For an auctioneer it is of utmost importance to design an auction mechanism that gives robust price signals which in turn increases auction performance. Information architecture and forward trading platforms are the two main information sources that could generate these price signals. However the traditional presumption that agents form rational expectations by accurately processing all available information in the online trading environment and forming their expectations accordingly has found mixed support. We develop a research model that empirically tests the impact of agent's attitudes on their price expectation through their trading behaviour. Using a unique data set, we tested our hypotheses on real ex ante forecasts, evaluated ex post, in an electricity day ahead auction context. This paper is one of the first to take an information-based view to study the trading behaviour of agents and their price expectations; with results that suggest a re-consideration of some of the conventional concepts.