Smart Gold Hunter wins most of its trades and loses money anyway. On the vendor's default preset our run closes 10,416 trades across 23 years of gold, 63.3% of them winners, and finishes with a net of -1,418.55 USD. Before we retrofit any commission at all the tester header already reads -585.27 USD with a profit factor of 0.95, so this result does not depend on our cost assumption. The account simply ends below where it started.
One deterministic tester run of version 2.0 on XAUUSD# M1, 100% real ticks, first trade 2003-05-06 and last trade 2026-08-12. We dissect it deal by deal against the tester journal, so pairing confidence is exact, journal coverage is 1.0, chronology violations 0 and duplicate deal ids 0. The catalog protocol runs a fixed 0.01 base lot on a 100,000 USD frame at 1:5000, which is a measurement setting and not a recommendation. On top of the tester result we retrofit commission at 3.50 USD per lot per side, which costs this run 833.28 USD across 20,832 deals.
The exit census explains the win rate in one line. Of the 10,416 trades, 9,886 ended at a stop and only 6 at a take profit, with 524 closed by signal or time. Of those stop exits, 6,443 closed in profit, which is 65.2% of them. That is a trailing stop doing the work. It banks a large number of small winners while the other 3,443 stop exits close in loss, which is close to a third of all trades. The input panel sets the stop at 1,500 points against a take profit of 250.
A high win rate built this way says nothing about profitability, and here it says the opposite. Average trade result is -0.14 USD after costs. Our sensitivity table puts the same figure the other way around. The run needs 14 cents less cost per trade to reach break even, and half a dollar more per trade turns it into -6,626.55 USD.
The loss is not a direction problem. Long trades lose 808.41 USD across 5,462 trades and short trades lose 610.14 USD across 4,954. Across 24 calendar years in the run, 18 are negative. The EA also holds long and short at the same time in 585 moments, which our structure dimension rates red, and it never stacks more than 2 positions at once, so this is not a grid. It is a symmetrical scalper that pays more in stops than it collects in trails.
The drawdown picture is unusual and worth reading carefully. Our end of day analysis finds exactly one drawdown episode. It starts at the peak on 2003-05-06, the first day of the run, reaches its trough on 2024-01-26 and never recovers. 6,071 trading days underwater and a depth of 1,720.30 USD, which is only 1.72% of the measurement frame. Small in percent, permanent in direction. The compound growth rate of the run is -0.06% per year.
The vendor ships four presets and we audited each of them on the identical world, window and cost model. This is the table we would want before buying.
Two of the four lose money over the full window. The two that make money make very little of it, and the best of them needs only 25 cents of extra cost per trade to fall to zero. All four presets carry a red structure rating and a red prop fit rating in our audit. Choosing the right preset is not a detail here. It is the difference between a small loss and a small profit, and none of the four produces the curve a buyer expects.
The vendor names no recommended deposit, so our capital replay falls back to 1,000 USD. On that basis 12.7% of the 20,832 possible start points end in a dead account. The two profitable presets kill nothing on the same basis and Prop Scalper kills 9.4%. Note what that means. Dead accounts fall in step with trade count across all four presets, from 12.7% at 10,416 trades to nothing at 8,359 and 5,065.
On the prop firm side our replay flags the consistency rule. Best day share exceeds the allowed percentage in 3 years under the funded profile we model, which is a structural conflict rather than bad luck. The prop fit lamp is red for every preset.
Over the traded span gold rose from 343.02 to 4,369.13 in the entry price series of this run, a move of 1,173.72%. Holding 0.01 lots for those 23 years would have earned 4,026.11 USD. The default preset earned -1,418.55 USD. That comparison is approximated from entry prices and holding gold means sitting through every gold crash since 2003, so it is not a fair risk comparison. It is a scale comparison, and the scale is what matters when a system asks 10,416 trades of your attention.
Smart Gold Hunter was published on 2026-03-25 and has been updated 7 times, most recently on 2026-08-23. Our test window ends on 2026-08-12, before that last update, so this run has no out of sample window at all. Of the tested history 8,359 days and 10,273 of the 10,416 trades fall before the release date. The 141 days between release and the end of our window contain 143 trades and +174.42 USD at an 84.6% win rate, which looks much better than the whole. We report it and we do not weight it, because 143 trades of a 10,416 trade record decide nothing.
One more date matters. We measured version 2.0. The market listing moved on to version 3.3 by 2026-08-26, and the vendor changed which preset is the default. Our numbers are version specific and dated, which is exactly why we publish the version next to every figure.
The deal level census, the preset matrix and the survival replay need full deal reconstruction, which is what we sell. Any catalog audit is $19, and EAs with several audited presets or versions come as one bundle. Your own tester report? The browser check is free.