Dataset Viewer
Auto-converted to Parquet Duplicate
meta
dict
results
list
{ "symbol": "XAUUSDm", "timeframe": "M5", "bars": 70952, "start": "2025-07-31T00:00:00", "end": "2026-08-04T00:00:00", "price_low": 3281.59, "price_high": 5595.35, "band_fraction_of_increment": 0.05, "horizon_bars": 12, "horizon_minutes": 60, "cooldown_bars": 24, "control_offsets": [ 0.3, ...
[ { "increment": 10, "round": { "levels_tested": 231, "approaches": 7112, "decided": 6890, "reject": 3731, "break": 3159, "inside": 222, "reject_pct": 54.2 }, "control": [ { "levels_tested": 232, "approaches": 7115, "decided": 689...

Do round numbers act as support and resistance on XAUUSD?

Every gold analysis page publishes a list of support and resistance levels, and those lists cluster on round numbers. The claim is that price is more likely to turn there. This dataset is the test.

Headline finding

Approaches decided Rejected Rejection rate
Round levels 11,670 6,602 56.6%
Non-round control levels 23,332 13,094 56.1%

Difference +0.45 percentage points, 95% CI [-0.65, +1.55] — the interval contains zero on 35,002 decided approaches. None of the four increments tested ($10, $25, $50, $100) produced an edge distinguishable from chance.

Roundness carries no measurable information about where gold turns.

Method

70,952 five-minute XAUUSD bars from a live MetaTrader 5 feed, 2025-07-31 to 2026-08-04, price range $3,281.59 to $5,595.35.

An approach is the first bar trading into a band around a level after being outside it. The outcome is judged 60 minutes later: reject if price ends back on the side it came from, break if it closes through and stays through. A 24-bar cooldown prevents price oscillating around a level from contributing correlated readings.

Each round level is matched against two control levels offset 30% and 70% into the same increment — prices nobody draws on a chart. Both groups use an identical band, so the comparison within each increment is like for like.

A caveat that belongs on the front of this file

Rejection rises from 54.2% at $10 to 69.8% at $100. That is not evidence that larger levels are stronger. The band scales with the increment, so a $100 level needs a $5 move to count as a break against $0.50 at $10. Round and control share the band, so each row is sound; the trend across rows measures band width, not the market.

Files

  • round_number_levels.csv — one row per level type per increment, with approaches, decided, reject, break and rejection rate
  • round_numbers.json — full output including per-control-offset breakdowns and run metadata

Full write-up

https://techkick.me/blog/gold-round-numbers-support-resistance

Citation

CC BY 4.0. Derived from the XAUUSD bar archive, DOI 10.5281/zenodo.21973215.

Downloads last month
29