Every audit in this catalog records the hour of the day at which trades were opened, and the answers are far less varied than the product pages suggest. Take the largest setup group of each of the 26 default audits and read off its busiest entry hour. For 15 of them that hour is 15, 16 or 17. Eleven sit on hour 15 alone. The remaining eleven audits are scattered across hours 1, 2, 7, 8, 9, 11 and 22.
Our engine splits a run into setup groups by the comment the robot writes on each entry, then counts the opening hour of every trade in a group and keeps the three most frequent. Timestamps are read exactly as the tester printed them, with no conversion, so the hour is broker server time and nothing else. The audit does not store the offset from any other clock, which is a real limit and worth stating before any number is read out of it.
One anchor we do hold. Our own note on broker financing records the daily rollover at hour 0 server time, which the broker itself describes as five in the afternoon in New York. Taking that description at face value, server hour 15 is eight in the morning in New York and server hour 17 is ten in the morning. Server hour 1 is the hour that begins one hour after the daily rollover, and that one needs no timezone at all, because the rollover defines hour 0 by construction.
The eleven audits whose largest group is busiest at hour 15 are Adaptive Gold Scalper MT5, Gold Atlas, Gold House MT5, Gold Snap, Gold Trade Pro MT5, Quantum Titan MT5, Smart Gold Hunter, The Gold Reaper MT5, ThunderGold Scalper, TwisterPro Scalper and Ultimate Breakout System. Scalping Robot Pro MT5 peaks at hour 16 across all 302,798 of its trades, and Logan MT5 and Smart Gold Impulse peak at hour 17. Waka Waka, which is not a gold robot, also peaks at hour 17.
This is partly a fact about our catalog rather than a discovery about the market. Of the 26 default audits, 21 record XAUUSD as their main symbol. Robots built for one instrument on one broker feed will tend to wake at the same time, and fourteen of those fifteen crowded audits are gold runs. The honest version of the headline is therefore narrower. 14 of our 21 gold audits open most of their trades in a three hour band, and 7 do not.
Those seven are the interesting ones, because they trade the same instrument on the same history and still choose a different clock.
By the broker's own clock description, hour 22 is three in the afternoon in New York and hour 1 is six in the evening. So two of these runs open most of their trades near the end of the New York afternoon, one in the hour right after the daily rollover and one in the hour after that. The last two sit at hours 7 and 9, which is midnight and two in the morning in New York. That is six distinct runs behind the seven audit files, against fourteen that open in the New York morning. All of these runs read the same XAUUSD tick series, so a difference in entry hour is a property of the robot and not of the feed.
The sharpest case in the catalog is Prop Firm Gold EA. Its run holds 9,538 trades in three setup groups, and for the largest of them the audit lists exactly one entry hour. The field keeps up to three, so one entry means the group has only one. All 4,759 trades of that group opened inside server hour 1, and all 4,759 were long.
The three groups look like three different robots wearing one name.
Per trade that is 0.30 USD for the biggest group, 0.92 USD for the second and 1.81 USD for the smallest. The setup that fires most often earns least per trade, by a factor of six against the one that fires least. The three group figures add to 6,769.79 USD, which is exactly the net profit the cost block reports for the whole run, and the three trade counts add to the 9,538 that the exit mix also counts. The split is a partition of the run and not a sample of it.
The clock and the holding time also agree with the cost block. A group that opens at hour 1 and holds for a median of 8.51 hours is closed long before the next rollover, and the audit records 0 overnight trades and a swap line of exactly 0 for this robot. Nothing here is inferred. The hour, the holding time and the financing line were measured separately and point the same way.
Three limits belong next to every number above. The first is coverage. The field keeps three hours out of a possible twenty four, so it says which hours are busiest and says nothing at all about how the rest of the day is filled. A robot with a sharp peak and a robot with a gentle one produce the same three numbers.
The second is the choice of group. For 9 of the 26 default audits the run has a single setup group, so the busiest hour describes the whole run. For the others it describes only the largest slice. Prop Firm Gold EA splits into three groups and its largest holds 49.9% of the trades. Logan MT5 splits into 76 groups and its largest holds 355 of 1,372 trades, which is 25.9%. For Logan the busiest hour of the largest group is a weaker summary than the same figure is for Gold Snap, and that difference is not visible in the number itself.
The third is meaning. We can tell you when a robot opens trades. We cannot tell you from a backtest whether that hour is a good one to trade, because the answer would depend on spread, slippage and fill behaviour at that hour on a live account, and a tester report carries none of those. What the hour is good for is comparison. When two robots sold as different products open their trades in the same hour on the same instrument with the same median holding time, that is a question worth asking the vendors.