Lizard placed 21,524 trades in our run and not one of them was still open on the next calendar day. Gold Atlas placed 10,999 and carried 2,188 of them into a following day, of which 640 are counted as spanning a weekend. Both figures sit on the audit page beneath every prop firm simulation, and one of the two means less than it looks like. What follows is how each is built, what can be read out of the pair, and where the weekend one stops being a measurement.
Both are counted off the paired trade list, so every entry has a known exit and both timestamps are available. The overnight count is the simple one. A trade counts if the calendar date of its exit is later than the calendar date of its entry. Nothing there is open to interpretation, and the same number also appears in the cost table of the audit, where it is the reason a run pays financing at all.
The weekend count is built from two tests joined by an or. A trade counts if entry and exit lie two or more calendar days apart, or if the exit falls on an earlier weekday than the entry while still carrying a later date. The second test is the week boundary itself, because the only way to end on an earlier weekday is to have crossed into a new week. The first test is not a weekend test at all. It is a proxy, and it is the one doing most of the counting.
A position opened on a Monday and closed on the Wednesday lies two calendar days apart. It passes the first test with no weekend anywhere inside it, and it is counted. A position opened on a Friday and closed on the Monday lies three calendar days apart and is counted by the same test, which is the case the number is meant for.
So the weekend count can only overstate weekend exposure, never understate it. Every real weekend crossing is at least two calendar days wide and is caught, and some mid week holds are caught along with it. The audit does not publish the split, which means we cannot tell you from the file how much of a given count is a Monday to Wednesday hold and how much is a genuine Friday to Monday one. Read a positive value as a ceiling. A 0 stays exact, because the ceiling only loosens a count once there is something in it.
By construction the weekend count is a subset of the overnight count, since both tests require the exit to carry a later date. Across all 33 audit files, 26 default runs and 7 variant runs, it is smaller than or equal to the overnight count in every single one. That is the check we would expect to hold, and it holds.
The gap between the two is therefore a count in its own right. Quantum Queen X carries 518 overnight trades and 29 weekend ones over 3,357 trades, so 489 of its overnight holds closed on the very next calendar day inside the same week.
Where the weekend count is 0 and the overnight count is not, every overnight hold closed on the next calendar day and none reached into a new week. Smart Gold Impulse is that case at scale, with 51 overnight holds out of 4,163 trades and 0 counted weekends. Thundergold Scalper held exactly 1 position into a following day across 984 trades. The 2.2 build in the Quantum Titan version matrix held 3 and spanned none.
Where the two are equal, the opposite holds and no overnight position closed on the next calendar day. Market Anomalies EA records 4 and 4, Pulse Engine records 13 and 13, and both are unlisted here so they stand without a link.
Sorted by the raw weekend count the top of the list is Gold Atlas with 640, then Gold Trade Pro with 423, Ultimate Breakout System with 178, Quantum Athena X with 155, Cryonx with 126, OilVector X with 111 and The Gold Reaper with 100.
Measured against the size of the trade list the order changes. Ultimate Breakout System counts 178 of 945 trades, which is 18.8 percent and the heaviest share in the catalogue. Gold Atlas counts 640 of 10,999, which is 5.8 percent, and The Gold Reaper counts 100 of 1,742, which is 5.7 percent. At the other end Lizard and Prop Firm Gold EA record 0 of 21,524 and 0 of 9,538. Two robots in this catalogue never once let a position see the next day.
Our engine holds a red flag for a run that carries positions across a weekend where the profile forbids it, and that flag has never been raised. We ship 6 prop firm profiles, and every one of the 28 catalogue entries runs the same 3 of them. All three allow overnight and weekend holding. The two profiles that forbid weekend positions are the funded standard account of one firm and the futures branch of another, and neither is in the list any audit runs. The flag is defined in our source and appears in none of the 33 files.
That belongs in the open rather than in a footnote. The weekend number is reported and it is not judged. If you trade a funded account whose rules end at the Friday close, this is the line to read yourself, because no lamp on the page reads it for you. The rules we do simulate carry their own dates, 2026-08-13 for the two firm profiles and 2026-08-12 for the generic one, and prop firm rules move.
The practical version is a single question to put to a seller. Not whether the robot holds overnight, which an equity curve will never tell you, but how many of its trades were still open when the market closed for the week. If the answer is not in the report it is in the trade list, and the trade list is where we counted ours.