Core Technology

Three systems.
One unfair advantage.

Three machine systems operate in real time — evaluating, optimising, and managing risk across changing market regimes. No human discretion in the live loop. No client capital in our infrastructure.

10B+
Data Points Processed
30+
Risk Metrics
<5ms
Signal Latency
24/7
Active Monitoring
SECTION · 01

The three machines.

Each system has one job and does it without sentiment. Together they evaluate every subsystem in production, build a Pareto-optimal portfolio, and pre-empt drawdowns before they materialise.

№ I · OPTIMISATION
α

NSGA-II Optimizer

Multi-objective evolutionary search

An elitist non-dominated sorting genetic algorithm exploring millions of portfolio combinations. Optimises three objectives simultaneously — maximising Smart Sharpe, minimising variance, minimising absolute weighted correlation — and surfaces a Pareto frontier of trade-offs.

  • Population & crossover · diversity injection
  • Tournament selection · elitism preserved
  • Pareto-rank non-dominated sorting
  • Re-runs across generations as data refreshes
№ II · EVALUATION
β

Scoring+ Framework

Proprietary 7-dimension evaluation

30+ underlying metrics across seven weighted dimensions evaluate every subsystem competing for portfolio inclusion. Dimension weights adapt dynamically to the current market regime, so what counts as "good" today is recomputed for tomorrow's conditions.

  • Performance · Smart Sharpe, alpha, beta
  • Consistency · regime-dependent stability
  • Risk · VaR, CVaR, tail moments
  • Drawdown control · recovery efficiency
№ III · RISK PREDICTION
γ

XGBoost Risk Engine

ML-driven pre-emptive de-risking

A gradient-boosted decision tree model trained on historical performance metrics, market indicators, macroeconomic variables, and correlation regime data. Predicts subsystem underperformance probability and drawdown likelihood — feeding pre-emptive risk-budget reallocations before drawdowns deepen.

  • Subsystem underperformance probability
  • Drawdown likelihood & magnitude
  • Correlation regime change detection
  • Continuous re-training on live results
SECTION · 02

How signals reach the broker.

From data ingestion to relayed instruction, the operational pipeline is fully automated. Five gates separate raw market data from the trade tickets the regulated partner executes on the client's behalf.

№ 01
Data ingestion
Market feeds, broker order flow, alternative data — normalised and stored in a distributed repository.
№ 02
Subsystem evaluation
Scoring+ runs across 30+ metrics in seven weighted dimensions, regime-adaptive.
№ 03
Portfolio construction
NSGA-II searches the Pareto frontier; an allocation matrix is selected based on current risk budget.
№ 04
Risk overlay
XGBoost forecasts feed dynamic risk budgeting; stress tests gate strategy deployment.
№ 05
Trade relay
Coordinated signal relay to the regulated partner firm, which executes against client accounts.
SECTION · 03

Capital and code,
kept apart.

Our infrastructure produces signals — never custodies funds. Client capital sits at the regulated broker, segregated and supervised by our regulated partners. We license the technology; the partner runs the firm.

DIVITAE ASSETS

Technology & research only.

Two entities (UAE + Cyprus) develop algorithms, infrastructure, and risk-management systems. We do not hold client funds, do not execute trades, and do not interact with end investors.

REGULATED PARTNERS

FCA, CySEC & SEC authorised execution & custody.

Independent partners authorised by the FCA, CySEC, and SEC are the exclusive licensees of our technology. Onboarding, KYC/AML, custody, execution, reporting, and regulatory obligations sit entirely with the partner.

SECTION · 04

Operating discipline.

Continuous oversight of every signal, every position, every metric. Four operational layers run alongside the three systems above — making sure the platform behaves the way it was designed to.

LAYER · 01

24/7 monitoring

System health, data integrity, signal latency, broker connectivity — observed continuously with automated alerting.

LAYER · 02

Independent audit

Live-account performance is verified independently by Alpha Verification under GIPS® standards. Investor login is available on request.

LAYER · 03

Stress & scenario testing

Historical regimes (2008, 2020, 2022, SVB) plus 10⁶ Monte Carlo synthetic paths per strategy. Deployment requires surviving the 99.5th percentile drawdown.

LAYER · 04

Continuous re-training

Risk and scoring models retrain on live data; dimension weights and risk budgets adapt as market regimes shift — without human discretion in the live loop.

Read the methodology in full.

The 7-dimension Scoring+ framework, the NSGA-II evolutionary process, the XGBoost risk model, and the 9-step infrastructure pipeline — documented in detail.

Open the methodology Speak with the team →