Built for the event that isn't in the model.

AI-managed tail-risk protection for pension funds, endowments and long-horizon investors....

Calm markets — protection cost-efficient, low decayTail event — convex payoff ↑
Positioning

Most portfolios are optimised for the market they've already seen.

Long-horizon institutions carry an unusual burden. They must compound capital across decades while meeting obligations that cannot be deferred — pensions to be paid, endowments to be drawn, beneficiaries who cannot wait for a recovery.

The mathematics of this are unforgiving. Diversification, which functions well in ordinary conditions, has a documented habit of failing precisely when it is needed, as correlations converge toward one.

The risk that matters to a long-horizon investor is not volatility. It is the rare, severe drawdown that permanently impairs the compounding path.

VectorFlow Capital is built for that specific risk, and for nothing else.

The asymmetry of loss

Portfolio loses−50%
Required to recover+100%

A portfolio that loses half its value must gain one hundred percent to return to where it began. This is the mathematics VectorFlow Capital is built against.

How it works

Three mechanisms, one purpose: convexity when it's needed most.

I.

Convexity, not prediction

The strategy holds positions whose value is designed to rise disproportionately as markets move sharply against the portfolio it protects. This is a structural property, not a forecast. No view on timing is required for the mechanism to function.

II.

Options-based construction

Protection is expressed primarily through derivative structures, sized and rolled with the explicit intention of being cost-efficient in calm markets while retaining meaningful convexity in stressed ones.

III.

AI-managed positioning

Structure selection, sizing and roll decisions are informed by machine learning models running on Verreaux — Cirrus AI's high-performance computing platform — drawing on volatility surfaces, correlation regimes and stress signals from real economic activity.

The argument

A hedge is only useful if you can afford to hold it.

The central problem in tail protection is not constructing a payoff that performs in a crisis. It is holding that position through the years in which nothing happens, without the carry cost eroding the returns the protection was meant to defend.

This is fundamentally an optimisation problem — one that rewards better modelling of volatility structure, better selection of instruments and strikes, and better discipline around when to add and when to sit still.

It is the kind of problem that scales with computational capability. Which is why we built the compute first.

The differentiator · real-economy signal

Our stress signals come from more than screens.

VectorFlow Capital sits inside Project Gaius — Cirrus AI's integrated industrial and AI programme in South Africa's Coega Special Economic Zone.

Financial markets are a compression of real-world conditions, and they compress with a lag. Supply chains, industrial throughput, input costs and health systems register stress in their own vocabulary, often before it is priced. An institution operating in all of those domains simultaneously sees a wider aperture than one watching the tape alone.

The closed loop

Protection that reinforces what it protects.

The institutions VectorFlow Capital is designed to serve are the same institutions that fund long-horizon industrial and health investment — including, within Project Gaius, coffee manufacturing, women's health innovation and AI infrastructure.

Capital that survives a crisis intact is capital that can keep funding the next decade of building. Capital that is impaired cannot.

Anti-fragility, in practice: when one part of the system is stressed, the others provide ballast. Industrial production, human health, financial intelligence and systemic risk management are structured to reinforce one another rather than fail together.

Who it's for

Designed for the patient and the obligated.

Institutional investors only — not a retail product
01

Pension funds

02

Endowments & foundations

03

Insurers

04

Sovereign & development funds

Credibility

The groundwork behind it.

Collaboration

Africa's largest financial institution. Use cases identified jointly in 2021, for implementation following operationalisation of Cirrus AI's industrial and health platforms.

Compute

Verreaux HPC. Exaflop-scale AI compute, deployed in Johannesburg — the most powerful AI computing platform on the African continent.

Provenance

Cirrus AI. Founded 2018. Eight years of technical development, industrial engineering and institutional collaboration across industry, academia and government.

Discipline

Aligned to established practice. Draws on convex tail-hedging discipline developed by specialist protection managers over two decades.

Contact

The next severe drawdown will arrive without notice.

VectorFlow Capital is being developed as part of Project Gaius. We welcome enquiries from institutional investors, trustees and their advisers.

Enquiries
Laura Mallory
Head of Marketing, Cirrus AI (Pty) Ltd
EMAIL LAURA MALLORY →