Two trading networks based on the transaction data of a stock and its warrant in Chinese financial market have been constructed to analyze the market information implied by microstructure of trading network and the influence of diversity of trading behavior on immediate price. The trading networks contain nearly 1 million traders. At a given significance level, we eliminate the noise traders, construct the "core" trading network, and find the noise traders have greater impact on immediate price. By analyzing the microstructure of trading network and identifying 30 network positions in the triadic motifs, the diversity of trading behavior is calculated by the frequency at which the trader appears at the network position. The traders with larger diversity of trading behavior have a more sensitive impact on immediate price. Microstructure of trading network provides a new perspective for studying trader behavior and price behavior.