Rule Collider

Backtests · GOLD · x_usdjpy_atr_z · long

Does x_usdjpy_atr_z above -0.05177 work on GOLD? 2023 to 2026 backtest

GOLD long when USDJPY ATR vs 50-bar average is above the threshold, tested at five thresholds taken from the feature’s own distribution on GOLD over this window: its 10th, 25th, 50th, 75th and 90th percentiles. Window 2023-09-30 to 2026-09-30, the last 30 percent held out of sample; stop 2.5 ATR, target 3 R, held overnight, max spread 30 pips, one-minute confirmation 2 pips, no hour filter. Fills and spread as measured on a demo fleet (how a result is judged). The direction and the exit are the ones the research catalogue carried for this pair; the thresholds are not.

Five thresholds, one rule shape

PercentileConditionIn sample R / tradesOut of sample R / tradestTrades to confirmLast quarter R / tradesVerdict
p10x_usdjpy_atr_z > -0.3816-187.8 R / 3429-169.9 R / 1328-2.9647-58.7 R / 449Active, net negative
p25x_usdjpy_atr_z > -0.2562-173.2 R / 3358-163.8 R / 1302-2.8672-58.1 R / 437Active, net negative
p50x_usdjpy_atr_z > -0.05177-163.1 R / 3112-137.9 R / 1174-2.5776-62.6 R / 395Active, net negative
p75x_usdjpy_atr_z > 0.2295-133.3 R / 2478-123.8 R / 930-2.5593-54.8 R / 302Active, net negative
p90x_usdjpy_atr_z > 0.5297-89.4 R / 1462-60.3 R / 582-1.51,024-25.6 R / 187No edge shown

R is the risk on one trade. The out-of-sample split carries the verdict: below 100 trades it is “no edge shown” whatever the t; a positive mean with t below 2 is “net positive, not yet distinguishable from chance”; the last column is how many trades that mean would need to reach t = 2.

Each threshold in full

x_usdjpy_atr_z > -0.3816 (p10)

Honesty panelActive, net negative

Traded 1328 times out of sample, net negative beyond chance (-169.9 R, -0.128 R per trade, t = -2.9); still active in the last quarter (-58.7 R over 449 trades).

In sample
2023-09-30 to 2025-11-05
-187.8 R
3429 trades · 32% wins
-0.055 R per trade · t = -1.9, within chance
about 3,777 trades to confirm this mean (-0.055 R per trade) at t = -2 · shown: 3429
spread 0.102 R per trade
Out of sample
2025-11-05 to 2026-09-30
-169.9 R
1328 trades · 29% wins
-0.128 R per trade · t = -2.9, beyond chance
about 647 trades to confirm this mean (-0.128 R per trade) at t = -2 · shown: 1328
spread 0.063 R per trade
Last quarter
last 91 days of the window
-58.7 R
449 trades · 29% wins
-0.131 R per trade · t = -1.7, within chance
about 638 trades to confirm this mean (-0.131 R per trade) at t = -2 · shown: 449
spread 0.084 R per trade
Costs. Spread 0.091 R per trade at 1.25 times the bar’s minimum (more than the trades made before spread); 796 fills in measured hot hours (01:00 at 3.9 times). Fill offset 0.08 stop-distances on every entry, measured on 732 demo fills. Note: the symbol's quoting precision changed on 2025-08-04; the spread field is converted with the point of each era. At 1.5 times the bar’s minimum the out-of-sample figure would read -186.7 R (-0.141 R per trade over 1328 trades).

Entry: stop order 2 pips past the signal close, cancelled after 4 bars; 4,996 signals placed, 4,757 filled, 239 expired; trigger state, an entry on every bar the conditions hold while flat.

Exit: stop 2.5 ATR (mean 90.5 pips, 2.50 ATR, 30 raised to the 10-pip floor); target 3 R; exits 66% at the stop, 22% at the target, 13% at the day end.

Filters: signals on every weekday; no entries 22:00 to 01:00 broker time (1,320 bars); max spread 30 pips (0 bars over it); M1 confirmation 2 pips (921 signals rejected).

