Reliably detecting insider trading is a major impediment to both research and regulatory practice. Using account-level transaction data, we propose a novel approach. Specifically, after extracting several key empirical features of typical insider trading cases from existing regulatory actions, we then employ a machine learning methodology to identify suspicious insiders across our full sample. Our identified outliers earn, on average, a significantly higher return relative to a random sample. Further, we find that the trading patterns of selected suspicious insiders exhibit similarities with the changes in a firm’s central decision-makers. We also find that insiders are more likely to use multiple accounts to trade around a major information event; we observe this via the IP address attached to each transaction. Taken together, our approach significantly augments an otherwise elusive ability to detect insider trading.
Note: Research papers posted on SSRN, including any findings, may differ from the final version chosen for publication in academic journals.