Here we discuss the model of Glosten and Milgrom (1985), which captures the notion of asymmetric information, explicitly characterizing the bid–ask spread. The model assumes risk neutrality, asymmetric information and a quote-driven protocol. Within this market protocol, dealers post bid and ask quotes that are subsequently chosen by their customers. The latter are asymmetrically informed and arrive at the market-place sequentially. The assumption of asymmetric information generates adverse selection costs, which oblige dealers to quote different prices for buying and selling, i.e. to open the bid–ask spread. The spread is the premium that dealers demand for trading with agents with superior information.

In Glosten and Milgrom (1985) it is assumed that customers' trades provide information about the future value of the risky asset. However, the model constrains trades to a size of one unit, so only the direction of orders (buy or sell) provides information to dealers. This hypothesis would subsequently be generalized in the model of Easley and O'Hara (1987), which embodies the signalling role of trade sizes. In this framework, market-makers set prices conditional on all the information they can extract from the order flow. Since, by observing the insiders' orders, liquidity providers can infer information on the future value, both models set out a process of price discovery and provide insights into the process of adjustment of quoted prices.