Monthly R, x_usdjpy_atr_z > -0.3816 on GOLDOne bar per month: the R earned that month; the held-out months in the full colour.2023-10: +22.4 R over 134 trades2023-102023-11: -11.4 R over 133 trades2023-12: -1.4 R over 127 trades2024-01: -28.3 R over 161 trades2024-02: -32.5 R over 146 trades2024-03: +21.9 R over 97 trades2024-04: -33.0 R over 145 trades2024-042024-05: -36.8 R over 159 trades2024-06: -31.7 R over 146 trades2024-07: +9.6 R over 140 trades2024-08: -12.2 R over 163 trades2024-09: +12.1 R over 136 trades2024-10: +2.6 R over 149 trades2024-102024-11: -12.7 R over 134 trades2024-12: -34.2 R over 138 trades2025-01: +10.8 R over 124 trades2025-02: +2.6 R over 137 trades2025-03: +15.9 R over 81 trades2025-04: +4.9 R over 126 trades2025-042025-05: -19.2 R over 133 trades2025-06: -25.3 R over 136 trades2025-07: -17.6 R over 143 trades2025-08: -25.1 R over 144 trades2025-09: +29.2 R over 142 trades2025-10: +7.5 R over 149 trades2025-102025-11: +6.8 R over 124 trades2025-12: +2.0 R over 145 trades2026-01: +14.8 R over 111 trades2026-02: +5.0 R over 93 trades2026-03: -42.5 R over 80 trades2026-04: -24.5 R over 132 trades2026-042026-05: -11.5 R over 72 trades2026-06: -67.2 R over 128 trades2026-07: -31.3 R over 148 trades2026-08: +17.8 R over 148 trades2026-09: -45.2 R over 153 trades+29 R-67 R

x_usdjpy_atr_z > -0.2562 (p25)

Honesty panelActive, net negative

Traded 1302 times out of sample, net negative beyond chance (-163.8 R, -0.126 R per trade, t = -2.8); still active in the last quarter (-58.1 R over 437 trades).

In sample
2023-09-30 to 2025-11-05
-173.2 R
3358 trades · 32% wins
-0.052 R per trade · t = -1.8, within chance
about 4,287 trades to confirm this mean (-0.052 R per trade) at t = -2 · shown: 3358
spread 0.101 R per trade
Out of sample
2025-11-05 to 2026-09-30
-163.8 R
1302 trades · 29% wins
-0.126 R per trade · t = -2.8, beyond chance
about 672 trades to confirm this mean (-0.126 R per trade) at t = -2 · shown: 1302
spread 0.063 R per trade
Last quarter
last 91 days of the window
-58.1 R
437 trades · 29% wins
-0.133 R per trade · t = -1.7, within chance
about 614 trades to confirm this mean (-0.133 R per trade) at t = -2 · shown: 437
spread 0.084 R per trade
Costs. Spread 0.091 R per trade at 1.25 times the bar’s minimum (more than the trades made before spread); 773 fills in measured hot hours (01:00 at 3.9 times). Fill offset 0.08 stop-distances on every entry, measured on 732 demo fills. Note: the symbol's quoting precision changed on 2025-08-04; the spread field is converted with the point of each era. At 1.5 times the bar’s minimum the out-of-sample figure would read -180.2 R (-0.138 R per trade over 1302 trades).

Entry: stop order 2 pips past the signal close, cancelled after 4 bars; 4,923 signals placed, 4,660 filled, 263 expired; trigger state, an entry on every bar the conditions hold while flat.

Exit: stop 2.5 ATR (mean 90.4 pips, 2.50 ATR, 28 raised to the 10-pip floor); target 3 R; exits 66% at the stop, 22% at the target, 12% at the day end.

Filters: signals on every weekday; no entries 22:00 to 01:00 broker time (1,909 bars); max spread 30 pips (0 bars over it); M1 confirmation 2 pips (933 signals rejected).

