AR-OUT-001·PL-01 — The output wall. Every figure answers one question. Stylised brand illustration of the laboratory, not a photograph of a physical facility. The charts and figures inside the plate are simulated.

Record AR-OUT-001 · Output gallery

What the bench
actually tells you.

The parameter laboratory is where a configuration is built. This is where it is read. Nine rails at the top, and then the whole of what they produce: the score and the safeguards that cap it, the headline figures, a conclusion the laboratory writes for itself, eleven ensemble figures drawn live, and the catalogue of every further visual output the application is specified to produce.

ENGINE DETERMINISTIC PREVIEW FIGURES 6 LIVE CATALOGUE SPECIFIED

Interactive demonstration — not a historical backtest

AR-OUT-001 · Control strip

Nine rails

AR-OUT-001 · Verdict

The score, and what caps it.

Hard caps are applied before any component is weighted. A configuration with a survival problem cannot buy its way past the cap with modelled return, which is the only way a single summary number is worth printing at all.

AR-OUT-001 · Figures 1 to 11

Eleven figures, drawn live
across an ensemble.

Every figure below is computed from 120 seeded histories of the configuration currently set on the strip — not from one representative history. That is the difference between this page and the parameter laboratory: the laboratory asks what moving a rail does, and answers with a median and a sensitivity. This page asks what the configuration produces, and the only honest answer to that is a distribution. Change a rail and all eleven redraw. None of them is a picture of a real account.

Why no equity curve, no underwater curve and no R histogram here

Those three are drawn in the parameter laboratory, where a single median history is the right reference for reading what a control change did. Repeating them here would show the same picture twice and tell you nothing new. What this page adds is everything a median history structurally cannot show: the spread around it, the histories that failed, and how the return relates to the decline it cost.

Figure 1 · Equity percentile fan Simulated

—

Figure 2 · Decline envelope across the ensemble Simulated

—

Figure 3 · Where the histories ended Simulated

—

Figure 4 · Deepest decline per history Simulated

—

Figure 5 · Outcome against decline endured Simulated

—

Figure 6 · Where the expectancy goes Simulated

—

Figure 7 · Risk sweep — survival and score Simulated

—

Figure 8 · Robustness surface — score across risk and concurrency Simulated

—

Figure 9 · Longest unbroken stretch below the peak Simulated

—

Figure 10 · Worst losing streak per history Simulated

—

Figure 11 · Same trades, different order Simulated

—

AR-OUT-001 · Specified output catalogue

Everything else the
application is specified to draw.

—

No placeholder charts

A figure that is specified but not built renders no picture here. Not a plausible chart with a badge on it, not a greyed-out mock — the name, and the word "specified". A catalogue that renders every entry identically is a catalogue that is lying about most of them.

Open the parameter laboratory How the full score is specified