# Noise Area Breakout on Bitcoin, around the clock, with 15-minute checkpoints - specification (RIP Backtest protocol v1.0 plus the 8 Oct 2026 rules)

Specification written and **frozen 2026-10-15 (approved by the owner the same day, draft 1 unchanged)**, before any 15-minute band, signal or trade was computed. Once frozen, any change of a rule is documented and treated as a separate test; bugs may be fixed. The owner asked the same day for both tests to be reported in one article.

## Why this test, and what is already known
- **Chosen by the owner on 2026-10-15, after seeing the results of the 30-minute test** (`na_btc_spec.md`, frozen and run the same day: both variants Buried, 1 of 6 gates, the long-and-short version near zero before costs). The owner asked for the same rule with decisions every 15 minutes.
- **This is a separate test, not a correction.** The published rule (Zarattini, Aziz and Barbon, 2024) decides every 30 minutes; the 30-minute test applied it as published and stays the main result. Checking every 15 minutes is a different setting of the same rule.
- **Stated in the article:** this setting was chosen after the 30-minute results on the same data were seen, so a good result here would be weak evidence; the 30-minute result is reported next to it whatever this one shows. The noise-area rule now has 22 variants on the site (18 on ES, NQ and gold, 2 at 30 minutes on Bitcoin, these 2).
- **The holdout (2024 to September 2026) has now been opened three times** on this file (Turtle, trend pullback, the 30-minute noise-area test) and is reported as used.

## What changes against the 30-minute test (everything else is identical)
- **Checkpoints every 15 minutes:** 00:15, 00:30, ... 23:45 UTC (95 a day), up to the last one at least 15 minutes before the 24:00 close.
- **sigma(m)** is computed at each 15-minute checkpoint, from the previous 14 days that have a minute at m (at least 10).
- **Entries and the trailing stop** are checked at every 15-minute checkpoint; fills at the open of the next minute, as before.
- **Two variants:** NA-BTC-24H15-LS (long and short) and NA-BTC-24H15-L (long only).

## Unchanged from `na_btc_spec.md`
The UTC day (00:00 to 24:00, every day of the week); the previous close; the bands; the entry, exit, flip and end-of-day rules; VWAP with the real minute volume; the Binance BTCUSDT 1-minute file (same SHA-256); costs of 0.05% of the price on every fill and no slippage beyond gaps; the funding scenario; results in percent of the entry price; rule 3 with gaps of 10 minutes or more; periods (development 2017-09-15 to 2021-06-30, validation 2021-07-01 to 2023-12-31, holdout 2024-01-01 to 2026-09-30); the six gates with **development t >= 2.5**; the control (each day's first entry moved by one random circular offset of at least 5% of the days, 500 runs, seeds 1000 + run).

## Engine and self-test (before the run)
- `na_btc_lib.js` and `na_btc_ref.js` take the checkpoint step as an input, 30 minutes unless told otherwise. `na_lib.js` stays untouched.
- Self-test, which must pass before any 15-minute result is printed: everything of the 30-minute self-test; **the 30-minute trades reproduced exactly** (same file as the published run); hand-worked 15-minute cases (95 checkpoints, sigma at 15-minute points, a decision at a quarter past); the independent re-implementation matched trade by trade for both 15-minute variants; the control at offset 0.

## Before the article
Results go to the owner first, next to the 30-minute results. The article covers both tests, the 30-minute one as the published rule and this one as the follow-up, and is published only with the owner's agreement.