VISION & ROADMAP
From 5 FX pairs to multi-asset. The mathematical framework generalizes to any equilibrium-seeking process: equities, rates, commodities. The architecture is designed to scale.
EXECUTION ROADMAP
CURRENT: LIVE
FX Execution Platform
5 currency pairs (CADJPY, AUDJPY, EURUSD, GBPUSD, USDCAD) deployed across multiple timeframes. The full system is operational and trading live.
- Fleet of live daemons across multiple timeframes
- 12-layer gate pipeline fully deployed on every daemon
- Real-time dashboard with 5 tabs: fleet monitoring, macro intelligence, performance analytics, gate diagnostics, and system health
- Macro intelligence layer live: session tracking (Sydney, Tokyo, London, New York), regime detection, macro filter gating, COT positioning, central bank calendar, commodity-currency correlations, and rate path scenarios
- Backtest-to-live gap identified, root-caused, and actively being closed through predictor recalibration
NEAR-TERM: 1 TO 4 WEEKS
Validation & Statistical Assessment
Completing the deployment validation cycle with rigorous statistical verification of live performance against backtested expectations.
- Hold buffer formal 5-fold walk-forward validation to confirm out-of-sample generalization
- Shadow deployment for validation of inactive strategies without capital risk
- Accumulating sufficient post-deployment trades per daemon for statistically meaningful performance assessment
MEDIUM-TERM: 1 TO 3 MONTHS
Pair Expansion & Predictor Calibration
Expanding the instrument universe while recalibrating the prediction models using live performance data to close the backtest-to-live gap.
- Signal refinement integration into the main optimization loop for improved entry and exit timing
- Predictor recalibration using live performance data
- Additional pairs: each new instrument optimizes in minutes per instrument
- External data integration: options volatility surfaces, order flow analytics
LONG-TERM: 3 TO 12 MONTHS
Multi-Asset & Institutional Infrastructure
Extending the mathematical framework beyond FX into asset classes with documented equilibrium convergence behavior, while building institutional-grade infrastructure.
- Multi-asset expansion: equity index futures, interest rate futures, commodities
- Institutional infrastructure: redundant cloud deployment with failover, dedicated database, co-location for latency reduction, compliance framework
- Advanced ML: extending gradient boosting ensemble beyond single-model gating, deep learning for sequence modeling, reinforcement learning for execution optimization
RESEARCH PRIORITIES
HIGH
Walk-Forward CV of Hold Buffer
Formal 5-fold walk-forward cross-validation of the hold buffer to confirm out-of-sample robustness and quantify generalization decay across time windows.
MEDIUM
Regime Detection Sensitivity
Systematic comparison of regime detection thresholds across all daemons. Quantify the false-alarm rate vs. detection-delay tradeoff for regime-change identification.
MEDIUM
Macro Filter Calibration
The macro intelligence layer is live with session tracking, regime detection, and a binary macro filter gate. Research focus: calibrate the filter's activation thresholds using accumulated live data to quantify its impact on drawdown reduction and trade quality.
LOW
Multiple-Testing Correction
Implement a stepwise multiple-testing correction to formally control family-wise error rate across the full ensemble of strategies and parameter sets.
LOW
Multi-Objective Optimization
Replace single-objective optimization with a multi-objective evolutionary approach to simultaneously optimize Sharpe ratio, max drawdown, and trade frequency, producing a Pareto front of strategy configurations.
SCALING VECTORS
CURRENT
5 FX PAIRS
↓
15+ INSTRUMENTS
MULTI-ASSET
CURRENT
2 TIMEFRAMES
↓
5+ TIMEFRAMES
M5 THROUGH D1
CURRENT
SOLO BUILD
↓
INSTITUTIONAL TEAM
QUANT + ENGINEERING
Full pair optimization takes minutes per instrument. Adding new instruments is a configuration change, not an engineering project.
The mathematical framework is asset-class agnostic. The stochastic process model captures equilibrium dynamics in any time series. The 12-layer gate pipeline, ensemble architecture, and validation framework transfer directly to new instruments with minimal adaptation.