One drawdown in this backtest began in June 2013 and ended in April 2024. It cost 30.41 USD. That single line explains ThunderGold Scalper better than any ratio does. The account spent 2,818 trading days below a previous high water mark while losing the price of a decent lunch, because the whole system moves so little. Over the full run the account gains 314.65 USD from 984 trades in twenty years, which is a real result and a very small one.
One deterministic tester run of version 1.3 on XAUUSD# M15, 100% real ticks, first trade 2006-05-04 and last trade 2026-08-12. We reconstruct the run deal by deal against the tester journal, so pairing confidence is exact, journal coverage is 1.0, chronology violations 0, duplicate deal ids 0 and the reconstructed balance matches the reported one to the cent. The catalog protocol uses a fixed 0.01 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 78.72 USD across 1,968 deals. That is 20.1% of gross profit, and our cost lamp is green here.
Read the vendor's tester header on its own and this is close to a perfect report. Net profit 393.37 USD before our retrofit, profit factor 1.73, recovery factor 14.8, maximum balance drawdown 26.13 USD or 0.03%, and a Sharpe ratio of 73.17. That header figure is 66 times the Sharpe we measure on the same run, and what it counts is 984 individual trade results, each of them tiny, on an account of 100,000 USD.
We compute the same statistic on the end of day balance curve, which is how any fund or prop desk would look at it. The answer is 1.11. Compound growth on the measurement frame is 0.02% per year. Neither number is wrong. They are the same run at different resolutions, and that factor of 66 is one of the widest we have measured. Whenever a sales page quotes a Sharpe ratio without saying what a period is, this is the gap it is hiding in.
Our end of day analysis splits the curve into 47 drawdown episodes. The deepest one is worth 31.79 USD. It peaked on 2024-09-09, bottomed on 2025-06-12 and recovered on 2026-01-06, which is 346 trading days underwater and 0.03% of the measurement frame. The second deepest is the one worth reading twice. Peak 2013-06-25, trough 2022-03-07, recovery 2024-04-12, depth 30.41 USD, 2,818 trading days underwater across 201 trades of which 72 lost.
A drawdown that lasts almost eleven years is normally a catastrophe. Here it is a rounding error, and that is the point. Percentage risk metrics are meaningless when the position size is fixed at 0.01 lots against 100,000 USD of capital. What survives the change of frame is the shape, and the shape says the equity curve went nowhere for a decade.
The record is thin in every direction. The account has a closed trade on 811 days. Between the first and the last trade there are 5,289 weekdays in our end of day series, so the robot is flat on about six days out of seven. That averages under fifty trades a year. 2018 carries no trade at all. 2007 and 2017 carry two each.
Seven of the twenty years in the record are negative, which our audit rates caution. None of them is dramatic. 2015 loses 7.88 USD, 2019 loses 5.11 USD, 2021 loses 2.76 USD. A system that produces losses this small also produces gains this small, and twenty years of it adds up to 314.65 USD on the measurement frame.
Now sort the result by year. 2026 delivered 179.51 USD from 186 trades at an 80.1% win rate, by 12 August. That is 57.0% of everything the system made across twenty years. The nineteen and a half years before it produced the remaining 135.14 USD. The immediately preceding year, 2025, was the busiest in the record with 198 trades and finished at -0.95 USD.
Concentration confirms it from the day side. 80% of the profit was made on 42 days, which is 5.2% of the 811 active days. Remove the twenty single best trades out of 984 and the net falls from 314.65 USD to 175.40 USD. Two percent of the trades carry 44% of the result. The best day in twenty years earned 12.95 USD on 2012-12-04 and the worst lost 5.35 USD on 2020-03-24, so nothing here is driven by an outlier disaster. It is driven by a small number of good days in a long stretch of nothing.
The exit census is unusually clean. All 984 trades ended at a stop. Not one closed at a take profit, by signal or by time. Of those stop exits, 706 closed in profit, which is the 71.7% win rate exactly. This is a trailing stop taking every decision about when to leave, and the win rate is a property of that mechanism.
The structure is otherwise as conservative as our checks can confirm. No martingale, no averaging into losses, never more than one position open, one symbol, 0 hedged moments, average holding time 134 seconds. Our structure lamp is green, which 4 of the 25 audits in our catalog manage. Worth noting alongside that, the input panel of version 1.3 contains no strategy parameters at all. Nineteen inputs cover lot size, a news filter that is switched off in this run, a holiday filter that is on, a broker type switch and six settings for the on chart panel. There is nothing to tune and nothing to overfit, which is a design choice with two sides to it.
The vendor recommends a deposit of 100 USD, so our capital replay uses that figure rather than a fallback. On that basis none of the 1,968 possible start points ends in a dead account, and the single observed path survives all 7,405 days of it. That is a genuine point in the product's favour and we say so plainly.
The same replay shows the other side of a 100 USD account. Re-expressed against that capital, the observed path carries a maximum drawdown of 31.79%, and all three prop firm floors we model are breached on the observed path itself rather than merely in simulation. Our prop fit lamp is red, and the reason is the consistency rule. Best day share exceeds the allowed percentage in 11 years under the funded profile we model. A system that earns on 42 days out of 811 conflicts with consistency rules by construction.
Over the traded span the entry price series of this run rises from 649.03 to 4,347.78, a move of 569.88%. Holding 0.01 lots through those twenty years would have earned 3,698.74 USD. The EA earned 314.65 USD on the same lot size, which is under a tenth. That comparison is approximated from entry prices and holding gold means sitting through every gold crash since 2006, so it is not a fair risk comparison. It is a scale comparison, and scale is the question a buyer should ask first.
ThunderGold Scalper was published on 2026-07-31 and updated 3 times in its first eight days, on 2026-08-01, 2026-08-04 and 2026-08-07. Our test window ends on 2026-08-12. The out of sample stretch after the last update is therefore 5 days and 3 trades, worth 0.19 USD. That is not an out of sample test, it is a rounding position, and we report it as such. The 974 trades before release carry a profit factor of 1.54 and effectively are the entire record.
One date counts in the product's favour. We measured version 1.3 and the market listing still shows version 1.3, so this audit describes the software that is currently on sale. That is not always the case, which is why every audit we publish names its version and its date.
The deal level census, the drawdown decomposition and the capital 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.