The stop loss that made money: 38,566 exits labelled SL that closed in profit

2026-09-04. Every figure quoted here is a field in a published audit JSON, and the listed audits are linked in place.

Smart Gold Impulse closes 4,146 of its 4,163 trades at a stop loss. Of those stopped trades, 3,672 ended in profit after commission and swap. That is not a contradiction and it is not a reporting error. It is what a trailing stop looks like in a trade list, and every audit we publish counts it, because a reader who sees a column full of the letters SL will otherwise read a column full of losses.

What the four exit classes are

The engine sorts every closed trade into one of four classes by reading the comment the terminal wrote on the closing deal. There is no close reason field in a tester report, so the label is the only thing there is to read.

SL. The closing comment begins with sl or carries an sl marker in brackets. It says the stop level was touched. It says nothing about the sign of the result.
TP. The same test for a take profit marker.
End of test. The position was still open when the run reached its last bar and the tester closed it. Across all 26 default audits this happens to six trades in total, three in Quantum Emperor, two in Waka Waka and one in Prop Firm Gold EA.
Signal or time. Everything else. The EA closed the position itself, or the comment carried no marker we recognise.

The four counts add up to the trade count of the run, which is the cheapest check on the block. Gold Snap reports 3,294 stops, 461 take profits and 14 in the last class, and its paired trade list holds 3,769 trades. Smart Gold Impulse reports 4,146 and 17 against 4,163 trades.

How often a stop loss pays

Twenty one of the 26 default audits close at least one trade at a stop. In twelve of those twenty one, more than half of the stopped trades ended in profit. The share is computed on the same net figure the rest of the audit uses, which is the raw result plus swap plus both commission legs, so a stop that closes at plus two cents gross and minus four cents net counts as a loss here.

The strongest case in the catalog is Quantum Emperor with 368 of 378 stopped trades in profit. Then come TwisterPro Scalper with 876 of 986, Smart Gold Impulse with 3,672 of 4,146 and Quantum Titan with 1,275 of 1,582. Add every audit together and 38,566 trades in the catalog were closed by a stop loss and still made money.

The other end of the range is just as clear and it is the reason the figure is worth printing. Scalping Robot Pro hits a stop 124,466 times and exactly one of those closes in profit. OilVector X manages two out of 846. Four audits sit at a flat zero, Market Anomalies EA with 0 of 80, Prop Firm Gold EA with 0 of 776, Pulse Engine with 0 of 92 and Range Breakout EA with 0 of 193. In those runs the stop is a stop in the ordinary sense, and the same column of SL labels means the opposite of what it means in the twelve above.

Why the trailing flag is false on six audits that trail

The audit also carries a one word verdict on this, a trailing stop signature, and it is true in eight of the 26 runs. Those eight are Adaptive Gold Scalper, CryonX, Gold Atlas, Gold House, Smart Gold Hunter, Smart Gold Impulse, ThunderGold Scalper and TwisterPro Scalper.

The flag needs two conditions at once. More than nine in ten of all trades must end at a stop, and more than one in five of those stops must be profitable. The second condition is the easy one. The first one is what makes the flag conservative, and six audits fail it while their stops are plainly being trailed.

Gold Snap. 3,294 stops out of 3,769 trades, and 1,884 of the stops in profit. It misses the gate because 461 trades ended at a take profit instead.
Quantum Emperor. 378 stops out of 559 trades, with the highest profitable share in the catalog. The gate does not care.
Quantum Titan. 1,582 out of 1,947, with 1,275 profitable stops.
The Gold Reaper. 926 out of 1,742, with 650 profitable stops.
Gold Trade Pro. 4,254 out of 6,305, with 2,157 profitable stops.
Lizard. 8,206 out of 21,524, with 5,623 profitable stops. Most of its trades end in the signal or time class, 13,074 of them.

We would rather leave the flag off and print the two counts than raise a flag on a threshold that was chosen by us. The counts are the measurement. The flag is a summary of the counts, and a reader who disagrees with our gate can see every number it was built from.

The audits with no stop at all

Five of the 26 close no trade at a stop, and their profitable share is empty rather than zero. Empty means there were no stopped trades to divide by, and zero means there were and none of them paid. That is a real difference and a summary table that prints both as 0 destroys it. The five are Logan, Quantum Athena X, Quantum Queen X, Quantum Queen and Waka Waka.

Waka Waka is the informative one. It closes 528 trades at a take profit and two at the end of the test, so its comment field demonstrably carries markers and the absence of stops is a fact about the run. Logan, Quantum Athena X and the two Queen entries close every single trade in the signal or time class, 1,372 and 3,976 and 3,357 of them. For those four we cannot tell from the exit mix whether the EA ran without a stop or whether it overwrote the comment that would have said so.

What the exit mix cannot say

The class is read from text, not from a field. An EA that writes its own comment on the closing deal lands in the signal or time class no matter what actually closed the position. This is the single largest limit of the block and it is why we publish the four raw counts rather than a sentence about exit style.

The stop level and the stop label come from different places. The audit reads the sl and tp levels of each position out of the tester log through the order number, while the class above comes from the comment. A run can therefore carry stop levels on every trade and still show no SL exits, which is exactly the ambiguity in the paragraph above.

Profitable does not mean well managed. A trailing stop that banks 3,672 small winners can still lose money overall if the unstopped remainder runs far enough. Smart Gold Hunter is the case in point, with 6,443 profitable stops out of 9,886 and an audited net result of minus 1,418.55 USD.

Quantum Queen and Quantum Queen X are the same measured run under two catalog entries, so the count of 26 audits contains 25 distinct runs. Both carry an empty share here, so nothing in this article is doubled by it.

Run these checks yourself

Count the SL rows, then count the negative ones. If most of the stopped trades are green, you are looking at a trailing stop and not at a losing streak.
Ask what the empty cell means. No stops and no profitable stops are two different reports, and only one of them is a statement about the strategy.
Check the take profit column too. A run with stops and no take profits at all is managing exits with one tool, and that tool is the one that fires in a fast market.
Read the net column, not the gross one. A stop that closes a hair above the entry is a loss once the round turn is paid, and the share above is already computed that way.

Every figure above is public in the audit files linked in place, and the category pages list the verdict lamps for free. Your own tester report? The browser check is free.

The honest limits. The exit class is a text heuristic over the closing comment and nothing more, so both the SL count and the signal or time count carry whatever the terminal and the EA chose to write. The trailing stop signature uses two thresholds we set ourselves, nine in ten and one in five, and the six audits listed above show what those thresholds exclude. All figures are single runs measured on our own tester world with the vendor default inputs, and the four unlisted audits named without links are older runs we do not rebuild. The runs behind this article end in August 2026, and a count of profitable stops is a description of how a robot exits rather than a judgement of whether it should be bought.