The Joint Estimation of Probability of Informed Trading: The Informational Role of Stock and Option Volume

This study develops a new methodology for jointly estimating the probability of informed trading from stock and option market, by extending the Easley, O’Hara, and Srinivas (1998) models and incorporating the trade size effect into the volume information. We introduce a powerful estimation technique, the EM-algorithm, to deal with the computational problems for the recently sample, in particular the heavily traded stocks. In calendar-time, we estimate the new measure probability of informed trading, SOVPIN, using data for individual NYSE-listed stocks and individual CBOE-listed stock options for 1996 to 2003. We find that SOVPIN has great predictive power of future stock returns and the risk premium persists regardless of the holding periods of portfolios. Further, in event-time before earnings announcements, we find that firms with good news tend to have relatively higher SOVPIN and higher informed put option trading volume, while firms with good news tend to have relatively lower SOVPIN higher informed call option trading volume. Our evidence supports the risk-based hypothesis for PEAD and reconciles with th