Short answer: across 72 versions on the 15-minute, 1-hour and 4-hour charts, 1 lost reliably and 1 passed our bar, and that one's return beyond gold's own rise is uneven across the years (breakdown below). We treat it as unproven.
We also tried to extend this test to all eight timeframes (1-minute to weekly). That run's random-series check raised one false alarm, so by the rule we set before running it, that run does not count. We are not reporting its numbers as results, and we will not re-run the check until it passes.
Read this first, what this test can and cannot tell you
- Gold only (XAUUSD), Deriv's price feed and chart candles on the UTC clock, January 2017 to September 2026 (9 years 8 months). Brokers on a different clock draw different 4-hour and daily candles; Deriv's daily chart includes a short Sunday candle.
- One precise, written definition of the concept. Your own version may differ, the exact rules are below so you can check.
- Every trade pays the broker's spread (at least 15 points, measured from real ticks). The spread around the daily reopen is wider than that, which makes results look slightly better, not worse.
- "Average per trade" is in basis points: 1 bp = 0.01% of the price (about $0.43 per 1-ounce position at $4,300).
- This is a record of what one rule did on past data. It is not advice, not a signal, and not a promise about the future.
What we tested
Swing points with 5 candles each side (and a wider 50-candle version). The first close beyond the latest unbroken swing is a break: a CHoCH if the trend was the other way, otherwise a BOS. Buy on bullish breaks, sell on bearish ones, at the next open; exits after 4, 12 or 24 candles. Charts: M15, H1, H4.
Results (5-candle swings, exit after 12 candles)
| Chart | Label | Direction | Trades | Win rate | Average per trade (bp, after spread) | Verdict |
|---|---|---|---|---|---|---|
| M15 | BOS | bullish → buy | 2581 | 47.8% | +0.76 | no edge (not significant) |
| M15 | BOS | bearish → sell | 2219 | 43.9% | -0.31 | no edge (not significant) |
| M15 | CHoCH | bullish → buy | 2774 | 48.4% | -0.80 | no edge (not significant) |
| M15 | CHoCH | bearish → sell | 2773 | 45.4% | -2.71 | no edge (not significant) |
| H1 | BOS | bullish → buy | 680 | 50.7% | +5.73 | no edge (not significant) |
| H1 | BOS | bearish → sell | 527 | 47.8% | +3.42 | no edge (not significant) |
| H1 | CHoCH | bullish → buy | 651 | 50.5% | +4.00 | no edge (not significant) |
| H1 | CHoCH | bearish → sell | 648 | 43.2% | -4.63 | no edge (not significant) |
| H4 | BOS | bullish → buy | 198 | 55.1% | +19.45 | too few trades to judge |
| H4 | BOS | bearish → sell | 149 | 48.3% | -3.62 | too few trades to judge |
| H4 | CHoCH | bullish → buy | 183 | 53.0% | +10.60 | too few trades to judge |
| H4 | CHoCH | bearish → sell | 183 | 46.4% | +1.65 | too few trades to judge |
The one version that passed, and why we don't trust it yet
Buying a bullish BOS on the 1-hour chart and exiting after 24 candles: 533 trades, 55.2% winners, +14.69 bp per trade after spread (t +3.45). Year by year:
| Year | Trades | Win rate | Average per trade (bp, after spread) |
|---|---|---|---|
| 2017 | 49 | 55% | -0.97 |
| 2018 | 44 | 55% | +6.66 |
| 2019 | 57 | 51% | +3.10 |
| 2020 | 61 | 67% | +32.33 |
| 2021 | 55 | 44% | +0.42 |
| 2022 | 52 | 48% | +4.89 |
| 2023 | 53 | 58% | +13.77 |
| 2024 | 56 | 57% | +5.05 |
| 2025 | 65 | 66% | +46.98 |
| 2026 | 41 | 44% | +26.60 |
Its return beyond gold's own rise, by period (bp per trade): 2017-19 +0.4 (150 trades), 2020-22 +11.4 (168 trades), 2023-24 +2.5 (109 trades), 2025-26 +27.5 (106 trades). In the stricter eight-timeframe accounting it would not clear the bar (t 3.45 against 3.65).
The one that loses
M15 CHoCH sell, exit after 4 candles: 2785 trades, 41.9% winners, -2.38 bp per trade, negative in 10/10 years.
How we tested
- The rules were written down and dated before the test ran; nothing was tuned afterwards.
- Signals use closed candles only; every entry is at the next candle's open. One trade at a time per version; a crossing signal that arrives while a trade is open is skipped.
- Many versions were tried, so the bar for "real" is raised to match (Bonferroni, two-sided 5%), and a result must also hold in at least 70% of years, in both halves of the sample, and beat gold's own rise over the same years.
- The same test runs on a random price series. If the random series "passes", the whole test is thrown out, and we say so. Random-series check for this 3-timeframe test: clean (strongest random result t 2.09).
Think we got it wrong? Show us. Send your rules, your data source and your result (form fields: your definition · timeframe · data source and dates · costs you charged · your result and sample size · a link or file). If your test holds up, we publish the correction on this page, dated, with your name if you want it.