Research tools for people who trust methodology over marketing hype. Curated and created by a retired academic data scientist.
The Z-Suite is a modular constellation of NT8 research tools — a Bayesian regime scorer, a multi-regime research strategy, a trade copier, and an account observer. Four independent products. Any one stands alone; any combination multiplies value. There is no required progression — you assemble what fits your research workflow. No performance promises. Every result we publish comes with its receipts.
NT8 chart indicator
A Bayesian setting scorer, not a classifier. Reads a 15-session lookback on an affiliated pair and emits a posterior probability over the strategy's settings — a probabilistic starting bias, not a verdict. When the evidence can't distinguish between two settings, it says so.
Learn more →Trade copier, v1.1.6
Copies orders from a master NT8 account to up to 6 follower accounts, across different instruments and brokers, and across live, sim or eval accounts. Catastrophe stops, reconnect watchdog, and stale-fire guards are built in.
Learn more →Live account panel + daily log
A pure observer: shows live P&L, trade count, win rate, and drawdown on the chart, and can write a daily record to CSV. Places no orders, runs no strategy, reads no strategy state.
Learn more →Shipping instance: ZStrategyNQ
An instrument-agnostic trading engine for CME products, delivered as per-instrument calibrated instances. Today's shipping instance, ZStrategyNQ, trades NQ paired with ES. Ships pre-calibrated and locked by default. Currently in out-of-sample validation — see the Research page for the live status.
Learn more →Most vendors in this space assert performance. We publish the artifacts that let you check it yourself — before you take our word for anything.
Our out-of-sample validation criteria are set before the data exists — published in advance, not fitted after the fact.
Results are shown across multiple repeated test runs, not a single cherry-picked pass.
ZIdentifier says "AMBIGUOUS" when the evidence can't distinguish between two possible settings, instead of forcing a confident call.
Out-of-sample results are published whether they pass or fail. A published miss is still evidence.
A preview of what's on the Research page, fully labeled. Full detail, methodology, and the live validation status live there.
| Window | Status | Result (simulation only) |
|---|---|---|
| Loading current validation status… | ||
Calibration-window results are periods the strategy was tuned on — strong results there are expected and are evidence of fit, not predictive skill. The out-of-sample windows are the real test. All figures are historical simulation results only and do not predict future performance. Full methodology and proof artifacts →
Get notified when the Jul–Aug out-of-sample result publishes, and when new products or research notes go live.
No spam, no advisory content. Unsubscribe anytime. [Placeholder form — backend TBD]