Monthly R, x_usdjpy_atr_z > -0.2562 on GOLDOne bar per month: the R earned that month; the held-out months in the full colour.2023-10: +26.1 R over 124 trades2023-102023-11: -12.8 R over 133 trades2023-12: +4.2 R over 122 trades2024-01: -28.8 R over 158 trades2024-02: -29.6 R over 144 trades2024-03: +22.5 R over 96 trades2024-04: -32.0 R over 138 trades2024-042024-05: -39.0 R over 160 trades2024-06: -28.3 R over 137 trades2024-07: +7.7 R over 141 trades2024-08: -17.9 R over 159 trades2024-09: +10.8 R over 137 trades2024-10: +1.5 R over 147 trades2024-102024-11: -15.1 R over 134 trades2024-12: -36.0 R over 139 trades2025-01: +12.9 R over 123 trades2025-02: +3.6 R over 132 trades2025-03: +20.4 R over 77 trades2025-04: +3.5 R over 124 trades2025-042025-05: -15.8 R over 133 trades2025-06: -25.3 R over 133 trades2025-07: -18.7 R over 138 trades2025-08: -21.0 R over 144 trades2025-09: +26.4 R over 143 trades2025-10: +13.5 R over 136 trades2025-102025-11: +9.4 R over 116 trades2025-12: -1.7 R over 145 trades2026-01: +19.1 R over 108 trades2026-02: +1.5 R over 93 trades2026-03: -40.3 R over 78 trades2026-04: -25.3 R over 132 trades2026-042026-05: -12.3 R over 72 trades2026-06: -62.0 R over 127 trades2026-07: -36.1 R over 148 trades2026-08: +13.6 R over 141 trades2026-09: -35.6 R over 148 trades+26 R-62 R

x_usdjpy_atr_z > -0.05177 (p50)

Honesty panelActive, net negative

Traded 1174 times out of sample, net negative beyond chance (-137.9 R, -0.117 R per trade, t = -2.5); still active in the last quarter (-62.6 R over 395 trades).

In sample
2023-09-30 to 2025-11-05
-163.1 R
3112 trades · 32% wins
-0.052 R per trade · t = -1.7, within chance
about 4,205 trades to confirm this mean (-0.052 R per trade) at t = -2 · shown: 3112
spread 0.099 R per trade
Out of sample
2025-11-05 to 2026-09-30
-137.9 R
1174 trades · 30% wins
-0.117 R per trade · t = -2.5, beyond chance
about 776 trades to confirm this mean (-0.117 R per trade) at t = -2 · shown: 1174
spread 0.062 R per trade
Last quarter
last 91 days of the window
-62.6 R
395 trades · 29% wins
-0.158 R per trade · t = -1.9, within chance
about 424 trades to confirm this mean (-0.158 R per trade) at t = -2 · shown: 395
spread 0.082 R per trade
Costs. Spread 0.089 R per trade at 1.25 times the bar’s minimum (more than the trades made before spread); 681 fills in measured hot hours (01:00 at 3.9 times). Fill offset 0.08 stop-distances on every entry, measured on 732 demo fills. Note: the symbol's quoting precision changed on 2025-08-04; the spread field is converted with the point of each era. At 1.5 times the bar’s minimum the out-of-sample figure would read -152.6 R (-0.130 R per trade over 1174 trades).

Entry: stop order 2 pips past the signal close, cancelled after 4 bars; 4,510 signals placed, 4,286 filled, 224 expired; trigger state, an entry on every bar the conditions hold while flat.

Exit: stop 2.5 ATR (mean 91.5 pips, 2.50 ATR, 27 raised to the 10-pip floor); target 3 R; exits 66% at the stop, 22% at the target, 12% at the day end.

Filters: signals on every weekday; no entries 22:00 to 01:00 broker time (3,651 bars); max spread 30 pips (0 bars over it); M1 confirmation 2 pips (805 signals rejected).

Monthly R, x_usdjpy_atr_z > -0.05177 on GOLDOne bar per month: the R earned that month; the held-out months in the full colour.2023-10: +21.5 R over 121 trades2023-102023-11: -13.8 R over 125 trades2023-12: -4.5 R over 108 trades2024-01: -34.9 R over 146 trades2024-02: -22.5 R over 135 trades2024-03: +19.6 R over 95 trades2024-04: -11.5 R over 122 trades2024-042024-05: -26.1 R over 139 trades2024-06: -29.4 R over 129 trades2024-07: +12.7 R over 130 trades2024-08: -14.2 R over 150 trades2024-09: -3.8 R over 127 trades2024-10: -2.2 R over 144 trades2024-102024-11: -5.9 R over 125 trades2024-12: -28.8 R over 122 trades2025-01: +7.7 R over 108 trades2025-02: +0.3 R over 131 trades2025-03: +22.6 R over 78 trades2025-04: -3.1 R over 124 trades2025-042025-05: -15.5 R over 125 trades2025-06: -32.8 R over 125 trades2025-07: -21.5 R over 132 trades2025-08: -15.0 R over 121 trades2025-09: +38.5 R over 113 trades2025-10: +5.3 R over 131 trades2025-102025-11: +19.1 R over 95 trades2025-12: +2.1 R over 138 trades2026-01: +14.9 R over 99 trades2026-02: +9.5 R over 85 trades2026-03: -34.8 R over 76 trades2026-04: -22.1 R over 116 trades2026-042026-05: -13.6 R over 61 trades2026-06: -56.3 R over 115 trades2026-07: -30.9 R over 132 trades2026-08: +11.2 R over 126 trades2026-09: -42.9 R over 137 trades+38 R-56 R

x_usdjpy_atr_z > 0.2295 (p75)

Honesty panelActive, net negative

Traded 930 times out of sample, net negative beyond chance (-123.8 R, -0.133 R per trade, t = -2.5); still active in the last quarter (-54.8 R over 302 trades).

