Investment Process
Lumenai’s investment process follows a repeatable five-step framework combining quantitative models with human oversight. Machine-learning techniques are incorporated to update portfolios as new data emerges, while experienced professionals monitor risk and methodology.
Step 01
Select Universe
We define the investment universe using systematic criteria informed by client objectives and liquidity requirements.
Key criteria:
Securities sourced from major indices
Strict standards for data integrity
Liquidity and pricing frequency requirements
Step 01
Select Universe
We define the investment universe using systematic criteria informed by client objectives and liquidity requirements.
Key criteria:
Securities sourced from major indices
Strict standards for data integrity
Liquidity and pricing frequency requirements
Step 02
Score & Rank
Supervised learning models evaluate securities based on model-derived signals that have historically aligned with certain market environments.
Key signals:
Volatility & trend stability
Fundamental strength
Risk-adjusted return measures
Pattern similarity to past performance
180,000+ data points
500 signals
20,000 features
Step 03
Cluster
Unsupervised learning groups securities with similar historical behavior to classify market regimes and identify outliers.
What this enables:
Market regime classification
Identification of factor characteristics
Removal of anomalous securities
Step 04
Fine Tune
Portfolio construction incorporates investor constraints and trading thresholds.
Portfolio controls:
ESG filters
Tax-related rules
Sector caps
Turnover thresholds
Portfolios are built using equal or optimized weights from the highest ranked securities.
Step 05
Execute & Oversee
Portfolios are implemented and monitored through Lumenai’s OMS/PMS infrastructure.
System monitors:
Market behaviour and drift
Corporate actions
Compliance rules
Performance reporting
This five‑step process is designed to adjust model inputs and exposures systematically as market conditions evolve. All model behavior is based on historically observed relationships and does not imply future performance.
Our Technology Engine
Powered by Data, Algorithms and Scientists
Data
Proprietary database
composed of traditional,
fundamental, alternative
and augmented data
Algorithms
Unique mathematical
models to transform data into insights
Scientists
Data Scientists, mathematicians & engineers
Frequently Asked Questions
Want to
learn more?
Let’s connect.
