Model a prop-firm evaluation
An evaluation turns on a target and a floor, and on the order in which your equity meets them. A ratio computed over the finished period answers a different question.
First passage, and why an average misses it
Two thresholds settle an evaluation: a target that has to be reached and a floor that has to be avoided. The outcome turns on the order in which the equity curve meets them, so a strategy with a strong finish and an early drawdown fails while a flatter one passes.
A Sharpe ratio, a total return or a maximum drawdown read over the finished period all average across that ordering. They describe the same series and answer a different question from the one an evaluation asks.
Open full sizeRules that report, against your own numbers
The rule family ships 17 nodes, of which 13 are validated by executable checks, and 4 carry a declared roadmap reason. Drawdown floors, a daily loss limit, a profit target, phase progression, minimum trading days, a consistency rule, an inactivity rule and a day-by-day account simulation each read the run you already produced.
The input is the run itself: the bar-by-bar timestamped P&L of the best configuration. Equity floor, daily loss, target, active days, the longest inactivity gap and the share carried by the best single day are read off that series, with no external data and no second simulation.
A rule node reports; it blocks no run and excludes no configuration. The numbers it uses come from the node you placed on the canvas, and a threshold you did not declare is treated as an absent rule, with the missing input named in the result.
Four rule nodes carry a declared roadmap reason and report it in place of a figure. They cover a correlation limit across open positions, an exposure cap in lots, a news blackout by event type, and a cap on concurrent positions. Each names the input it lacks, so the gap is readable where it applies.
Your rulebook is entered by you, and checked by you
Presets carrying a firm's name are absent from the shipped catalogue. Two consequences follow, and both are yours to hold. The published wording of your firm is where the numbers come from, and the transcription into the nodes is a step you perform and verify.
What the shipped catalogue offers in exchange is that no threshold is assumed on your behalf. A rule you did not enter reports as absent, so a gap in your transcription reads as a gap and not as a pass.
Open full sizeFee, phases, funded account, payouts
An estimate of the full cycle is available beside the rules. Entry fee, two evaluation phases, the funded account, payout cycles with a profit split, an optional fee refund on the first payout and optional retries are simulated by bootstrapping the daily P&L of your run.
The simulation uses a fixed seed, so the same run and the same settings produce the same estimate. Below the minimum number of trading days declared in the node, the result is reported as insufficient. The estimate is informational within that run, and it changes no verdict.
Open full sizeWhat is being claimed, and what is left out
The rule nodes report against the series a run produced, using the thresholds entered on the canvas.
- Scope
- The reading covers the bars that were loaded and the best configuration of that run, in the account currency implied by your risk declaration.
- Effect
- Simplifications are declared in the result: static drawdown anchors, an i.i.d. bootstrap without autocorrelation, and a risk model driven by the declared risk per day.
- Control
- Each declared simplification is named in the note attached to the estimate, next to the figure it affects.
Production packaging requires the strategy engine and fails if its source is missing. A prop-firm demonstration may use a generic cross-sectional fallback when marked as a non-deliverable demo. Such a fallback is not a prop-firm evaluation result.
No result described above is a statement about future performance, and the harness carries no signal to trade. A result concerns the series that was submitted.