We develop a new likelihood-based approach to sign trades in the absence of quotes. It is equally efficient as existing MCMC methods, but more than 10 times faster. It can deal with the occurrence of multiple trades at the same time, and noisily observed trade times. We apply this method to a...