Hunt Join logo
Hunt Join Machine Learning · Portfolio Optimisation
Hunt Join - since 2014

Machine
learning,
applied.

Portfolio optimisation is a craft that sits at the intersection of mathematics, data, and judgment. We built a place where practitioners can develop all three through hands-on work, not passive watching.

11+ years of workshops
delivered nationally
38+ regions reached
across South Africa
6k participants who have
completed a programme
Our programmes Get in touch
Workshop participants working through a machine learning portfolio exercise

The problem that started everything

Most ML education treats finance as an afterthought - a dataset you plug into a generic model. The output looks plausible on screen but falls apart the moment real constraints appear: transaction costs, regime shifts, correlation instability. We kept seeing capable people trip over exactly these gaps.

Hunt Join was built around a single conviction: that portfolio optimisation deserves its own dedicated learning environment, not a borrowed one from another discipline.

Practical over theoretical

Every module is structured around a real decision a portfolio manager or quant analyst would face. Participants work with actual market data, implement models in Python, and stress-test their assumptions against historical scenarios. The goal is not a certificate - it is a repeatable skill.

Workshops are designed so that someone working through them alone in Limpopo has the same depth of experience as someone attending a session in Cape Town. Equal access was a design requirement, not an afterthought.

Where we are now

The curriculum now spans reinforcement learning for dynamic allocation, factor model construction, and risk-parity frameworks. Each area is taught through step-by-step assignments with checkpoints that surface misconceptions before they calcify.

We are a small team and we intend to stay that way. Depth over breadth.

Close-up of a data visualisation from a portfolio optimisation exercise

Participants work with live data pipelines during each workshop session - not simplified toy datasets.

  • Specificity over generality

    Every lesson targets one concrete skill, not a broad category.

  • Transparency about difficulty

    We tell participants upfront which parts are hard and why.

  • Regional access as a standard

    All materials are built to work without a fast connection.

  • Curriculum updated each quarter

    Models that were cutting-edge in 2021 are not taught as current.

The people behind the curriculum

A small group with specific backgrounds - no generalists, no growth hackers. Each person on the team has worked in the domain they teach or supports.

Tarquin Botha, Lead ML Research Scientist at Hunt Join

Tarquin Botha

Lead ML Research Scientist

Tarquin spent eight years building quantitative strategies for an asset manager in Johannesburg before moving into education. He designs the core ML modules and keeps the curriculum aligned with what practitioners actually encounter in production environments.

Quantitative Strategy

Ines Ferreira

Curriculum Architect

Ines studied applied mathematics and then spent several years teaching finance at a university in Pretoria. She structures how ideas sequence across a programme - deciding what a participant needs to understand before they can meaningfully tackle the next concept.

Applied Finance & Pedagogy

Dariusz Kwiatkowski

Platform Engineer

Dariusz built the interactive exercise infrastructure that runs every assignment on the platform. His focus is on making data pipelines accessible - participants should be able to run a real backtest without spending half a session configuring their environment.

Data Pipelines & Tooling