Phase 5 / 07:00 UTC study
XAUUSD London opening range, tested on real M1 data
We built a fixed 07:00-07:29 UTC gold range on 1,362 qualifying weekdays and observed the next four hours without filling missing candles or assuming an intrabar path.
In 1,362 fixed-clock tests, the 07:00 UTC XAUUSD 30-minute range broke on 99.6% of days. The high broke first on 51.1% of ordered breaks, and resolved 0.50R continuation measured 70.6%. These are price-event rates, not strategy returns.
What the fixed 07:00 UTC / London test actually found
The 30-minute box qualified on 1,362 UTC weekdays. Its median high-low size was $3.85 per ounce, with a middle-half interval of $2.75 to $6.12. During the next four clock hours, at least one boundary broke on 1,356 tests, or 99.6%. A narrow opening box followed by a much longer observation window naturally creates a high touch rate; the number does not describe an executable win rate.
First direction was balanced. The high broke first on 51.1% of ordered first breaks, and the median first break arrived 6 minutes after the range ended. Both boundaries broke during the same evaluation window on 48.1% of tests. The last observed bid closed back inside the original box on 26.8%. Those path statistics show why “the range broke” is incomplete without a stop, target, and ordering rule.
Under the 0.50R ordered definition, 910 tests extended half a box beyond the first-broken boundary before 379 reached the opposite side first. After excluding 0 same-minute ambiguous and 67 unresolved paths from that resolved rate, continuation measured 70.6%. The underlying counts remain visible so a reader can choose a different denominator.
| Target | Eligible | Target first | Opposite first | Ambiguous | Unresolved | Continuation* | False break* |
|---|---|---|---|---|---|---|---|
| 0.25R | 1,356 | 1,094 | 237 | 0 | 25 | 82.2% 95% CI 80.0-84.2% |
17.8% |
| 0.50R | 1,356 | 910 | 379 | 0 | 67 | 70.6% 95% CI 68.1-73.0% |
29.4% |
| 1.00R | 1,356 | 634 | 519 | 0 | 203 | 55.0% 95% CI 52.1-57.8% |
45.0% |
*Resolved denominator only: continuation-first plus opposite-first. Same-minute ambiguity and unresolved observations are reported separately.
07:00 UTC / London compared with the 13:30 UTC / U.S. clock
The matching 13:30 UTC / U.S. 30-minute sample has 1,357 tests and a median opening range of $7.49. Its any-break rate is 98.1% and its resolved 0.50R continuation rate is 72.9%. On 1,357 dates where both clocks qualified, the two opening-range sizes had a Pearson correlation of 0.655. The 13:30 UTC range was larger on 87.5% of matched dates.
This is a paired descriptive comparison, not evidence that one clock should always be traded. The boxes occur at different points in the day, inherit different news exposure, and use different prices. The first ordered direction agreed across the two clocks on only 49.8% of dates with usable first breaks in both, which is close to an even split.
| UTC weekday | Tests | Median range | Any break | 0.50R continuation* | 0.50R false break* | Both sides |
|---|---|---|---|---|---|---|
| Monday | 271 | $3.97 | 100.0% | 73.5% | 26.5% | 40.6% |
| Tuesday | 276 | $3.97 | 99.6% | 68.7% | 31.3% | 50.0% |
| Wednesday | 275 | $3.72 | 100.0% | 70.2% | 29.8% | 49.1% |
| Thursday | 273 | $3.78 | 100.0% | 69.7% | 30.3% | 52.4% |
| Friday | 267 | $3.75 | 98.1% | 70.9% | 29.1% | 48.3% |
*Resolved ordered outcomes. These weekday slices were inspected after the full sample and are descriptive, not independently confirmed signals.
How to turn the observation into a real backtest
- Freeze the exact clock, range length, quote source, and daylight-saving policy.
- Specify whether a wick, close, buffer, or retest triggers entry.
- Model spread and slippage on both entry and exit, especially around scheduled U.S. releases.
- Keep ambiguous M1 paths unresolved or replay them from finer data; do not award them to the preferred outcome.
- Choose the rule on a training period, then evaluate it once on later untouched dates.
One exact source, fully fingerprinted
The Phase 5 analyzer reads the production XAUUSD binary directly. It verifies the four-byte candle count, expected byte length, every timestamp, and every OHLC row before calculating a range. The published JSON includes aggregate tables plus all 8,149 qualifying clock-and-duration observations.
| Source | M1 candles | Bytes | UTC coverage | SHA-256 |
|---|---|---|---|---|
| XAUUSD_M1.bin | 1,871,487 | 37,429,744 | 2021-01-03 to 2026-04-30 | b3472e52f51bae343be941a8e8fb6dbdbc8749896c907b6ef77144b9b8c5cd63 |
0 duplicate timestamps0 out-of-order rows0 non-minute timestamps0 invalid OHLC rows
Open the Phase 5 evidence JSON
The file header declares 1,871,487 rows and its 37,429,744 bytes match exactly. Calendar coverage is 98.6% across eligible UTC weekdays. Missing periods are disclosed and never filled.
Exactly how every opening range was tested
1. Fix the clock
The two anchors are 07:00 UTC and 13:30 UTC. They never shift for daylight saving. “London” and “New York” are search-friendly clock labels, not assertions that one venue opens at the same UTC minute all year.
2. Build the box
The analyzer takes the highest bid high and lowest bid low in the first 15, 30, or 60 minutes. Every expected opening-range minute must be present; a gap disqualifies that test.
3. Observe four hours
The next 240 clock minutes form the evaluation window. At least 210 observed M1 candles are required. High-only, low-only, both-side, and no-break tests all remain visible.
4. Preserve order
Target-versus-opposite ordering starts on the minute after the first break. If both events share one later M1 candle, the record is ambiguous because OHLC cannot reveal the intrabar sequence.
Continuation targets are 0.25R, 0.50R, and 1.00R beyond the first-broken boundary, where R is that test's opening-range size. A high break and low break are strict inequalities; merely touching the recorded boundary is not counted.
Continue through the Phase 5 evidence
Replay the same market in the free XAUUSD backtester. Add your own entry, costs, stop, exit, and holdout period before treating any historical pattern as a strategy.
Questions answered by this dataset
Is a boundary-break rate the win rate of an XAUUSD strategy?
No. It reports whether an observed bid high or low crossed the opening-range boundary during the next four hours. No executable entry, stop, exit, costs, or position size is included.
Does the study adjust London and New York clocks for daylight saving?
No. The anchors stay fixed at 07:00 UTC and 13:30 UTC for reproducibility. The session names are clock-window shorthand.
What does 0.50R continuation mean?
After the first ordered break, a later complete M1 candle must extend half of that test’s opening-range size beyond the broken boundary before an earlier complete minute crosses the opposite boundary.
Can an AI crawler verify the XAUUSD calculations?
Yes. https://fxabsolute.com/research-data/phase-5-evidence.json publishes the source fingerprint, methodology, validation, exclusions, aggregate statistics, and every qualifying observation.