How the account was tested
Each configuration is replayed over the full dataset to measure its strategy metrics. Separately, a fresh evaluation starts on the first complete session of every calendar month. After a pass, funded trading starts at a new $25,000 balance on the following session. Passing attempts continue until first payout eligibility, account breach, or the end of the data. Unresolved attempts include accounts that stopped finding structural trades small enough for their remaining allowance. Monthly attempts overlap and their counts do not establish independent probabilities.
Sizes tested are 1 NQ, 1 / 5 / 7 MNQ, and 1 ES. The smaller 1-MNQ configuration was added to examine how structural stops interact with a $1,000 account loss allowance. Every size fits the initial funded scaling tier. The $600 figure is a per-trade net target in that target model. A 1R target is calculated from entry to structural stop, with fees and slippage deducted from the result.
What “passed” means here
The count gate requires more than 100 full-history trades. A survivor requires at least 100 training trades plus positive training and validation P&L. Ranking uses validation P&L only. The final 20% is reported for inspection. These market dates have been used in earlier studies, so the final segment is exploratory. The family-wise test adjusts the validation evidence for searching every configuration.
Research coverage
The two ICT confluence families combine a prior-session liquidity reference, sweep, displacement, structure break, FVG retracement, premium/discount and optional OTE. They are specific mechanical approximations. Adding every discretionary ICT concept does not define one unambiguous algorithm, and the available single-instrument data does not support a synchronized cross-market SMT test. Historical scheduled news, order-book liquidity and actual fill queues are absent.