Most firms assign clients to a model from a menu. At Convexity, every portfolio is built from the ground up — starting with how you think about investing.
You come to the table with a view on risk and return — how much downside you are willing to accept, the return profile you are targeting, and how you would like the portfolio to behave over time. We build around that preference, seeking to manage to a risk budget rather than a benchmark.
You are working toward a specific outcome — retiring at a certain age, funding a major purchase, or building wealth designed to transfer to the next generation. We work backward from that goal, calibrating risk to the return that may be required to reach it on time.
Every portfolio follows the same six-step analytical process — designed so that each decision is grounded in research, not intuition. The goal is a portfolio that sits on the efficient frontier: the point where every unit of risk is working as hard as it can.
Every accessible asset class catalogued — no category excluded before filters are applied.
Every investable instrument across every accessible asset class is catalogued. No category is excluded before the filters are applied. The result is a complete candidate set — ten asset classes evaluated across four dimensions: diversification value, cost, tradability, and structural complexity.
Produces: Complete candidate set — ten asset classes, four evaluation dimensions View the Investment UniversePrice behavior mapped instrument-by-instrument — diversification quantified before entry.
Each instrument's price behavior is mapped against every other. The matrix reveals what diversification actually looks like before a single position is taken — measured across normal conditions and stress periods alike. Correlations that appear stable in calm markets often break down precisely when diversification is needed most.
Produces: Full correlation matrix — stress-tested across multiple regimesReturn per unit of risk maximized — position-level limits discipline the output.
The model seeks to maximize return per unit of risk across all candidates. Position-level limits then discipline the model's output — preventing any single instrument from dominating the portfolio regardless of how favorable its risk-return profile appears in isolation. The optimizer seeks the frontier; the constraints are designed to keep the portfolio on it.
Produces: Optimized weight vector — risk-budgeted at the position level Explore the Efficient FrontierModel output tested against conditions it cannot yet price — judgment where math ends.
Quantitative output is evaluated against conditions the model cannot yet price — regime shifts, liquidity dislocations, geopolitical risks, and structural changes that require judgment rather than calculation. The model provides the starting point; experience determines whether the output survives contact with reality.
Produces: Adjusted allocations — model output tempered by market intelligenceNo market timing — the analytical case is the only basis for entry.
Implemented without market timing. The analytical case is the only basis for entry — not price levels, not near-term forecasts, not sentiment. Each allocation is mapped to client-specific constraints, tax considerations, and custodian requirements before a single trade is placed.
Produces: Final target portfolio — mapped to client constraints and custodianEvery trade staged for fill quality — the portfolio you hold matches the analysis.
Every trade is staged for fill quality and reconciled post-execution. The goal is for the portfolio you hold to match the portfolio the analysis determined you should hold. Drift is monitored continuously, and rebalancing is triggered by analytical thresholds — not calendar dates.
Produces: Live portfolio — verified against target within reconciliation toleranceTell us how you think about your portfolio. We'll show you how we'd build it.