# Turtle Trading (System 1 and System 2) on daily EUR/USD - specification (RIP Backtest protocol v1.0 plus the 8 Oct 2026 rules)

Specification written and **frozen 2026-10-11 (approved by the owner the same day, draft 1 unchanged)**, before any Turtle level, signal or trade was computed on EUR/USD.

## Why this test
- Chosen by the owner on 2026-10-11 as the out-of-market check named in the gold Turtle post ("the same rules, unchanged, on markets and years we have not used"). It was chosen **after** seeing the gold result, and that is its purpose: if the rules carry an edge beyond gold's long rise, System 2 should also beat the control here. Stated in advance: a result on EUR/USD alone confirms nothing; a clear failure here weakens the gold result.
- Earlier tests on this EUR/USD file: the FX study, the 1-hour golden cross and the Fibonacci 61.8% pullback. No daily Turtle number on EUR/USD was looked at.

## Rules
**Identical to `tt_spec.md`** (frozen 2026-10-09) and to the readings listed at the top of `tt_lib.js`: N, System 1 with the skip rule and the 55-day failsafe, System 2, the 2N stop, the 10- and 20-day exits, pyramiding to 4 units, R = 2N at the first entry, one position at a time, long and short. Four variants: **EUR-TT-S1-1U, EUR-TT-S1-4U, EUR-TT-S2-1U, EUR-TT-S2-4U**. Only the market-specific values change:
- **One tick = 0.00001** (5-digit quotes).
- **Costs:** about **1 pip per round trip per unit** (0.15 pip on every fill plus 0.7 pip commission), the cost model of every EUR/USD test on the site.
- **Financing:** baseline none. Scenario A: **2% a year** of the entry price per calendar day held, charged to longs and shorts alike. Scenario B: **today's IC Markets EUR/USD swap**, if the owner provides the rates from the MT4 specification before the run (not a historical swap); otherwise not run.
- **Rates for scenario B (provided by the owner before the run, read 2026-10-11 09:43 Bucharest from the MT4 EURUSD specification):** swap type in points, **swap long -7.99, swap short +1.31**, triple swap on Wednesday, 5 digits (1 point = 0.00001), contract size 100,000. Per unit and night: long -0.0000799, short +0.0000131 in price (about -$7.99 and +$1.31 per lot). One rollover at each 17:00 New York, Monday to Friday, Wednesday counting three, as in the gold test.

## Data
- EUR/USD, IC Markets MT4 demo 1-minute bid history, **2004-01-01 to 2026-09-28 15:46 New York** (the file used by the Fibonacci test, SHA-256 in `mt4_prefix_hashes.json`), daily bars 17:00 to 17:00 New York, days with fewer than 30 minutes not used for levels or N.
- **Clock:** the labels are about one hour early in most weeks of 2004-2014 (clock audit, 9 Oct 2026). The run is repeated on the corrected copy (`clock_audit_fix.js`); gates and verdicts must match, or the difference is reported.
- **Holes (rule 3):** `data_holes_eur.json`; trades open across a suspect or unclear gap are counted, and every gate and verdict is recomputed without them.

## Periods, gates, control
- Periods by first entry: development 2004-2012, validation 2013-2019, holdout 2020-2026 (**already used** by the earlier EUR/USD tests).
- Six gates as in `tt_spec.md`, but **development t >= 2.5**: with the gold test, eight Turtle variants have now been tried in total (the protocol raises the bar above four). Fewer than 30 trades: Inconclusive.
- Control: the time-shifted entry control of `tt_spec.md`, 500 runs, seeds 1000 + run.

## Engine
`tt_lib.js` gets one additive change: the market (data, tick, cost, financing) becomes a parameter, with gold as the default. **Before any EUR/USD result:** the gold run must reproduce exactly (every trade and gate), and the self-test must pass again, including the independent re-implementation matched trade by trade on EUR/USD for all four variants.

## Before the article
Results go to the owner first; one article, one market, published only with the owner's agreement.