<p>Accurately forecasting stock prices is challenging owing to the inherent volatility, non-linearity, and uncertainty of financial markets. Traditional econometric and statistical models often fall short in capturing these dynamics, while advanced machine learning approaches may face limitations such as high computational cost and overfitting. To tackle these challenges, this exploration proposes a hybrid Artificial Bee Colony–Decision Tree (ABC-DT) scheme for stock price projection. The ABC optimizer is employed to fine-tune the Decision Tree framework, enhancing its predictive capacity and robustness. Using nine years (2013–2022) of daily OHLCV (open, high, low, close, volume) data from the Nikkei 225 index, the ABC-DT model is compared with standalone DT and other hybrid models (PSO-DT, ALO-DT). Experimental results show that ABC-DT achieves superior performance with an R² of 0.9735 and a MAPE of 0.4919, outperforming benchmark models in both accuracy and reliability. Cross-validation and feature ablation studies further confirm its consistency and robustness. The contributions of this work are threefold: (1) the integration of swarm intelligence (ABC) into the DT framework for improved hyperparameter optimization, (2) a comprehensive empirical evaluation using a large and volatile stock index dataset, and (3) demonstration of the scheme’s efficacy in reducing forecast error and improving decision-making support for investors and analysts. Overall, the ABC-DT model provides a computationally efficient and dependable framework for stock market forecasting.</p>