Gold Trade Pro is one of the better tester runs in our catalog. Over 23 years of real gold ticks (2003 to 2026, daily chart), v6.7 produced 6,305 trades at a profit factor of 1.83 and a relative balance drawdown of 1.20% on the deep measurement frame. The data quality lamp is green, the log pairs exactly, and not one trade breaches the per position loss limit of the IQ Capital profile at full tested size. Then the prop replay runs the consistency rule and the lamp turns red anyway. In 8 separate years the run breaks the 30% consistency requirement, and that single mechanism is the finding of this audit.
Standard catalog protocol on our documented data world. XAUUSD tick history from 2003.05.05 to 2026.08.22, every tick based on real ticks, 10 ms execution delay, fixed 0.01 base lot on a 100,000 USD measurement frame at 1:5000. The frame is a measurement setting that keeps the tester from censoring the trade sequence. Capital math is answered on the vendor recommended 1,000 USD basis. The tester journal covers the run completely and the engine reconstructed all 6,305 trades with exact pairing confidence.
The structure forensics count up to 18 simultaneous positions and 4,745 stacked entries, 1,154 of them split tickets. The direction is the healthier kind: 3,302 adds go with the move and only 274 against it, with a median size escalation of exactly 1.0. This is pyramiding into strength, not averaging a losing basket, the same geometry that distinguished The Gold Reaper from the grid class. The martingale detector stays green.
The calendar is where the profile cracks. 80% of the total profit was earned in 71 trading days, which is 3.9% of the 1,805 days on which the EA traded. Three calendar years end negative. A system that earns in rare bursts collides with any rule that demands evenness, and the consistency rule is exactly that. The red lamp and the concentration caution are the same fact seen from two sides.
The replay sweep across position sizings finds zero account deaths at every step from full size down to one eighth, and zero per position loss breaches anywhere. On pure survival this run is clean. The withdrawal line at full tested size prints 54% a year on the vendor deposit basis. What fails is the shape of the equity curve inside each payout period: 8 years contain at least one window where a single stretch contributes more than 30% of the profit. A funded account on those terms does not blow up, it gets its payouts questioned.
The report head prints a Sharpe ratio of 3.89. Our end of day recomputation lands at 1.44, with costs taking 9.7% of gross profit (commission 770.18 USD, swap 678.63 USD over the run). The gap is aggregation, not fraud, and it is why we recompute instead of copying. One number deserves genuine credit: the slice after the last vendor update (2026-05-20 to 2026-08-11, the only part that is guaranteed out of sample) nets +856.73 USD over 58 trades at a profit factor of 2.44. That is 84 days against 51 recorded updates and 23 years of visible history, so we weigh it as a good sign, not as proof.
The full sizing matrix, the year by year consistency table and the payout replay sit in the catalog audit. Your own tester report goes through the same engine: the browser check is free, the full audit is $19.