Detecting insider trading: The theory and validation in Korea Exchange

This paper provides an empirical basis for identifying insider transactions by deriving a theoretical model, which incorporates the relationship between insider transactions and time series of stock returns. Thus, this model enables us detecting insider transactions by applying stock return time series. We show that when there is an insider transaction in the market, time series can be derived as an ARMA(1,1) process having closed solution coefficients. For validation of the model, we test publicly released insider transactions and reverse takeover events using the minute-by-minute stock price data. The selected events show higher pass rate of the detection criteria than the current detection system which shows that our model produces smaller Type II error than the existing post transaction-based cumulative abnormal return model.