Scholarlink

<p>Corporate insiders trade for various reasons, often possessing Material Non-Public Information (MNPI). Accurately determining whether a trade was conducted with MNPI is a challenging task due to its complexity. The work presented here focuses on two critical aspects, first, accurately detecting Unlawful Insider Trading (UIT), and second, identifying the key features explaining accurate classification. It demonstrates how combining Shapley Values (SHAP) and Causal Forest (CF) can effectively reveal these explanatory features. The findings reported herein underscores the importance of causality in effectively identifying, interpreting, and explaining the UIT. It necessitates considering alternative scenarios and their potential outcomes. Within a high-dimensional feature space, the proposed architecture integrates state-of-the-art techniques. The resulting framework exhibits high classification accuracy and provides robust feature rankings through SHAP and causal significance assessments using CF, facilitating the discovery of unique causal relationships. The analysis demonstrates statistically significant relationships between the outcome and several key features, including director status, price-to-book ratio, return, and market beta. These features significantly influence the likelihood of the outcome, suggesting potential links between insider trading behavior and factors such as information asymmetry, valuation risk, market volatility, and stock performance. This analysis draws attention to the complexities inherent in financial causality, suggesting that while initial descriptors may offer intuitive insights, a deeper examination is required to fully understand their nuanced and often uncertain impacts. Nevertheless, these findings reaffirm the architectural flexibility of decision tree models. By incorporating heterogeneity during tree construction, these models effectively uncover latent structures within trade, finance, and governance data, thereby characterizing fraudulent and non-fraudulent behavior while maintaining reliable results.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Detecting and Explaining Unlawful Insider Trading: A Shapley Value and Causal Forest Approach to Identifying Key Drivers and Causal Relationships

  • 期刊论文

  • 发表日期：

    2026年4月15日

查看全文
  • Krishna Neupane,
  • Igor Griva,
  • Robert Axtell,
  • William Kennedy,
  • Jason Kinser

摘要

Corporate insiders trade for various reasons, often possessing Material Non-Public Information (MNPI). Accurately determining whether a trade was conducted with MNPI is a challenging task due to its complexity. The work presented here focuses on two critical aspects, first, accurately detecting Unlawful Insider Trading (UIT), and second, identifying the key features explaining accurate classification. It demonstrates how combining Shapley Values (SHAP) and Causal Forest (CF) can effectively reveal these explanatory features. The findings reported herein underscores the importance of causality in effectively identifying, interpreting, and explaining the UIT. It necessitates considering alternative scenarios and their potential outcomes. Within a high-dimensional feature space, the proposed architecture integrates state-of-the-art techniques. The resulting framework exhibits high classification accuracy and provides robust feature rankings through SHAP and causal significance assessments using CF, facilitating the discovery of unique causal relationships. The analysis demonstrates statistically significant relationships between the outcome and several key features, including director status, price-to-book ratio, return, and market beta. These features significantly influence the likelihood of the outcome, suggesting potential links between insider trading behavior and factors such as information asymmetry, valuation risk, market volatility, and stock performance. This analysis draws attention to the complexities inherent in financial causality, suggesting that while initial descriptors may offer intuitive insights, a deeper examination is required to fully understand their nuanced and often uncertain impacts. Nevertheless, these findings reaffirm the architectural flexibility of decision tree models. By incorporating heterogeneity during tree construction, these models effectively uncover latent structures within trade, finance, and governance data, thereby characterizing fraudulent and non-fraudulent behavior while maintaining reliable results.