Guo Dong Hou

Two key challenges in financial distress prediction are pronounced class imbalance between majority and minority classes and the persistent misclassification of hard-to-learn samples. To tackle these issues, this study proposes an ensemble framework called Dynamic Weight-Adjusted Random Forest Boost (DWARFB). The proposed method incorporates a dynamic sample-weighting mechanism that leverages cumulative misclassification information, adaptive minority–majority class ratios to address class imbalance issue, and a real-time performance-driven strategy to integrate models’ prediction results. The effectiveness of DWARFB is evaluated using a financial dataset from the China Stock Market & Accounting Research (CSMAR) database. Comparative experiments against eight benchmark Random Forest (RF) approaches show that DWARFB delivers superior balanced performance, and a stable precision–recall trade-off, which effectively reduces both false negatives and false positives in the prediction. Moreover, a loss-based feature contribution metric provides economically meaningful insights into the key financial determinants of distress, enhancing model interpretability. Overall, DWARFB demonstrates strong reliability and adaptability and offers a practical solution for early financial distress warning in imbalanced and dynamic financial environments.

One of the key challenges in financial distress data is class imbalance, where the data are characterized by a highly imbalanced ratio between the number of distressed and non-distressed samples. This study examines eight resampling techniques for improving distress prediction using the XGBoost algorithm. The study was performed on a dataset acquired from the CSMAR database, containing 26,383 firm-quarter samples from 639 Chinese A-share listed companies (2007–2024), with only 12.1% of the cases being distressed. Results show that standard Synthetic Minority Oversampling Technique (SMOTE) enhanced F1-score (up to 0.73) and Matthews Correlation Coefficient (MCC, up to 0.70), while SMOTE-Tomek and Borderline-SMOTE further boosted recall, slightly sacrificing precision. These oversampling and hybrid methods also maintained reasonable computational efficiency. However, Random Undersampling (RUS), though yielding high recall (0.85), suffered from low precision (0.46) and weaker generalization, but was the fastest method. Among all techniques, Bagging-SMOTE achieved balanced performance (AUC 0.96, F1 0.72, PR-AUC 0.80, MCC 0.68) using a minority-to-majority ratio of 0.15, demonstrating that ensemble-based resampling can improve robustness with minimal impact on the original class distribution, albeit with higher computational cost. The compared findings highlight that no single approach fits all use cases, and technique selection should align with specific goals. Techniques favoring recall (e.g., Bagging-SMOTE, SMOTE-Tomek) are suited for early warning, while conservative techniques (e.g., Tomek Links) help reduce false positives in risk-sensitive applications, and efficient methods such as RUS are preferable when computational speed is a priority.

This research investigates financial distress prediction, a crucial concern in corporate finance and risk management. Employing the Z-score methodology, traditionally used in financial risk prediction, the study calculated the distress value of companies. Utilizing Out-of-Bag (OOB) error metrics, the study confirms the acceptability of Random Forest for this task. The results demonstrated that the Random Forest regression achieved low error rates (MSE, RMSE, MAE) and high explanatory power (R ² ) on training data, indicating strong predictive capability. However, performance varied significantly on unseen test data, suggesting the need for further tuning to enhance generalization. Optimizing the model based on the OOB error resulted in the lowest error of 0.1667 with 700 regression trees, yielding an average MSE of 391.43, RMSE of 17.92, MAE of 11.29, and R2 of 0.41 on test data. Despite these improvements, the model's performance on unseen data remains suboptimal due to the limited number of indicators and an insufficient number of distressed companies in the samples. Further research needs to focus on data comprehensively, feature selection and data imbalances.

Financial distress is when a company faces significant difficulties meeting its financial obligations and maintaining profitability, leading to bankruptcy, insolvency, and severe economic losses. Therefore, early warning for companies at risk of financial distress is vital for business stakeholders to take timely corrective actions and avoid adverse outcomes. Existing financial distress predictions often rely on historical datasets, incorporating various indicators collected through varied methodologies and experts’ opinions. The challenge arises in discerning which indicators are pivotal for predicting corporate distress, as their influence and relevance may vary. This study proposed a machine learning framework to eliminate variations of different experts’ knowledge when selecting pivotal indicators. Data containing 4006 companies and 204 indicators was extracted from CSMAR. The Chi-square test is employed to select significant indicators. The correlation of these selected indicators is modeled using the C4.5 decision tree. Results showed that this selected feature set is closely aligned with those obtained when utilizing all features in the data. A thorough comparison of the indicators selected by the expert revealed notable distinctions. Features chosen by the Chi-square test are related to financial ratios and also exhibit a pronounced focus on societal attention, shareholding concentration, and market dynamics.