ABSTRACT In highly class‐imbalanced financial distress prediction, standard classifiers often fail as they assume balanced class distributions and equal misclassification costs. We propose a heterogeneous voting ensemble method that addresses extreme class imbalance by synergizing sampling, cost‐sensitive learning, and ensemble techniques. First, coordinating sampling and cost‐sensitive learning reduces model bias in identifying distressed firms. Sampling rebalances class distributions, which reduces representation bias and provides an effective data foundation for cost‐sensitive learning. Cost‐sensitive learning assigns higher costs to distressed firms, adjusting the optimization objective to minimize total misclassification costs. Second, by aggregating diverse base classifiers, the heterogeneous voting ensemble reduces potential additional variance introduced by sampling and cost‐sensitive learning, thereby improving generalization. Using data from China, Poland, and the United States, we show that the proposed method significantly outperforms benchmark models in detecting distressed firms. Moreover, the improved predictions translate into reduced misclassification costs, particularly when the cost difference between distressed and healthy firms is greater. Ablation experiments validate the effectiveness of each component, where sampling contributes most to detection accuracy for distressed firms, followed by cost‐sensitive learning, while the ensemble ensures model stability. These results highlight that combining sampling, cost‐sensitive learning, and ensemble techniques is crucial for stakeholders (e.g., investors, creditors, and regulators) to detect rare events such as financial distress.

ABSTRACT I compare the forecasting performance of a wide range of univariate and multivariate forecasting models for both headline and core inflation in Germany. The analysis is conducted using a full sample containing quarterly data from 1995 to 2024, which includes the COVID‐19 pandemic and subsequent inflation surge, and a shortened sample spanning the years 1995–2019, a period of low and stable inflation. I firstly find evidence to support a recurrent finding in the literature that univariate models are hard to beat but show that this holds particularly for single‐indicator models, at times of stable inflation. Furthermore, upon testing 34 indicators for German inflation in both single‐indicator and combination (i.e., factor, pooling, and BVAR) models, I find that the performance of multivariate models is improved by (i) including the years 2020–2024 in which inflation was high and volatile and (ii) by combining information contained within the indicators. Specifically, I find that the multivariate models improve upon the univariate benchmark by up to 20% in the full sample, providing evidence for the performance of forecasting models to vary with the economic environment. The analysis also sheds light on which indicators are particularly useful for forecasting inflation in Germany. Additionally, it appears that multivariate models provide a greater forecasting gain over the univariate benchmark for headline, rather than core, inflation. Finally, a short case study of the inflation pickup following the COVID‐19 pandemic reveals that the multivariate models, particularly the factor, BVAR, and price expectations single‐indicator models, most accurately forecasted inflation at this time.

ABSTRACT Enterprise bankruptcy exerts effects on financially linked firms‘ operations. Such risks may propagate through shareholding relationships, potentially amplifying financial distress and threatening the stability of the broader system. Enterprise risks originate not only from internal factors but also from complex equity relationship networks. Consequently, there is a critical need for enterprise bankruptcy prediction models to support investment and operational risk management. We propose hypergraphs and bidirectional attention‐based dual graph neural networks (HBA‐DGNN) as an innovative approach for predicting enterprise bankruptcy. It consists of two main components. The first component, the hypergraph embeddings of categorical features (HECF) module, can effectively capture higher order relationships among enterprises. Simultaneously, the bidirectional attention‐based GNN (BAG) module quantifies the importance of equity relationships based on enterprise attributes and networks. We conduct an empirical study on the model of Evergrande Group, which faced a debt crisis and caused systemic risks in China's financial market. The HBA‐DGNN demonstrates superior predictive performance compared to baseline models, achieving an average improvement of over 20%. Additionally, attention coefficients in the BAG significantly correlate with enterprise bankruptcy, effectively identifying critical edges and nodes responsible for risk contagion. The HBA‐DGNN effectively predicts enterprise bankruptcy, supporting corporate operations, financial investment, and market supervision.

ABSTRACT The persistent uncertainty surrounding the Trump administration's commitment to addressing climate change raises concerns that global financial stability could be compromised through spillovers from US climate‐related risks. This study examines this hypothesis by analyzing the dynamic interactions between global financial instability and US climate risks using nonlinear time‐varying connectedness based on a TVP‐VAR framework and VAR‐based local projection causality methods. It distinguishes between physical and transition risks and considers various forms of systemic financial stress at the global level, including banking and funding stress, equity market stress, safe asset stress, and market volatility, in order to inform the prediction of future global financial instability. Our findings highlight that global financial instability reacts differently depending on the nature of US climate risks. We show that transition risks have stronger predictive relationships than physical risks on global financial stability, especially after major events such as the Paris Agreement and the COVID‐19 pandemic. Whereas US physical risks tend to propagate instability via safe asset markets, transition risks exert more direct pressure on credit and equity markets. In particular, safe asset markets, which absorb shocks under US physical risks, can also transmit these shocks to credit and equity markets under transition risks. This suggests the presence of flight‐to‐safety versus flight‐to‐quality behaviors among global investors, depending on the specific type of US climate risks. Overall, these findings show how climate‐based financial mechanisms can strengthen early‐warning signals that inform the forecasting of systemic financial risk.

