This article develops a model in which patterns in buy and sell volume, order imbalances, and expected price changes arise endogenously. The model covers cases in which the market maker is competitive and is a monopolist. Our results provide an explanation for the existence of patterns in mean returns within the trading day and across trading days. Over the past decade studies of stock-price transaction data have revealed a number of systematic patterns in the volume of trading, the expectation and variability of returns, and the size of bid-ask spreads, both within the trading day and across trading days. The regularity and pronounced nature of these patterns is surprising, and in most cases the patterns were not predicted by any theory. In this article we consider patterns in mean returns and show how these can arise as a consequence of the market-making process and the presence of privately informed traders. The nonconstancy of measured expected returns on stocks is well documented.