Our guiding principles
Simplicity over complexity
Elegant solutions outlast elaborate ones
Rigorous testing
Hypotheses are tested, not assumed
Reproducibility and transparency
Methods withstand independent scrutiny
Safety and robustness
Constraints protect against unforeseen risks
Collaboration
Research and risk work as one
Continuous validation
Live testing confirms what backtests reveal
How our advanced machine learning pipeline works
Our model consumes technical, fundamental, macro and alternative data. After rigorous checks, signals are generated, tested and selected to build a holistic view of every stock.
- Data
- Pre-processing
- Signal generation
- Portfolio contruction
- Continuous learning
- Execution
Ensuring robustness
Each step is robustly designed to build a systematic investment process whilst filtering out noise and signals.
Statistical integrity throughout
We maintain strict standards for data quality and mathematical correctness.
Avoiding overfitting
Models are tested on unseen data to confirm genuine predictive power.
Sensitivity testing
We stress-test assumptions and examine how models behave under different conditions.
Reproducibility checks
Independent teams verify that results can be replicated and methods are sound.
Peer scrutiny
Research undergoes critical examination by independent committees.
The people behind the research
Our team brings together physicists, mathematicians, engineers and computer scientists. We invest heavily in people, tooling and infrastructure, ensuring that clarity and reproducibility are built into how we work. Research, investment and risk collaborate from the start, not as separate functions.