XAUUSD support and resistance levels, priced
Every chart package draws support and resistance lines. None of them says how often price gets there. XAUUSD is at 4616.68 and its hourly ATR is 19.9 dollars, as of 2026-08-21 19:00 UTC.
A support or resistance level is a price you expect to matter, and the only thing that makes one worth drawing is whether price arrives at it. That is what the table below measures — not whether a level "holds", which is a judgement nobody can settle, but how often it is reached at all, by distance, from this market's own history.
| Level | Price | Away, dollars | ATR | 4h | 8h | 24h | 48h | Typical time |
|---|---|---|---|---|---|---|---|---|
| round 4800 | 4800.00 | ▲ 183.3 | 9.22 | 0% | 0% | 2% | 7% | 69h |
| round 4700 | 4700.00 | ▲ 83.3 | 4.19 | 1% | 2% | 20% | 36% | 37h |
| round 4600 | 4600.00 | ▼ 16.7 | 0.84 | 38% | 53% | 78% | 84% | 5h |
| yesterday's high | 4540.80 | ▼ 75.9 | 3.82 | 2% | 4% | 25% | 40% | 31h |
| yesterday's close | 4526.19 | ▼ 90.5 | 4.55 | 1% | 2% | 18% | 32% | 37h |
| value area high | 4515.00 | ▼ 101.7 | 5.12 | 1% | 1% | 14% | 27% | 42h |
| round 4500 | 4500.00 | ▼ 116.7 | 5.87 | 0% | 1% | 10% | 22% | 47h |
| yesterday's POC | 4490.40 | ▼ 126.3 | 6.35 | 0% | 1% | 8% | 19% | 50h |
| value area low | 4469.10 | ▼ 147.6 | 7.42 | 0% | 0% | 6% | 14% | 56h |
| yesterday's low | 4450.36 | ▼ 166.3 | 8.37 | 0% | 0% | 4% | 10% | 60h |
| last week's high | 4449.45 | ▼ 167.2 | 8.41 | 0% | 0% | 4% | 10% | 61h |
| last week's low | 4310.38 | ▼ 306.3 | 15.41 | — | — | — | — | — |
Typical time is the median wait among the times price did reach that far, so it says nothing about the times it never got there — read it next to the percentage, not instead of it. The spread around it is wide: of the times a line 1.5 ATR away is reached at all, a quarter take more than a day.
Test your own support or resistance level
The same table, the same lookup. Type any price and it answers for that instead.
Stop loss and take profit: which one prints first
Two levels, and the only thing that matters is which prints first. That is not something the table above can answer — it gives the odds of reaching each one, and says nothing about the order. This is measured separately, on 1.9M hourly bars: how often the target came before the stop. It covers thirteen instruments rather than the table's seventeen: this sub-study predates the most recent instrument and has not been re-run since.
Entry at the market, both distances measured from there, no spread and no slippage. A bar that covers both levels does not say which it printed first, and those are counted as the stop — the pessimistic reading. At normal distances that is a rounding detail; at a quarter of an ATR it is a third of all cases, and the box says so when it happens.
Where the numbers come from
2.2M hourly bars across 17 instruments, 2000 to 2026. Built on everything before 2024 and then checked against everything after, which it had never seen: across 216 cells the median error was 0.6 percentage points and the largest 4.5.
Distance is measured in ATR rather than in dollars, so a quiet Tuesday and a payroll Friday ask the same question. Below eight hours the hour of day is used as well — worth up to 21 points there, almost nothing beyond a day, and taken from your clock rather than from the build's. Upside and downside are kept apart and come out within about two points of each other at nearly every distance. An earlier version of this page said a level below was reached noticeably more often; that gap was a bug in how the forward window was cut, not a fact about the market.
These levels are not claimed to be special. Every one we tested — the point of control, the value-area edges, yesterday's extremes — is reached about as often as a line drawn the same distance away on the other side. That is not a caveat, it is the whole argument: it is precisely why one table can answer for any line, including one you drew yourself.
Descriptive only. No entry, no exit, no spread, no position sizing, and not a forecast — a base rate for how far price travels, nothing more.
Turnmarks marks the hour a push runs out on 17 charts, and publishes what happened after every past marker, per instrument, so the tool can be judged before it is paid for. How a turn marker is decided, or open the charts — free plan, no card.