Using Convex Optimisation to Balance Alpha, Costs, and Portfolio Risk

Summary

This article demonstrates a convex optimisation workflow for a crypto perpetual futures portfolio. It combines expected returns estimated from cross-sectional momentum and carry features with a breakout signal, then uses a covariance estimate to represent portfolio risk. Trading costs and portfolio constraints enter the optimisation directly, allowing target holdings to reflect several tradeoffs at once. The article compares this flexible setup with an earlier no-trade-buffer heuristic, which is simpler to understand but less direct for adding risk models and other constraints.

The simulation explores how risk aversion and propensity to trade affect the optimiser’s choices, and shows a net exposure constraint limiting the portfolio’s net long position. The forecasts, rather than the optimiser, supply any potential alpha. The approach depends on reliable expected-return and covariance estimates; the author cautions that covariance estimates can be uncertain and that optimiser decisions may be sensitive to them. The example illustrates a research method, not proof that the signals or resulting strategy will remain profitable.

Key ideas

• The optimiser balances forecast returns, trading costs, risk estimates, and portfolio constraints.
• Expected returns come from momentum, carry, and breakout features in the crypto futures example.
• Risk aversion and propensity to trade shape the optimiser’s trading decisions.
• Optimisation helps use forecasts but does not create alpha.
• Covariance estimation uncertainty can materially affect portfolio choices.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.