In sample
2023-09-30 to 2025-11-05
-133.3 R
2478 trades · 32% wins
-0.054 R per trade · t = -1.6, within chance
about 3,967 trades to confirm this mean (-0.054 R per trade) at t = -2 · shown: 2478
spread 0.091 R per trade
Out of sample
2025-11-05 to 2026-09-30
-123.8 R
930 trades · 29% wins
-0.133 R per trade · t = -2.5, beyond chance
about 593 trades to confirm this mean (-0.133 R per trade) at t = -2 · shown: 930
spread 0.054 R per trade
Last quarter
last 91 days of the window
-54.8 R
302 trades · 27% wins
-0.182 R per trade · t = -2.0, within chance
about 314 trades to confirm this mean (-0.182 R per trade) at t = -2 · shown: 302
spread 0.065 R per trade
Costs. Spread 0.081 R per trade at 1.25 times the bar’s minimum (more than the trades made before spread); 448 fills in measured hot hours (01:00 at 3.9 times). Fill offset 0.08 stop-distances on every entry, measured on 732 demo fills. Note: the symbol's quoting precision changed on 2025-08-04; the spread field is converted with the point of each era. At 1.5 times the bar’s minimum the out-of-sample figure would read -133.7 R (-0.144 R per trade over 930 trades).

Entry: stop order 2 pips past the signal close, cancelled after 4 bars; 3,589 signals placed, 3,408 filled, 181 expired; trigger state, an entry on every bar the conditions hold while flat.

Exit: stop 2.5 ATR (mean 92.3 pips, 2.50 ATR, 19 raised to the 10-pip floor); target 3 R; exits 66% at the stop, 22% at the target, 12% at the day end.

Filters: signals on every weekday; no entries 22:00 to 01:00 broker time (6,120 bars); max spread 30 pips (2 bars over it); M1 confirmation 2 pips (653 signals rejected).

Monthly R, x_usdjpy_atr_z > 0.2295 on GOLDOne bar per month: the R earned that month; the held-out months in the full colour.2023-10: +8.5 R over 92 trades2023-102023-11: -12.0 R over 94 trades2023-12: -1.3 R over 103 trades2024-01: -25.9 R over 110 trades2024-02: -21.6 R over 97 trades2024-03: +14.0 R over 92 trades2024-04: -10.4 R over 108 trades2024-042024-05: -25.5 R over 117 trades2024-06: -9.8 R over 92 trades2024-07: +11.1 R over 99 trades2024-08: -16.6 R over 114 trades2024-09: -5.0 R over 106 trades2024-10: -8.6 R over 109 trades2024-102024-11: -8.8 R over 96 trades2024-12: -23.0 R over 96 trades2025-01: +5.8 R over 93 trades2025-02: +12.0 R over 104 trades2025-03: +28.3 R over 57 trades2025-04: +12.2 R over 90 trades2025-042025-05: -18.7 R over 106 trades2025-06: -28.1 R over 95 trades2025-07: -26.2 R over 110 trades2025-08: -20.8 R over 100 trades2025-09: +25.3 R over 89 trades2025-10: +17.9 R over 99 trades2025-102025-11: +20.5 R over 84 trades2025-12: -15.4 R over 121 trades2026-01: +8.9 R over 93 trades2026-02: +4.2 R over 66 trades2026-03: -20.1 R over 56 trades2026-04: -15.9 R over 82 trades2026-042026-05: -14.1 R over 42 trades2026-06: -43.2 R over 94 trades2026-07: -35.0 R over 102 trades2026-08: -8.3 R over 100 trades2026-09: -11.5 R over 100 trades+28 R-43 R

x_usdjpy_atr_z > 0.5297 (p90)

Honesty panelNo edge shown

No edge shown out of sample: -60.3 R over 582 trades is -0.104 R per trade, within what chance produces (t = -1.5); about 1,024 trades would be needed to confirm this mean at t = -2, shown: 582.

