Jul 2026· Signal Processing and Communications Applications Conference· pp. 1-4· 0 citations· 10 references
Abstract
Portfolio construction aims to balance expected return and risk through effective asset allocation. This study proposes a portfolio formation framework that integrates machine learning-based return prediction with Markowitz mean–variance portfolio optimization. Random Forest, XGBoost, Multilayer Perceptron, and Support Vector Regression models are employed to predict the cross-sectional excess returns of stocks using financial indicators derived from technical and macroeconomic variables. These predictions are incorporated into the portfolio optimization process to determine portfolio weights. The resulting strategies are evaluated against benchmark portfolios including an equal-weighted portfolio and the BIST 100 index. Empirical results show that machine learning-based stock selection improves portfolio performance. In particular, the XGBoost-based portfolio achieves the best results with an annual return of 75.30% and a Sharpe ratio of 1.80. SHAP analysis further indicates that momentum and price-based technical indicators play a dominant role in model predictions.
The stock market provides several investment opportunities, but selecting suitable securities and deciding their portfolio allocation requires careful consideration of expected return and risk. This study develops a portfolio optimization framework that combines machine learning-based stock price prediction with Modern Portfolio Theory (MPT). Historical daily data for 21 companies listed on the National Stock Exchange (NSE) were collected from Yahoo Finance for January 2020 to December 2025. The data were cleaned and prepared for exploratory analysis, return calculation and prediction. Linear Regression, Decision Tree, Random Forest and XGBoost were considered for stock-price prediction and evaluated using MAE, MSE, RMSE and R² Score. The project results identify Random Forest as the overall best-performing model. Predicted prices were then used with MPT to determine portfolio weights. The final recommendation selected M&M, SUNPHARMA, BHARTIARTL, NTPC and LT, with allocations of 22.37%, 21.21%, 21.01%, 18.29% and 17.12%, respectively. Portfolio performance was evaluated using annual return, annual volatility, Sharpe ratio, CAGR and maximum drawdown. The framework provides a systematic decision-support approach for combining predictive analytics with portfolio optimization.
Keywords- Machine Learning, Modern Portfolio Theory, NSE, Portfolio Optimization, Stock Price Prediction.
Subhash Naidu B, R. N. Kulkarni· International Journal of Cre...· 0 citations
Constructing an optimal portfolio is a crucial step for investors in balancing the trade-off between expected return and investment risk. This study aims to construct an optimal portfolio comprising two stocks, EMDE and MTDL, by applying the Markowitz mean-variance model to minimize return variance at a specific return level. The methodology employs mean-variance optimization, estimating expected return, variance, and covariance based on historical return data for both stocks to determine efficient portfolio weights. The analysis reveals that the optimal portfolio consists of 19.27% EMDE stock and 80.73% MTDL stock. This combination yields an expected portfolio return of 1.33% with a return standard deviation of 8.43%, reflecting a more efficient risk-return profile compared to an allocation in a single stock. These findings indicate that diversification between EMDE and MTDL can improve portfolio risk characteristics without significantly sacrificing returns. Consequently, investors are advised to consider this combination as part of their asset allocation strategy, particularly those with moderate risk preferences who prioritize mean-variance efficiency. This study provides empirical evidence regarding the application of the Markowitz model in the Indonesian stock market and serves as a reference for future research involving a broader range of assets and data periods.
Zahra Rohadatul Aisylah, Ferdiansyah Saputra, Arief Surya Lesmana et al.· Digital Business and Entrepr...· 0 citations
The capital market plays an important role in the economy by providing investment instruments for investors and financing sources for companies. A capital market portfolio consists of a collection of financial assets, such as stocks, constructed to achieve an optimal return while reducing investment risk. Mean-variance portfolio construction is highly sensitive to parameter estimation errors. Therefore, a robust estimation approach is employed to obtain more stable parameter estimates by minimizing the influence of outliers. This study aims to construct an optimal stock portfolio through diversification, determine stock weights using a two-constraint mean-variance model with robust S-estimation, calculate the expected return and risk, and evaluate portfolio performance. The analysis was conducted using the closing prices of stocks included in the IDX30 Index from October 2024 to September 2025. The results identified nine stocks with positive expected returns from five different sectors. Based on the stock selection criteria, two optimal portfolios were constructed. Portfolio 1 consists of ASII, BRPT, INDF, PGAS, and TLKM, whereas Portfolio 2 consists of ASII, ANTM, INDF, PGAS, and TLKM. Portfolio 1 generates an expected return of 0.137% with a risk of 2.226%, while Portfolio 2 generates an expected return of 0.097% with a risk of 1.319%. Based on the Sharpe and Treynor ratios, Portfolio 1 demonstrates relatively better performance than Portfolio 2.
Anis Faiqo Tuzzainiyah, E. Sulistianingsih, Nurfitri Imro'ah· Jambura Journal of Mathemati...· 0 citations
Portfolio optimization is a fundamental aspect of investment management that focuses on constructing a portfolio capable of delivering the highest possible return while minimizing investment risk. Modern Portfolio Theory (MPT), introduced by Harry Markowitz, provides a quantitative framework for selecting an optimal combination of assets based on their expected returns, variances, and correlations. This study examines the application of Modern Portfolio Theory in optimizing investment portfolios by evaluating the trade-off between risk and return across different asset classes. Historical financial data are analyzed to estimate expected returns, standard deviations, and covariance among selected securities. The efficient frontier is generated to identify portfolios that maximize returns for a given level of risk, while diversification is employed to reduce unsystematic risk. The findings demonstrate that a well-diversified portfolio designed using MPT can significantly improve investment performance compared to investing in individual assets. The study also highlights the practical significance of portfolio optimization in assisting investors, financial analysts, and portfolio managers in making informed investment decisions aligned with their financial objectives and risk tolerance. Overall, the research emphasizes that Modern Portfolio Theory remains a valuable and effective approach for achieving efficient asset allocation and enhancing long-term portfolio performance in dynamic financial markets. Keywords: Portfolio Optimization, Modern Portfolio Theory (MPT), Investment Management, Risk-Return Trade-off, Asset Allocation, Portfolio Diversification, Efficient Frontier.
K. Naveen, Amita Johar, T. Meghana· International Journal of Dat...· 0 citations
This study presents an integrated framework for multi-market portfolio optimization that integrates machine-learning-based return forecasting with classical and downside-oriented risk models. Using daily data for Bitcoin, Ethereum, BNB, Microsoft, and Tesla, the XGBoost algorithm is employed to predict short-term returns, after which the forecasts are used to construct optimal portfolios under the Mean-Variance (MV), Mean-Semi-Variance (MSV), and Mean-Absolute Deviation (MAD) models. A naïve equal-weighted portfolio is included as a benchmark. The results show that the predictive models provide useful signals for portfolio construction, with all optimized strategies achieving markedly lower risk than the naïve portfolio for the same expected return level. Tangency portfolios, particularly the Maximum Sortino and Maximum Sharpe-AD configurations, delivered the highest efficiency gains and the strongest performance on out-of-sample data. Overall, the findings demonstrate that combining machine learning with risk-sensitive optimization tools can improve decision-making in markets characterized by volatility and mixed asset dynamics.
P. Peykani, Daniyal Sabour, Cristina Tanasescu· An International Journal of...· 0 citations
