a losing position without re-running the checklist.** The original thesis may be dead; the market has just been telling you so. Re-thesis or exit. 7. **Skipping the post-mortem because the loss was small.** Small unrecorded losses are how process drift starts. Grade every closed trade, even scratch trades. ## Versioning - v1.0 (2026-06-15) — initial. Supersedes the placeholder `decision-process.md` (now removed). Changes to this process are recorded in `90-archive/process-changelog.md` with a one-paragraph justification. Quarterly process audit checks: are the trades of the last 90 days consistent with this doc?