The failure this catalog is built to catch is averaging into losers, because that is the geometry that lets a single session take an account. Adaptive Gold Scalper does none of it, and the audit should say so before it says anything else. Volume never rises after a loss, 0.0% of the time. At most 4 positions were open at once, and long and short were never held together, 0 moments. Only 217 of 3,891 entries were stacked on an existing position, and 9 of those went against the open trade. Every one of the 3,891 trades closed on a stop, and our trailing detector fired, which is the mechanical signature of a stop that follows the price. Maximum drawdown across 23 years of gold ticks was 0.24%.
This is what a disciplined scalper looks like in a report. The problem is a different one.
Our drawdown episode table sorts by depth, and the deepest episode is also the one still running at the end of the test. The balance peaked at $100,511.15 on 2016-03-09. It bottomed at $100,273.07 on 2026-01-19 and it never recovered. The run finished at $100,381.48, still below the 2016 high, after 2,720 trading days underwater.
Split the 23 years at that date and the shape is plain. The first thirteen years produced $511.15. The following ten gave back $129.67. The audit's headline profit of $381.48 is what is left of a strategy that stopped working roughly a decade ago and kept trading anyway.
Gross before costs was $694.17. Our commission retrofit at $3.50 per lot per side charged 7,782 deals and took -$311.28. Swap took -$1.41 across 17 overnight trades. Net $381.48, and costs are 45% of gross profit. Over 3,891 trades that is under eighteen cents earned and eight cents paid per trade, with just under ten cents surviving.
Our sensitivity section makes the consequence explicit. The break even cost shock is $0.10 per trade. Ten cents of extra spread, slippage or commission per trade and 23 years of work is flat. At an average size of 0.01 lots that is not a large amount of friction to imagine.
In fairness the edge does not hang on a handful of lucky sessions. Remove the 20 most profitable trades out of 3,891 and the run still ends at +$191.55, which is 50.2% of the total. Nothing here depends on a few sessions in the way a grid does. The edge is real. It is simply too thin to pay for anything.
Adaptive Gold Scalper was published on 2026-01-12. Of the tested history, 8,238 days and 3,573 trades lie before that date, at +$278.41 and a profit factor of 1.19. The window from release to the last vendor update on 2026-07-19 covers 189 days and 284 trades at +$92.38, profit factor 1.51.
The window after the last update, the only slice of a backtest that is genuinely out of sample, is not empty here. It runs 24 days and 34 trades at +$10.69 and a profit factor of 2.16. Twenty four days and thirty four trades prove nothing on their own, and we will not pretend otherwise. Of the 20 audits in our catalog that carry a release split, 7 have nothing at all in that window.
Our withdrawal replay sweeps everything above the start balance to the owner on the last trading day of each month. Across 279 months it paid in 53 of them, $509.64 in total, which is 0.02% per year. The median payday was $6.74 and the best month was $46.78. The longest dry streak is 126 months, which is ten and a half years without a payout.
There is a detail in that replay worth reading twice. After the sweeps, the account ends at $99,872.04, below where it started. The owner took $509.64 out and the balance is short by the difference. That is not an error. It is what monthly skimming does to a strategy whose gains and losses arrive in that order.
Same instrument, same lot size, no strategy. Gold moved 1,047.09% across the tested span, from 384.484 to 4,410.36. Holding 0.01 lots through it would have made $4,025.88. The EA made $381.48, about ten times less, while opening 3,891 positions and paying commission on every one of them.
The comparison flatters neither side completely. Buy and hold carries overnight and weekend risk that this EA almost never takes, and it would have sat through drawdowns far deeper than 0.24%. Still, 59.7% of what profit exists here came from the long side, in a market that rose 11.5 times, and the total is a tenth of what the move itself paid.
Against IQ Capital Classic the account survives at full size and at every reduced size we test, with 0 floor breaches and 0 per position loss breaches throughout. That clean sheet is the normal case in this catalog rather than a distinction. 19 of our 25 audits come through the same sweep with zero breaches of either kind. The modelled withdrawal rate at full size is 0.02% per year, and below half size it rounds to 0.00%.
The one rule it does trip is consistency, which caps how much of a year's profit a single day may carry, and it is breached in 1 year. Concentration explains why that is even possible on numbers this small. 80% of the profit was made in 29 of the 2,362 trading days, which is 1.2% of them. Our concentration verdict is the only red lamp on this audit.
100% real ticks from 2003.05.05, 1,409 absent minutes out of 7,897,519 minute bars, 0 randomizer prints, so the run is deterministic and another tester can reproduce it. 0 failed entries, 0 chronology violations, 0 duplicate deal ids, log coverage 1.0, and the reconstructed final balance matches the report to the cent.
Two figures belong to the tester head rather than to our cost ladder, and we label them as such. The report head prints a profit factor of 1.43 and a Sharpe of 20.52, both before our commission retrofit. Our own end of day Sharpe on the same balance curve is 0.66 and the compound growth rate is 0.02% per year. A head Sharpe above 20 is arithmetic, not a quality statement. It divides a tiny mean by a tiny per trade variance.
The MQL5 page listed this EA at $499 when we checked on 2026-08-25, and the vendor recommends a $500 deposit. Our measurement frame runs a fixed 0.01 lot on $100,000, so the dollar path does not change with account size, and a reader on the recommended deposit can read the same $381.48 as the 23 year result. We flag the caveat our own audit flags, which is that margin requirements are not modelled in that view, so the direction of that reading is optimistic rather than conservative.
We also replay the observed trade path from every one of the 7,782 deal timestamps as a fresh start on the recommended $500 and count how many hit zero. The answer is 0.0%. Of the 18 audits in our catalog that carry the same survival map, 13 also read 0.0% and 5 read double digits, the worst of them 97.0%. Nothing here is going to take your account. Nothing here is going to pay you either.
The deal level census, the release split and the withdrawal replay need deal reconstruction, and that 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.