Is your portfolio actually working for you?
Most allocation decisions still rely on assumptions that data has long since complicated. This workshop explores how machine learning changes what's possible - not as a shortcut, but as a more honest lens on risk and return.
When the material gets hard
Difficulty is expected here. What matters is what's in place when you hit it - and there's more than you might expect.
Async Q&A that actually gets answered
Questions posted between sessions receive written responses from the same practitioners who deliver the workshops - not a support queue. Turnaround is typically within one working day, sometimes faster.
Peer cohorts working through the same problems
Each intake is small enough that participants recognise each other's work. Collaborative exercises are built into the structure - not optional extras. You are not working through this alone.
Code review on submitted assignments
Submitted work on Python-based exercises gets annotated feedback - what ran, what didn't, and why. The focus is on understanding the logic, not just fixing syntax.
Reference materials stay accessible after completion
Notebooks, reading lists, and recorded walkthroughs remain available after the workshop ends. The point is retention, not just completion.
What the experience tends to feel like
There is no clean linear path through this material. Most participants describe something closer to a cycle - confusion, then clarity, then a harder question. That pattern is expected and built into how the programme is structured.
Disorientation is normal at the start
The first sessions surface assumptions you didn't know you had. Mean-variance thinking, for instance, breaks down in ways that feel uncomfortable before they feel useful. That friction is the point.
Momentum builds around week three
By the time you're running your first regularised regression on historical return data, the earlier theory starts clicking. The shift from abstract to applied is where most participants report feeling the work become real.
The end is less a finish line than a reference point
Participants leave with a working project, annotated code, and a clearer sense of what they don't yet know. That last part is intentional - it's more useful than false confidence.
The people behind the content
The workshop was developed by practitioners who have worked in quantitative finance and applied ML research - not academics writing about the field from a distance. Their backgrounds shape what gets taught and what gets left out.
Teboho Nxumalo
Lead Instructor · Quantitative Methods
Spent eight years building factor models for a mid-size asset manager before moving into education. His sessions on covariance estimation and shrinkage methods draw directly from that work.
Oksana Bilyk
Curriculum Design · ML Research
Holds a research background in reinforcement learning applied to dynamic asset allocation. She designed the assignment structure and the sequencing of hands-on exercises across the programme.
Dariusz Kwiatkowski
Data Infrastructure · Applied Sessions
Handles the technical scaffolding - the datasets, the notebooks, the environments that participants work in. His applied sessions focus on what breaks in production and why that matters before you get there.
Years of programme development
The curriculum has been revised through multiple cohorts since Hunt Join was founded. What exists now reflects what actually worked, not what looked good on paper.
Participants rated the programme
An aggregate rating of 4.6 across 183 reviews. The critical responses are read carefully - they tend to point at the same friction points the team is already working on.
Provinces with active participants
The online format was built to serve participants across South Africa equally - not as an afterthought, but as a core design requirement from the beginning.