This paper develops a novel approach to information-based securities trading by characterizing the hidden state of the market, which varies following a Markov process. Extensive simulation demonstrates that the approach can successfully identify market states and generate dynamic measures of information-based trading that outperform prevailing models. A sample of 120 NYSE stocks further verifies that it can better depict trading dynamics. With this sample, we characterize the features of information asymmetry and belief dispersion around earnings announcements. The sample is also applied to the study of the co-movements of trading activities due to private information or disputable public information.