This study aims to explore the role that artificial neural networks (ANNs) can play in improving the measurement of financial performance, and how that improvement might be reflected in the quality of financial reporting. The study also draws a comparison between the predictive efficiency of ANN models and multiple linear regression (OLS) in explaining the effect of discretionary accruals on the financial performance indicators of Iraqi banks listed on the Iraq Stock Exchange over the period 2016–2023. Three financial performance indicators were employed: profit margin, return on assets (ROA), and return on equity (ROE). Discretionary accruals were measured using the Modified Jones Model. A Multilayer Perceptron (MLP) neural network was applied and its results were compared to those of the traditional model using Leave-One-Out Cross-Validation (LOO-CV). The findings revealed variation in model efficiency depending on the financial indicator used. Neural networks outperformed multiple linear regression in predicting profit margin, which reflects their ability to capture nonlinear relationships among financial variables. On the other hand, the OLS model performed better in explaining ROA and ROE, suggesting that the superiority of neural networks is not absolute but rather depends on the nature of the financial indicator being studied. The results also showed that discretionary accruals have an inflationary effect on certain financial performance indicators. The study concludes by stressing the importance of using artificial intelligence techniques in an integrated manner alongside traditional statistical methods to support the quality of financial analysis and enhance the reliability of financial reports.