ZIdentifier, ZCopier, and ZStrategyNQ are each built on the same standard: extensive research into NQ and ES market behavior, careful calibration using tick-by-tick market replay data, and testing against data the strategy was never tuned on.
Exact reproduction of a backtest depends on your platform, broker feed, and data download matching ours — for an outside user, that's not a standard we can honestly promise and deliver. So instead of asking you to trust a number you can't reproduce, we show you the strategy in action, running, uncut, and publish everything behind it.
Most tools in this space claim to pass evals. Ours doesn't, by design — and saying so plainly is the first thing we want you to know about how we operate.
We're not in the business of predicting markets. We're in the business of holding ourselves to one standard, set before we see the result, and reporting it the same way whether it flatters us or not.
A second out-of-sample window (Oct–Nov 2026) follows the same standard — see the forward-commitment scoreboard below.
We test on four trading windows a year, chosen to sit off contract-roll periods and span different market conditions. We label which periods the strategy was tuned on and which it wasn't — most vendors don't.
| Window | Status | What it tells you |
|---|---|---|
| Loading current validation status… | ||
Why show the calibration windows at all? Because the honest signal isn't any single window's result — it's the comparison. If the out-of-sample windows hold up near the calibration windows, that resemblance is meaningful evidence. If they degrade, we publish that too. Anyone who shows you backtests without telling you which periods the system was built on is showing you a photograph and calling it a pulse.
A forward commitment, not a highlight reel. Misses stay on the board.
Loading…
ZIdentifier is a Bayesian setting scorer, not a decider. It hands you a posterior probability distribution over the strategy's settings — when the evidence can't separate two of them, it says so instead of forcing a confident label.
This is an illustrative exhibit of an AMBIGUOUS reading, not a live feed. A confident call looks the same, with one setting's posterior clearly ahead. Both states are shown on the Live Demo page in real time.
Two things you'll see referenced throughout this site — what they actually mean.
ZStrategyNQ ships with its calibration protected: the core tuned parameters are held at their validated values automatically (status: LOCKED). You can unlock it to research your own configurations — the choice is yours, and it's transparent — but once unlocked, the tool runs your values, and the validated calibration no longer applies.
Every proof video on this site is produced locked. What you watch is the shipped, protected calibration — not a special demo tune.
An on-screen readout that watches a run and reports whether the account would have breached a prop-firm evaluation limit — a measurement tool, not a trading behavior. It changes no trades.
We include it because it tends to report the uncomfortable truth quickly: at full-size NQ scale, this strategy typically breaches a small eval account almost immediately. We'd rather show you that number than hide it — it's the clearest illustration of the disclaimer at the top of this page.
Mechanism-level write-ups from internal development — what didn't work, and why. Calibration values are never disclosed here; the point is showing our reasoning, not our tuned numbers.
A mechanism-level look at why ThreeOut-style fade setups become structurally unprofitable once a market enters sustained directional trend — and how the signal set is gated off in that environment.
How ZIdentifier ensures every displayed score is computed on clean current-contract data, and what silent correlation contamination looks like when that gate is missing.
Data gaps get disclosed here directly, never worked around quietly. As of this build, no gaps have been identified in the published windows.
No missing sessions currently on record for Jan–Feb, Apr–May, or the in-progress Jul–Aug window. This section will be updated if any are found during validation processing.
Note for researchers re-running these windows themselves: if you load the same settings file and replay the same window on your own machine, expect results that are close but not bit-identical to ours, and not bit-identical between your own repeated passes either. NT8 tick Market Replay is sensitive to run-to-run initialization — an "initial-value" effect — so any single replay is best read as one draw, not a fixed constant. This does not appear within a single continuous run (which is what the proof videos are); it shows up only when comparing separate passes. Judge the strategy by its behavior and shape across a window, not the last decimal of a single figure.