№ I · SCORING+
Scoring+ Evaluation
30+ metrics across 7 weighted dimensions evaluate every subsystem for inclusion in the active portfolio.
A multi-layered quantitative methodology integrating evolutionary optimization, machine learning risk prediction, and 7-dimensional subsystem evaluation — designed to produce decorrelated, regime-adaptive portfolios at institutional scale.
A four-stage pipeline integrating subsystem evaluation, evolutionary optimization, correlation analysis, and machine learning risk prediction.
30+ metrics across 7 weighted dimensions evaluate every subsystem for inclusion in the active portfolio.
Evolutionary multi-objective search explores millions of portfolio combinations for Pareto-optimal allocation.
Multi-dimensional analysis reduces simultaneous drawdowns through deeply decorrelated portfolio construction.
XGBoost models continuously monitor and adapt portfolios to changing market regimes in real time.
Seven weighted dimensions with 30+ underlying metrics. Dynamic weighting adapts to market regimes in real time.
Absolute return · Smart Sharpe Ratio · Alpha generation · Benchmark beta.
Rolling window stability · Regime-dependent behaviour · Return autocorrelation.
VaR & Conditional VaR (CVaR) · Max DD / DD length / DD std dev · Tail risk (skewness & kurtosis).
Recovery efficiency · Stop-loss compliance · Dynamic risk allocation.
Stress period performance · Tail correlation · Black swan resilience.
Trade frequency & timing · Execution efficiency · Slippage analysis.
Average holding duration · Relative Net Asset Exposure · Entry/exit rule efficiency.
Multi-dimensional correlation analysis across subsystems, assets, and time — far beyond traditional pairwise comparison.
Analysis identifies correlations between returns of different subsystems over time to combine strategies with complementary performance characteristics.
A three-dimensional correlation matrix captures relationships between individual assets across all subsystems, evolving over time to show how asset correlations shift under different market conditions.
Time-series analysis detects regime changes and time-dependent patterns that static measures would miss.
The correlation analysis informs multi-objective optimisation by simultaneously minimising portfolio variance, minimising absolute weighted correlation, and maximising the Smart Sharpe Ratio.
State-of-the-art evolutionary algorithm exploring millions of portfolio combinations across billions of data points.
The process repeats over many generations, producing a Pareto frontier of optimal portfolios — each representing a unique trade-off between risk-adjusted return, variance, and correlation.
NSGA-II processes vast quantities of data impossible for human analysts. Millions of portfolio combinations tested against billions of data points.
Multi-dimensional correlation analysis achieves diversification levels far exceeding simple asset allocation strategies.
Iterative evolutionary process allows rapid adaptation to changing market conditions and correlation regime shifts.
Algorithmic processing eliminates cognitive biases and emotional responses that lead to suboptimal investment decisions.
Proactive, adaptive, and comprehensive — eight layers of protection beyond portfolio allocation.
A machine learning model trained on historical performance metrics, market indicators, macroeconomic variables, and correlation data. Predicts subsystem underperformance probability and drawdown likelihood for pre-emptive risk mitigation.
An adaptive risk-parity approach dynamically adjusts risk allocation across subsystems based on predicted risk-return characteristics and current market conditions.
Historical scenarios (2008 GFC, 2020 COVID, 2022 Crypto Deleveraging, SVB) plus Monte Carlo simulation generating 10⁶ synthetic return paths per strategy. Strategies must survive the 99.5th percentile drawdown path before deployment.
Integrated liquidity constraints and transaction cost models ensure strategies remain executable under stressed market conditions.
End-to-end pipeline from data collection through execution and reporting.
Active research areas ensuring strategies remain at the cutting edge.
Many-objective evolutionary algorithms for increasingly complex optimisation problems beyond the current three objectives.
Quantum-hybrid models exploring incomparably larger solution spaces. Targeting near-real-time model retraining — currently impossible with classical computing alone.
Autonomous systems that adjust to regime changes in real time, enabling instantaneous strategy and risk management recalibration.
Satellite imagery, supply chain analytics, NLP-derived sentiment, and other non-traditional data sources to enrich analytical capabilities.
10+ additional strategies currently in development, progressively expanding coverage across asset classes and market conditions.
This document describes proprietary research methodology developed by Divitae Assets for informational and transparency purposes only. It does not constitute investment advice, a financial promotion, or an offer of any financial service. Divitae Assets is a technology and research company — it does not provide investment services, portfolio management, investment advice, or accept client funds. All trading strategies are licensed exclusively to independent regulated partners authorised by the FCA, CySEC, or SEC depending on jurisdiction, which are responsible for all client-facing regulated activities. Past performance is not indicative of future results. Trading in financial instruments carries a high degree of risk and may result in the loss of all invested capital.