A vendor names a deposit and we measure on a different one. Our protocol runs every backtest on 100,000 USD at a fixed 0.01 lot, which is a measurement frame and not a trading account. The number a buyer actually funds is the vendor's own recommendation, and across the catalog it runs from 100 USD to 2,000 USD. So the engine takes the largest drawdown of the measured run in dollars and divides it by a ladder of account sizes. In 6 of the 22 default audits that carry a row for the vendor deposit, that one drawdown is larger than the deposit itself.
The ladder is fixed at 1,000, 2,500, 5,000, 10,000, 25,000 and 100,000 USD, with the tested balance and the vendor deposit added and duplicates dropped. Nothing is re run and no trade moves. With a lot size that never changes, the dollar result of every trade is the same whatever the account holds, so one dollar drawdown can be expressed against any account. Gold House shows the arithmetic in a single line. Its drawdown is 30.02 percent of the 2,000 USD the vendor recommends, 24.02 percent of 2,500, 12.01 percent of 5,000, 6.00 percent of 10,000, 2.40 percent of 25,000 and 0.60 percent of the 100,000 the run was measured on. The measured run goes through a 6 percent floor on every rung up to 10,000 USD and first clears it at 25,000.
Beside that column the audit prints a second one. For three prop rule profiles it asks whether the account would have gone through the floor, once for the measured path, which the table marks as a hit, and once as a percentage across 1,000 resampled paths. The first profile is our funded IQ Capital Classic copy with a 6 percent end of day floor, the second an FTMO challenge shape with a 10 percent static floor, the third a generic 6 percent trailing set. All floor figures below come from the first one.
The same measured drawdown, read against the frame and then against the deposit the vendor names. Scalping Robot Pro reads 53.23 percent on the frame and 5,323.40 percent on 1,000 USD. Logan reads 30.25 percent and 1,512.40 percent on 2,000 USD. Lizard reads 2.55 percent and 1,018.32 percent on 250 USD. Gold Snap reads 0.54 percent and 269.99 percent on 200 USD. Smart Gold Hunter reads 1.72 percent and 172.03 percent on 1,000 USD. Gold Trade Pro reads 1.39 percent and 138.55 percent on 1,000 USD.
A drawdown above 100 percent of the account is arithmetic and not an observed margin call. The trades happened on the frame, where the account was never close to being closed out, and the ladder only asks what the same dollars would have meant on a smaller balance. The audit states the limit of that itself, margin is not modeled, so a small account may not have been able to hold the positions at all, and the direction of that error is optimistic.
At the vendor deposit the measured run goes through the 6 percent floor in 18 of the 22 audits. The four that do not are Quantum Athena X at 0.03 percent of 1,000 USD, Quantum Queen X at 0.05 percent, Quantum Queen at the same 0.05 percent, and OilVector X at 3.36 percent. The two Queen entries are one measured run counted twice, our deal level census found 6,714 identical deals across both tester runs, and the older listing stays in the catalog as the republish record.
Four further default audits carry the ladder but no vendor row, because their runs come from vendor reports on another broker and carry no deposit of ours to mark. They are left out of every count above rather than filled in.
In three of the 26 default audits at least one row shows a hit next to a resampled probability of 0.0 percent. Gold House carries five such cells, Gold Snap six and Lizard one. Gold House at 5,000 USD is the plainest case, the measured path went through the floor and not one of the 1,000 resampled paths did.
That is a property of the convention rather than a defect. The floor check sweeps every gain back to the starting balance on the same day, so the account never builds a cushion and a deep excursion needs several losing weeks in a row. The resampling draws five day blocks in a circle, which keeps a week together and scatters the weeks. A long slow decline therefore survives in the measured order and dissolves in nearly every reshuffle. The audit already calls its ruin figure a lower bound because the full firm rule sets are stricter than a bare floor, and this is a second reason for the same warning.
Lizard makes the point at the top of the ladder. At 25,000 USD, which is a hundred times the deposit the vendor recommends, the measured run still goes through the 6 percent floor while the resampled paths do it in 0.1 percent of cases.
Every audit in the catalog carries this table. The floor replay behind it is taken apart in the prop firm sweep and the resampling method in the reshuffle piece. Your own tester report? The browser check is free.