Stawarz Marcin
Department of Applied Informatics and Mathematics in Economics, Faculty of Economic Sciences and Management, Nicolaus Copernicus University in Torun, Torun, Poland.
PLoS One. 2025 Jun 24;20(6):e0326809. doi: 10.1371/journal.pone.0326809. eCollection 2025.
This study aims to analyze the development of global stock exchanges by integrating clustering, classification, and Shapley Values to identify growth patterns and understand the differences in market characteristics and dynamics. The research applies the K-means algorithm for clustering, which enables the segmentation of exchanges based on their similarities. This is followed by using the random forest algorithm to classify these clusters and evaluate the importance of various features. Shapley Values are employed to interpret the contribution of individual variables to the model's predictions, considering all possible combinations of features. The empirical analysis is based on data from 82 stock exchanges worldwide, sourced from organizations such as the World Federation of Exchanges and the International Monetary Fund. Key variables used include market capitalization, trading value, the number of listed companies, and share turnover velocity. The results highlight the significant heterogeneity among exchanges, with major markets like those in China and the United States forming distinct clusters due to their size, capitalization, and high trading activity. This distinction underscores their dominant position in the global financial landscape. Moreover, exchanges that have emerged from mergers, such as Euronext and NASDAQ Nordic, demonstrate superior characteristics compared to their peers, indicating that consolidation can be an effective strategy for competing with larger markets and enhancing global competitiveness. The study's findings show that integrating clustering, classification, and Shapley Values is a robust approach for uncovering complex structures within financial markets. This approach provides deeper insights for market participants and policymakers into the growth patterns and strategic positioning of stock exchanges, offering valuable implications for future market development and competition strategies.
本研究旨在通过整合聚类、分类和夏普利值来分析全球证券交易所的发展，以识别增长模式并了解市场特征和动态的差异。该研究应用K均值算法进行聚类，这使得能够根据证券交易所的相似性对其进行细分。接下来使用随机森林算法对这些聚类进行分类，并评估各种特征的重要性。考虑到特征的所有可能组合，采用夏普利值来解释单个变量对模型预测的贡献。实证分析基于来自全球82家证券交易所的数据，这些数据来源于世界证券交易所联合会和国际货币基金组织等机构。使用的关键变量包括市值、交易价值、上市公司数量和换手率。结果凸显了各证券交易所之间存在显著的异质性，中国和美国等主要市场由于其规模、市值和高交易活跃度而形成了不同的聚类。这种差异凸显了它们在全球金融格局中的主导地位。此外，通过合并形成的证券交易所，如泛欧证券交易所和纳斯达克北欧证券交易所，与其同行相比表现出更优越的特征，这表明合并可能是与更大市场竞争并提高全球竞争力的有效策略。该研究结果表明，整合聚类、分类和夏普利值是揭示金融市场复杂结构的一种稳健方法。这种方法为市场参与者和政策制定者提供了关于证券交易所增长模式和战略定位的更深入见解，对未来市场发展和竞争战略具有重要意义。
