Analysis of Historical Market Data to Justify Portfolio Decisions

Roman Shevchenko

Bakalářská práce

Abstract:

The main objective of this bachelor thesis is to develop a practical portfolio optimisation model based on historical market data using statistical analysis and financial indicators to justify investment decisions. The study uses a quantitative approach with Python and its specialised libraries (yfinance, NumPy, Pandas, and SciPy) to implement the Markowitz mean-variance optimisation model. The methodology includes the calculation of key risk-adjusted indicators such as annual volatility, Sharpe ratio, and historical value at risk (VaR). To ease empirical analysis and visualisation, was developed a web-interface based on the Streamlit framework. Empirical research conducted over a ten-year period (2013–2022) for five different asset classes demonstrated that data-driven optimisation significantly improves portfolio performance compared to naive, equally weighted diversification. The algorithm identified two main optimal distributions: a portfolio with minimum variance, which successfully reduced annual volatility to 8.81%, and a portfolio with maximum Sharpe ratio, which maximised risk-adjusted return with a Sharpe ratio of 0.651 (exceeding the benchmark of 0.370). The results confirm that mathematical optimisation, which is highly dependent on asset correlation structures, provides a better compromise between risk and return and serves as a powerful, data-driven tool for strategic asset allocation.

Keywords

Portfolio Optimization, Historical Market Data, Stock market analysis, Investment Strategy, Risk and Return Analysis

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 18. 3. 2026

Obhajoba závěrečné práce

• Obhajoba proběhla 25. 5. 2026
• Vedoucí: PhDr. Ing. Tomáš Maier, Ph.D.
• Oponent: Pavel Kotyza, Ing.

Česká zemědělská univerzita v Praze

Provozně ekonomická fakulta

Bakalářský studijní program:
Informatics