In sample
2023-09-30 to 2025-11-05
-89.4 R
1462 trades · 32% wins
-0.061 R per trade · t = -1.4, within chance
about 3,043 trades to confirm this mean (-0.061 R per trade) at t = -2 · shown: 1462
spread 0.089 R per trade
Out of sample
2025-11-05 to 2026-09-30
-60.3 R
582 trades · 29% wins
-0.104 R per trade · t = -1.5, within chance
about 1,024 trades to confirm this mean (-0.104 R per trade) at t = -2 · shown: 582
spread 0.052 R per trade
Last quarter
last 91 days of the window
-25.6 R
187 trades · 28% wins
-0.137 R per trade · t = -1.1, within chance
about 578 trades to confirm this mean (-0.137 R per trade) at t = -2 · shown: 187
spread 0.060 R per trade
Costs. Spread 0.078 R per trade at 1.25 times the bar’s minimum (more than the trades made before spread); 254 fills in measured hot hours (01:00 at 3.9 times). Fill offset 0.08 stop-distances on every entry, measured on 732 demo fills. Note: the symbol's quoting precision changed on 2025-08-04; the spread field is converted with the point of each era. At 1.5 times the bar’s minimum the out-of-sample figure would read -66.4 R (-0.114 R per trade over 582 trades).

Entry: stop order 2 pips past the signal close, cancelled after 4 bars; 2,148 signals placed, 2,044 filled, 104 expired; trigger state, an entry on every bar the conditions hold while flat.

Exit: stop 2.5 ATR (mean 92.9 pips, 2.50 ATR, 7 raised to the 10-pip floor); target 3 R; exits 67% at the stop, 22% at the target, 11% at the day end.

Filters: signals on every weekday; no entries 22:00 to 01:00 broker time (9,935 bars); max spread 30 pips (7 bars over it); M1 confirmation 2 pips (364 signals rejected).

Monthly R, x_usdjpy_atr_z > 0.5297 on GOLDOne bar per month: the R earned that month; the held-out months in the full colour.2023-10: -11.2 R over 52 trades2023-102023-11: -6.8 R over 54 trades2023-12: -8.2 R over 58 trades2024-01: -14.8 R over 56 trades2024-02: -18.1 R over 55 trades2024-03: +0.9 R over 65 trades2024-04: +5.9 R over 68 trades2024-042024-05: -9.4 R over 73 trades2024-06: -11.8 R over 69 trades2024-07: +18.0 R over 74 trades2024-08: -14.7 R over 73 trades2024-09: -12.1 R over 63 trades2024-10: +11.7 R over 47 trades2024-102024-11: +4.5 R over 52 trades2024-12: -22.6 R over 52 trades2025-01: +7.0 R over 52 trades2025-02: +7.6 R over 60 trades2025-03: +11.3 R over 45 trades2025-04: +20.9 R over 40 trades2025-042025-05: -20.8 R over 65 trades2025-06: -16.0 R over 55 trades2025-07: -3.1 R over 58 trades2025-08: -4.0 R over 43 trades2025-09: +1.6 R over 61 trades2025-10: +0.9 R over 66 trades2025-102025-11: -4.1 R over 59 trades2025-12: +3.4 R over 64 trades2026-01: +17.0 R over 55 trades2026-02: +18.0 R over 42 trades2026-03: -20.8 R over 32 trades2026-04: -11.2 R over 54 trades2026-042026-05: -12.1 R over 28 trades2026-06: -31.1 R over 67 trades2026-07: -13.9 R over 65 trades2026-08: -9.9 R over 61 trades2026-09: -1.8 R over 61 trades+21 R-31 R

What x_usdjpy_atr_z measures

USDJPY ATR vs 50-bar average (ratio), on M5 bars. USDJPY ATR(14) minus its 50-bar average, divided by that average.

Test your own variant

Rule Collider runs one rule on one symbol over years of broker M5 data with the fills and the spread a demo fleet measured, and shows the out-of-sample figure first. Early access goes to the waitlist in signup order.

Join the waitlist

Historical simulation on broker M5 data with a measured fill offset, a measured spread factor, swap and a one-minute confirmation model. Past results do not predict future results. This is a research tool, not investment advice. Hours are broker time, the chart time in MT5. Built 2026-11-04 from runs ending 2026-09-30; refreshed monthly.