The Objective
The primary role of the XTNL entity is to construct a lasting financial legacy through the systematic arbitrage of quantifiable, statistically verified inefficiencies in the foreign exchange market.
This is not a speculative venture. The business operates on a single, non-negotiable premise: every unit of capital deployed must be justified by a proven statistical edge, sized by a risk model derived from that edge, and audited continuously against live performance.
The systematic capture of recurring, statistically verifiable pricing inefficiencies in the EUR/USD spot market.
Disciplined reinvestment of realised returns, sized precisely against a risk model derived from Monte Carlo tail-risk analysis.
The long-horizon objective: a self-sustaining compounding architecture that outlasts any single market regime or edge lifecycle.
Research Framework
Every component of the XTNL system — the statistical models, the risk engine, the execution architecture, the walk-forward validation — exists to answer one or more of the following questions with empirical rigour.
System Architecture
The XTNL architecture is deliberately decoupled into distinct layers, ensuring that no single component — including the edge itself — can compromise the integrity of the whole.
The Statistical Engine
A multi-dimensional analytics pipeline that computes SQN, expectancy, CVaR, walk-forward OOS metrics, and regime labels across 14 distinct data subsets — continuously auditing the health of the underlying edge.
The Risk Architecture
Position sizing derived entirely from a Monte Carlo Conditional Value at Risk (CVaR) model. Risk allocation is not fixed — it is dynamically computed from the tail-risk distribution of the edge's own historical outcomes.
The Execution Layer
A deterministic, firmware-level execution system that removes the human operator from real-time capital decisions. The operator provides coordinates; the firmware determines sizing and executes without discretionary override.
IP / Infrastructure Decoupling
The infrastructure chassis — analytics, risk engine, execution firmware — is permanently owned and continuously refined. The intellectual property (the specific edge) is treated as a replaceable payload, isolated so that edge decay never corrupts the architecture.
Deployment Status
XTNL operates across multiple data universes with different validation statuses. The figures below represent the current state of the production system as reported by the analytics pipeline.
SESSION_FILTERED — Primary Core
ELITE · SQN 4.253FULL_OPTIMAL — Aggregate Universe
SUPERB · SQN 5.211LIVE Execution — Current Deploy
CAUTION · SQN 0.064Note on live vs. forward-test data: The SESSION_FILTERED (N=106) and FULL_OPTIMAL (N=308) datasets span the full forward-test execution period. The LIVE sample (N=29) represents actual capital-at-risk trades in the current deployment phase. A small live N at CAUTION status is expected at this stage of the deployment lifecycle — statistically, 29 trades is insufficient to make probabilistic claims about live performance.
Governance
When the person managing capital has different incentives from the person who owns it, interests diverge. XTNL treats this as an engineering problem — every misalignment vector is addressed with a deterministic, code-enforced constraint. The operator is not trusted. They are measured.
Commission gating
The operator earns zero commission if execution efficiency drops below 88% or if they miss more than 25% of qualified opportunities. Profitability alone does not qualify — quality of execution is the gate.
Performance-gated risk
Position sizes are algorithmically tied to the operator's demonstrated efficiency score. Below 80%, sizes are cut to 60%. Below 40%, the system halts. These are firmware constraints — not policies the operator can override.
Capital lock mechanism
Fresh capital is not injected unconditionally. The system requires three consecutive weeks of ≥ 88% efficiency before unlocking the capital pool. A single poor week resets the streak to zero.
Lucky trade isolation
The system explicitly identifies trades where the outcome was good despite poor execution. These are flagged as 'lucky' and excluded from efficiency calculations — the operator does not benefit from accidental profit.
A third governance layer — a large language model auditor — reads the weekly performance report and produces a qualitative assessment that no statistical model can replicate: detecting narrative inconsistency, identifying rationalisation patterns, and flagging commentary that contradicts the observable data.
Read the full governance architecture →Explore
Institutional Prospectus
Full 11-section technical thesis covering statistical validation, architecture, risk models, and capital scaling.
Read →InteractiveInteractive Simulator
1,000-iteration Monte Carlo engine. Adjust every parameter — edge decay, tax, capital injection, drawdown halt — in real time.
Read →AnalyticsSystem Data
All production metrics, R-distribution charts, hourly performance heatmap, SQN benchmarking, and WFO validation tables.
Read →