Portfolio optimisation

Markowitz and asset selection for a Sharpe contest

2026MIAX master's, Instituto BME

Figure 1. Efficient frontier with the maximum-Sharpe portfolio

Summary

Diversification analysis, Markowitz optimisation and asset selection over 50 anonymised assets, for a class competition scored on annualised Sharpe ratio. The maximum-Sharpe portfolio reached 1.16 on the training data.

What it does

A class competition scored on annualised Sharpe ratio, over 50 anonymised assets and 1,760 days of returns. The code is split into modules: data preparation, equal-weight portfolios and diversification, Markowitz optimisation, asset selection by diversification and Sharpe, a comparison of points along the frontier, and a final check of the constraints before export.

Result

The maximum-Sharpe portfolio reached 1.16 on the training data. That is an in-sample figure, so it says more about the method than about how the portfolio would do next.

Figures

**Figure 2.** Total risk against number of assets, split into specific and systematic risk

**Figure 3.** Weights of the optimised portfolios by size

**Figure 4.** The 15 assets with the lowest beta

Built with

PythoncvxpyNumPypandasSciPymatplotlib