In the published article, the following typographical errors are present in the reference list.These errors

ABSTRACT This paper presents an evaluation of the accuracy of machine learning (ML) techniques in forecasting the realized volatility of West Texas Intermediate (WTI) crude oil prices. We compare several ML algorithms, including regularization, regression trees, random forests, and neural networks, to several heterogeneous autoregressive (HAR) models. The results show that the ML and regularization methods are efficient, even when there are only three predictors: daily, weekly, and monthly realized volatility (RV) lags. In addition, when the ML and regularization methods are applied, the results become more pronounced over longer forecasting horizons, as well as for weekly and monthly horizons. These ML methods are effective in approximating long‐term realized volatility. Furthermore, we found that additional explanatory variables for real‐time currency exchange contain valuable information to forecast the RV of crude oil prices.

ABSTRACT This paper forecasts the quantile connectedness among Global CSR and Sustainable Indices, considering daily prices extending from December 30, 2019 to March 27, 2024. Even if both CSR indices and sustainability indices assess companies on nonfinancial performance, they partially differ in scope and focus. Where corporate social responsibility (CSR) indexes primarily evaluate how businesses fulfill their ethical and social responsibilities, emphasizing aspects such as corporate governance, labor practices, community engagement, sustainability indices take a broader, long‐term view, assessing how companies align with sustainable development goals (SDGs), manage risks, and produce resilience. The examined indices are based on the above distinction and respond accordingly in terms of their prices. To capture both CSR and sustainable indices, four stocks of each based on market capitalization are considered. For the empirical computation, quantile vector autoregression (QVAR) is employed. The results reveal a high connectedness among the examined markets in the lower, middle, and upper quantiles. In addition, the financial shocks due to natural and man‐made outbreaks do not impair the interconnectedness among the examined indices. Further, the S&P_TSX Renewable Energy and Cl (RSPRECTI) indices showed the least transmission of the shock in lower and middle quantile, which may be preferred by investors and portfolio managers in their portfolio baskets. This article offers practical implications to markets' investors and stakeholders.

ABSTRACT This paper examines the significance of technical indicators in the context of machine learning. We commence by constructing a machine learning–based cross‐sectional asset pricing model that incorporates 124 previously identified asset pricing characteristics and fourteen macroeconomic factors. Subsequently, we assess widely regarded technical indicators in their ability to influence machine learning–based model accuracy. Our findings indicate that Bollinger Bands, moving averages, momentum reversal, MIN–MAX retracement, trender, and the parabolic time/price system have a significant influence on the predictability of machine learning–based models. Interestingly, we find that a well‐curated machine learning model with technical indicators is comparable in performance to models with much larger asset pricing characteristics. We also document a time‐varying phenomenon in the predictability of asset pricing models. Finally, long–short portfolio analysis suggests that active investment using technical indicators and machine learning can yield positive alpha.

ABSTRACT Forecasting high‐frequency stock prices is a significant challenge due to inherent noise, non‐stationarity, and complex market dynamics. Conventional models often struggle to effectively extract meaningful signals from raw price and volume data. To address this, we propose an innovative hybrid framework, EWOA‐VMD‐ATT‐BiGRU, which introduces several key novelties to enhance prediction accuracy. First, our model uniquely decomposes both price and volume sequences as an innovative feature engineering, which allows for the unveiling of multi‐scale market characteristics and effectively mitigates signal interference. Second, we employ an enhanced whale optimization algorithm (EWOA) to adaptively optimize the variational mode decomposition (VMD) parameters, ensuring a more precise and data‐driven signal separation. Finally, a Bidirectional GRU network integrated with an Attention mechanism (ATT‐BiGRU) is utilized to dynamically weigh the importance of the decomposed features for superior prediction. Empirical results on the high‐frequency SSE index demonstrate our model's superior performance. Compared to the second‐ranked model, it achieves reductions in MSE, RMSE, MAE, MSLE, MAPE, and SMAPE by 6.23%, 3.17%, 3.22%, 5.87%, 3.16%, and 3.15%, respectively. Notably, our strategy of decomposing both price and volume yields a substantial improvement, reducing key error metrics by over 40% compared to an equivalent non‐decomposed model. Our proposed decomposition‐based hybrid model offers a more precise and robust approach for forecasting of high‐frequency stock prices.

ABSTRACT This study introduces a realized volatility fuzzy time series (RV‐FTS) model that applies a fuzzy c‐means clustering algorithm to estimate time‐varying latent volatility states and their corresponding membership degrees. These memberships are used to construct a fuzzified volatility estimate as a weighted average of cluster centroids. The final volatility forecast is generated through an exponentially weighted moving average (EWMA) mechanism that combines the most recent fuzzified volatility estimate with the previous forecast, governed by the smoothing parameter . The two hyperparameters are estimated using a rolling‐window cross‐validation approach. Our empirical study is based on volatility forecasts for 14 major stock market indices, covering more than 20 years of data. We predict 1‐ to 22‐day‐ahead volatility and compare the RV‐FTS model with nine standard volatility model benchmarks: generalized autoregressive conditional heteroscedasticity (GARCH), ARFIMA, AR, heterogeneous autoregressive (HAR), EWMA, and random forest models, as well as conditional combination forecasts. We find that, in the short‐term, day‐ahead setting, the RV‐FTS model tends to outperform the benchmark models under the mean squared error loss and performs similarly to the best models under the QLIKE loss. The conditional combination forecast shows that across all markets and multiple forecast horizons, there are periods when the weight of the RV‐FTS model in the conditional combination of eight models reaches 50% or more. The volatility timing strategy also shows that the RV‐FTS model leads to higher cost‐ and risk‐adjusted returns compared with a benchmark volatility model.