Does the first hour set the day?
Two reports this category sells by name, and they are the same question twice: opening candle continuation — if the first hour is green, how often does the session close green? — and the initial balance, the first hour's high and low. We ran both on 68,678 sessions across 17 markets. One of them finds nothing at all; the other fires on 99.7% of days.
Where the open is
Before any of this can be measured you have to decide what "the first hour" means, and getting that wrong is how the naive version of these reports goes wrong. There is no first hour of a currency day in UTC. So FX and metals use the London open and the index markets use their own cash open, both resolved through each city's clock for every single day so daylight saving cannot shift them. Crypto is excluded: it has no open, and inventing one is exactly the mistake to avoid.
Continuation, and the control that kills it
The obvious way to test this is to count: of the days whose first hour closed green, how many closed green overall? That number will land somewhere near 50% and look like something. The control is what makes it readable — we re-pair each day's first hour with a different day's session. The same number of green first hours, the same number of green sessions, and the link between them destroyed.
| Market | Sessions | Green first hour → green session | Control | Δ | z | One side of the IB breaks | Control | Neither side |
|---|---|---|---|---|---|---|---|---|
| EUR/USD | 5,914 | 48.3% | 48.5% | -0.1 | -0.14 | 47.2% | 44.9% | 0.27% |
| GBP/USD | 5,218 | 50.4% | 49.6% | +0.8 | 0.78 | 47.0% | 44.5% | 0.33% |
| USD/CHF | 3,240 | 51.1% | 51.4% | -0.3 | -0.26 | 46.3% | 44.4% | 0.37% |
| USD/JPY | 4,574 | 52.5% | 52.0% | +0.4 | 0.44 | 46.6% | 43.5% | 0.42% |
| AUD/USD | 5,721 | 48.8% | 49.1% | -0.4 | -0.42 | 45.1% | 43.9% | 0.14% |
| USD/CAD | 5,931 | 49.1% | 50.0% | -0.8 | -0.93 | 34.9% | 35.7% | 0.05% |
| NZD/USD | 5,169 | 50.3% | 50.4% | -0.1 | -0.15 | 44.4% | 43.8% | 0.10% |
| EUR/JPY | 5,239 | 50.6% | 50.4% | +0.2 | 0.22 | 51.5% | 48.5% | 0.52% |
| GBP/JPY | 4,171 | 52.9% | 52.1% | +0.8 | 0.72 | 50.6% | 47.2% | 0.26% |
| EUR/GBP | 2,454 | 47.4% | 47.8% | -0.4 | -0.24 | 48.1% | 47.1% | 0.20% |
| Gold (XAU/USD) | 6,043 | 51.8% | 50.5% | +1.3 | 1.46 | 37.2% | 35.2% | 0.13% |
| Silver (XAG/USD) | 5,187 | 52.6% | 53.1% | -0.5 | -0.56 | 32.4% | 35.5% | 0.04% |
| US30 (Dow) | 2,811 | 51.9% | 53.2% | -1.3 | -0.98 | 68.1% | 67.8% | 1.89% |
| NAS100 | 1,663 | 57.0% | 56.2% | +0.9 | 0.5 | 70.3% | 69.9% | 2.41% |
| SPX500 | 1,663 | 55.3% | 53.8% | +1.5 | 0.89 | 66.2% | 65.9% | 1.56% |
| GER40 (DAX) | 2,711 | 52.8% | 52.7% | +0.1 | 0.06 | 63.1% | 62.0% | 1.25% |
| UK100 (FTSE) | 969 | 49.7% | 51.9% | -2.2 | -1 | 63.8% | 62.0% | 0.52% |
The initial balance fires on almost every session
The first hour's range survives the rest of the session untouched on a median of 0.33% of days. A filter that admits 99.7% of sessions is not a filter — it is the same defect the opening range has, and it is fatal in the same way.
The half that does survive is the split between a session that breaks one side and one that breaks both — the trend-day story the report is really sold on. One side only happens on a median 47.23% of sessions against 44.94% for a session that keeps its own bars, its own hours and its own sizes and loses only their direction. That is a real gap of +1.16 points — and it is smaller than every spread on our board.
What we would actually say
- Decide what "the open" means before you measure anything about it. For FX there is no natural one, and picking an arbitrary hour puts the arbitrary hour in your answer.
- Continuation is nothing. Seventeen markets, sixty-eight thousand sessions, not one weak flag.
- The first hour is a range, not a signal. It tells you roughly how far this session is likely to travel — the axis that survives a holdout — and nothing about which way.
Full method and every market in section 20 of the evidence page.
Other reports this category sells, tested
Every claim we have tested, including the ones that held, is on the reports index.
We publish the tests that failed as loudly as the ones that held. Every number on this site carries its sample size, and every claim carries what it was measured against.
Read the whole evidence page →