Understand the Control Center
The Control Center offers a view into dataset provenance, candidate training, validation, specialist research, monitoring and deployment policy. Screenshots document a specific development or replay environment. Some features and readiness states vary by license, data availability and backend configuration. This is a guide to what each interface reports, not a claim that every shown state is production ready.
Datasets & lineage
Review the immutable dataset registry, historical ledger lineage, row counts and trainability status.
How this screen is used
The dataset view distinguishes historical observations from a dataset that is eligible for model training. Check lineage warnings before attempting a new build.
A SOURCE_MISSING notice means one or more referenced ledger files were not available in the captured environment. It is not a successful dataset build.
Candidate training
Prepare a candidate model from an eligible immutable dataset.
How this screen is used
The workflow shown is Dataset → Model family → Calibration → Review → Submit candidate. Available families and calibration methods depend on the installed version and server configuration.
A candidate is an experiment, not an approved production model. The screenshot also displays a monitoring timeout.
Model validation
Inspect backend-reported validation metrics and sample sufficiency.
How this screen is used
Metrics may include AUC, F1, Brier score, log loss, calibration and sample count. Validation decisions are reported by the server; the interface does not calculate them locally.
An INSUFFICIENT_SAMPLE or FAILED verdict means that run should not be described as production ready.
Model registry
Understand model identity, version, lifecycle and readiness.
How this screen is used
The registry describes a controlled lifecycle: Candidate → Validated → Shadow → Champion → Production → Retired. Progression is subject to backend checks and deployment policy.
An empty registry view does not demonstrate an active model.
Research overview
Review historical trade results and research readiness.
How this screen is used
The overview brings together completed trades, win/loss counts, realized PnL and Flow AI status. It is a research dashboard: figures are historical or replay-context observations, not a forecast.
RESEARCH_READY is a research-row threshold and does not independently authorize production model promotion.
Context / L2 state
Understand how contextual and Level 2 components report operating modes.
How this screen is used
This view distinguishes supported modes, current modes, readiness and previously observed ledger state for Context, L2 and Flow AI. The interface labels the reported state as read-only.
Unavailable and NOT_VERIFIED states mean that the active capability was not confirmed in the captured view.
BM specialists
Explore research readiness for seven distinct liquidity concepts.
How this screen is used
The specialist table covers Liquidity Wall, Flash Magnet, Absorption, Exhaustion, Iceberg, Spoofing and Whale. It compares samples, eligibility, observed wins/losses, quality and feature availability.
INSUFFICIENT means that a specialist does not yet meet the displayed readiness criteria. This table is not a live signal feed.
Deployment controls
Learn how a validated model is considered for production.
How this screen is used
Production control exposes deployment readiness, active deployment, champion eligibility and rollback availability. The backend remains the authority for eligibility and deployment decisions.
NOT READY / NONE / UNAVAILABLE in this example indicate that no production deployment was available at capture time.
Monitoring & diagnostics
Check service health and operating status.
How this screen is used
The monitoring page groups server health, license, runtime, dataset lineage, active model, validation jobs, mode mismatch and related diagnostics.
HEALTHY server status does not mean the ML pipeline is production ready; other readiness checks can still fail.
Server logs
Trace operational activity through timestamped events.
How this screen is used
Use severity, source and message columns to review API calls and investigate behavior. Save or copy log entries when requesting technical support, after removing private information.
Routine HTTP 200 log entries confirm an individual request response, not end-to-end trading-system readiness.
Server-managed settings
Locate runtime and configuration information.
How this screen is used
The settings screen reports runtime mode, API port, model artifact references and license status, as supplied by the server. The displayed configuration is read-only in this interface.
Development paths, ports and artifact filenames are environment-specific and should not be treated as public setup defaults.
About & legal
Review software attribution, ownership and risk information.
How this screen is used
The About & Legal screen contains product identity, licensing and intellectual-property notices, technical warnings and a futures-trading risk disclosure.
Historical, simulated, replay, machine-learning and backtest performance do not guarantee future results.
Explore Xtrend Algo V10
Discover order-flow research, automation, included indicators and the wider NinjaTrader workflow. Trading carries risk; results depend on market conditions, configuration, risk management and how the system is used.
Explore Xtrend Algo →