Phase 1 · Weeks 1–4 · 30 Days

Manufacture evidence, not opinion.

The backtest phase produces the dataset the rest of the system runs on. 35+ logged cases. Win Rate computed. R:R computed. Personal Playbook built from your data — not someone else's setups. One Gate Check at the end. Pass or repeat.

35+ cases · 4 review days · ~2 hrs/day · 2-hour fallback

§A

Daily Operating Protocol

~2h standard · 2h fallback also valid

▾

STANDARD

~2 hours, in order

120 min

DRC

§1 Temperature Check · sets Risk Mode (5–10 min)

DRC

§2 Market Context · BTC bias · BTC.D · ETH/BTC · regime · 4H/1H structure · narrative (10–15 min)

ATLAS

Reference 7-condition checklist (after Day 24, Atlas is the source) (10 min)

Open Replay → Pre-Trade case → advance bars → Post-Trade + Execution + Outcome (60–90 min)

DRC

§5 EOD Process + Adherence Score · §6 Self-Regulation Audit (10–15 min)

FALLBACK

2-hour cap (tight days)

120 min

DRC

§1 Temperature (5 min)

DRC

§2 Market Context — short form (10 min)

Core backtest — minimum 1 case, ideally 2 (90 min)

DRC

EOD + TCJ close (15 min)

Fallback is a deliberate downgrade, not a skip. The DRC streak must remain intact across all 30 days.

§B

What to Record · Where

DRC · TCJ · Weekly Review · Atlas

▾

Event	DRC	TCJ	WR	Atlas
Morning start	D§1 Temp → Risk Mode	—	—	Read-only (Day 25+)
Market read	D§2 Context	—	—	Bias + Regime check
Open BT case	—	T§0 + §1 Pre-Trade	—	Read §3 checklist
Close BT case	—	T§2 + §3 + §4	—	Queues weekly signal
No trade today	D§3 → No-Trade	TNo-Trade Case	—	—
End of session	D§5 + §6 EOD	—	—	—
Loss pattern	—	—	—	A§7 Error Library
Review day (7,14,21,30)	§1 + §5 + §6	Sync Report	W§1–§8 full	§6 Best/Worst · §7 reconcile

Week 1 · Setup & Learning

Days 1–7 · Environment + Replay + Example cases

▾

Week 2 · 20 BTC Backtest Trades

Days 8–14 · Log only · no judgment

▾

Week 3 · ETH + Advanced Conditions

Days 15–21 · ETH + BTC.D + false-setup detection

▾

Week 4 · Playbook + Readiness

Days 22–30 · Atlas build · Drawdown · Gate Check #1

▾

§C

Valid Backtest Case · 10 Criteria

All must be true · else Case Incomplete

▾

Case Type = Backtest in TCJ §0 · Phase = 1

Ticker = BTC/USDT or ETH/USDT only — no other pairs in Phase 1

Pre-Trade written before outcome bar visible in Replay — no result contamination

Thesis contains structure ref + trigger ref + driver ref (all three)

Invalidation is price-precise — a number or "1H close below X" — not a feeling

R:R ≥ 1.0 at planned TP1

Post-Trade closed same session — no overnight "I'll finish tomorrow"

Sections 2 + 3 + 4 all filled at close

Outcome computed from R Realized, not assigned manually

Lesson uses process language, not P&L language

§D

Invalid Cases · Recognized Failure Modes

Rejected from rollups · flagged for repair

▾

Failure	Why It's Rejected
Thesis without trigger	Can't separate setup from prediction
Invalidation as feeling	No measurable failure point
Outcome bar visible at Pre-Trade	Result-knowledge contamination
Grade C taken as normal trade	Violates DRC §3 block · backtest-study only
Edited Thesis after entry	TCJ §3 #12 self-deception breach
Lesson written in P&L language	Wrong layer — process, not P&L
No-Trade with no category	Avoidance, not discipline
Pair outside Atlas scope	Setup outside system (only BTC, ETH)
Timeframe below 1H referenced	Class A1 ban (T3)
Skipped §2 / §3 / §4 at close	Case Incomplete

§E

Labeling · Wins · Losses · No-Trades · Breaches

Outcomes by R Realized · Plan Match orthogonal

▾

Label	Definition	Counts In
Win	R Realized ≥ +0.8R · all sections complete	Win Rate numerator
Loss	R Realized ≤ −0.9R · all sections complete	Win Rate denominator
Breakeven	R Realized −0.2R to +0.2R	Excluded from Win Rate
Scratch	−0.9 to −0.2 or +0.2 to +0.8 (manual exits)	Flagged in Execution Score
No-Trade	Setup not taken · valid reason category	Discipline Counter
Rule Breach	Any of 12 TCJ §3 breach items	Process Score component d
Case Incomplete	Failed §C validity rules	Excluded · repair in WR

Plan Match is orthogonal to outcome.
A Loss with Plan Match = Full and Was Thesis Correct? = Y is a clean loss — it strengthens the playbook, not weakens it. A Win with Plan Match = Broken is a sloppy win — it does not strengthen the setup; it surfaces a behavioral risk for the Error Library.

§F

Worked Examples

3 good · 5 bad · all on BTC / ETH

▾

§G · TCJ Sync Report

Run weekly · Days 7, 14, 21, 30

Cases Logged

Wins

Losses

Breakeven

Scratch

No-Trade Cases

Sum of Win R (+R)

Sum of Loss R (−R, signed)

Sum of Planned R:R

Rule Breaches (week)

Win Rate—

Avg R Realized—

Avg R:R Planned—

Expectancy (per trade, R)—

Set Day # above 0 to assess against the case-count floor.

§H · Gate Check #1 — Day 30

Tap the dot to mark each criterion green. Five of these five must be green to enter Phase 2. Criterion #5 activates in Phase 3 only.

Win Rate ≥ 45%

across 35+ Backtest cases · TCJ Sync Report

Avg R:R ≥ 1.2

across closed cases · TCJ Sync Report

Atlas Playbook Completeness ≥ 70%

§1, §2, §3, §5 complete for all three setups

Drawdown Protocol Written

Atlas §0 toggle = Y · non-empty §1 text · 3-loss rule locked

DRC Streak 30 / 30

no skipped days, even No-Trade days

Gate #1 Result

Mark criteria to compute decision

§I

How to Think During Backtesting

8 frames · execution language only

▾

Each case is a sentence the system can read. Not a story you tell yourself. If TCJ would reject the case, the trade didn't happen.

Result is downstream of process. A Win that broke plan is information about you, not the setup. A Loss that respected invalidation is information about the setup, not you.

The bar at Pre-Trade is the only state you're allowed to know. Replay's job is to enforce this. Yours is not to cheat it.

No-Trade is a valid output. A day with one A-grade No-Trade is stronger than a day with three impulse cases.

Numbers come from quantity, not certainty. 35 cases produce a Win Rate. 8 cases produce a feeling. Hit the case floors before drawing conclusions.

The Error Library is the most valuable file in the system. Wins teach a setup; losses teach a setup; broken-plan trades teach you yourself.

Atlas comes from data, not opinion. Days 22–24 codify the patterns Days 8–21 already proved. Atlas does not invent.

A Gate Check #1 fail is information, not failure. The 14-day extension exists because entering Phase 2 with bad data is worse than a delay.