GOLD V1: tests and results
Below are the historical test results of 12 GOLD profiles over five years: this is not a forecast and not a guarantee of returns. The tables cover all closed trades of a profile, longs and shorts together, including the + and ++ signal classes.
Test results#
Period: September 16, 2021 – September 15, 2026.
| Timeframe | Trades | Win rate | PF | Drawdown (1×) | Loss streaks (all / LONG / SHORT) |
|---|---|---|---|---|---|
| 30 seconds | 1,710 | 73.51% | 1.205 | 11.24% | 4/3/5 |
| 1 minute | 1,552 | 70.43% | 1.242 | 10.40% | 5/4/5 |
| 2 minutes | 1,604 | 71.57% | 1.227 | 9.26% | 4/5/5 |
| 3 minutes | 1,572 | 75.76% | 1.202 | 10.11% | 4/5/5 |
| 5 minutes | 1,111 | 71.38% | 1.345 | 11.65% | 5/5/5 |
| 15 minutes | 685 | 67.15% | 1.333 | 7.99% | 4/5/4 |
| 30 minutes | 305 | 59.02% | 1.445 | 5.23% | 4/4/4 |
| 45 minutes | 255 | 61.57% | 1.469 | 9.52% | 4/5/5 |
| 1 hour | 203 | 68.47% | 1.818 | 8.96% | 5/3/4 |
| 2 hours | 119 | 62.18% | 2.264 | 4.20% | 5/3/4 |
| 3 hours | 123 | 75.61% | 2.252 | 5.21% | 3/3/4 |
| 4 hours | 109 | 74.31% | 1.749 | 8.23% | 3/4/4 |
Loss streaks are given in this order: all trades / LONG only / SHORT only. In subgroups, streaks can be longer.
Profit from closed trades is calculated with reinvestment at 100% of current capital (1x) per position, with no open trade at the end.
These are observed results, not a limit on future losses.
How to read the numbers#
Win rate is the share of profitable trades after model costs.
PF in this report is the sum of positive percentage results divided by the sum of negative results (ignoring the minus sign). A PF of 1.75 is not a 175% return.
Compare PF together with the number of trades: a high score on a small sample is less meaningful.
Each timeframe in GOLD V1 mode was tested separately. The model uses different logic on different timeframes: not only the scale changes, but also the trade parameters. So each timeframe is a separate strategy, and results from different timeframes cannot be combined into a single portfolio, even if the same trades appear on different timeframes.
Longs and shorts separately#
Values are given as "number of trades / win rate / PF". Longs and shorts on the same timeframe differ noticeably, so the overall result does not show the full picture by direction.
| Timeframe | LONG | SHORT |
|---|---|---|
| 30 seconds | 1,031/76.04%/1.196 | 679/69.66%/1.223 |
| 1 minute | 1,035/75.46%/1.259 | 517/60.35%/1.183 |
| 2 minutes | 1,221/75.51%/1.197 | 383/59.01%/1.372 |
| 3 minutes | 1,177/77.57%/1.224 | 395/70.38%/1.147 |
| 5 minutes | 646/65.48%/1.387 | 465/79.57%/1.256 |
| 15 minutes | 512/64.45%/1.292 | 173/75.14%/1.470 |
| 30 minutes | 150/55.33%/1.523 | 155/62.58%/1.278 |
| 45 minutes | 149/65.77%/1.581 | 106/55.66%/1.227 |
| 1 hour | 106/72.64%/1.749 | 97/63.92%/2.050 |
| 2 hours | 55/65.45%/2.221 | 64/59.38%/2.343 |
| 3 hours | 65/76.92%/2.329 | 58/74.14%/2.107 |
| 4 hours | 70/75.71%/1.714 | 39/71.79%/1.872 |
Signal classes + and ++#
Rows within one direction overlap: "+ and above" includes "++", and both are part of the overall pool of longs or shorts. They should not be added together. Class statistics are a subset of all trades, not a separate test of trading only plus-rated signals. Values are given as "number of trades / win rate / PF"; "none" means the class is not available.
| Timeframe | LONG+ and above | LONG++ | SHORT+ and above | SHORT++ |
|---|---|---|---|---|
| 30 seconds | 697/79.91%/1.461 | 141/87.23%/2.549 | 324/73.15%/1.440 | 88/84.09%/2.651 |
| 1 minute | 701/79.74%/1.584 | 140/89.29%/3.432 | 230/66.96%/1.773 | 103/79.61%/3.113 |
| 2 minutes | 999/77.18%/1.322 | 172/85.47%/1.501 | 252/61.11%/1.580 | 130/63.85%/1.749 |
| 3 minutes | 886/79.46%/1.343 | 475/81.68%/1.558 | 174/76.44%/1.572 | 86/80.23%/1.950 |
| 5 minutes | 463/68.03%/1.495 | 172/73.84%/2.030 | 256/82.42%/1.511 | 111/84.68%/1.766 |
| 15 minutes | 358/67.04%/1.436 | 69/81.16%/3.052 | 83/83.13%/2.360 | 39/89.74%/4.288 |
| 30 minutes | 122/58.20%/1.711 | 37/70.27%/2.908 | 101/69.31%/1.932 | 50/76.00%/2.570 |
| 45 minutes | 72/70.83%/1.998 | 37/78.38%/2.983 | 72/62.50%/1.629 | 34/73.53%/2.715 |
| 1 hour | 81/76.54%/2.150 | 35/85.71%/3.949 | none | none |
| 2 hours | none | none | none | none |
| 3 hours | 44/81.82%/3.215 | 29/86.21%/4.535 | none | none |
| 4 hours | 53/81.13%/2.364 | 32/87.50%/3.849 | none | none |
Note
- Results are uneven over time. On 30s – 404 of 679 shorts fell in the incomplete 2026. On 2m – the overall PF in the incomplete 2021 was about 0.891; on 3m – in 2022 it was about 0.953: both below 1.
- Small samples are less meaningful. For example, the SHORT++ group on 15m has only 39 trades.
- In subgroups, loss streaks can be longer than the overall one. On 5m, in the LONG+ and above group, the streak reached 6 while the overall streak was 5.
What the test includes#
- The model uses 15-second Gold quotes.
- Costs: 0.003% commission per side, 0.20 USD spread per ounce, 0.05 USD slippage per ounce.
- Swap (rolling a position over to the next day) is not included.
Parameters were selected on the same history (September 16, 2021 – September 15, 2026). This is not an independent check on new data and not a guarantee of future results. Your broker's or exchange's quotes and conditions may change the outcome.
Why the numbers in the account table (ACCOUNT TABLE – MODEL) differ from the report
The table on the chart shows the best range of the last 10–50 trades, while the report covers all five years. PF in the report is based on percentage results, while PF in the table is based on USD and depends on position size. These values cannot be compared directly.