Abstract: In order-driven financial market, the submission and cancellation of market orders and limit orders are often found to be showing non-stationary and agglomerative characteristics.Quantitatively evaluating the exciting relationship between these limit order book events is an essential topic of market microstructure researching, such as the dynamic modeling of the limit order book and the behavioral characteristics of traders.Using all the orders of sample stocks ubmitted by traders in Shenzhen Stock Exchange, high frequency dynamic limit order book is rebuilt and state-dependent Hawkes process model is construoted to analyze the self-excitation and co-excitation effects among various limit order book events.Specifically, two sets of states are built, using bid-ask spread to measure instant market liquidity and queue imbalance to measure buy or sell pressure; and at the same time,2 sets of limit order book events according to the type and the aggressiveness of all the orders. In total, our model is tested under 3 different settings. Empirical results show that, in China’s order-driven stock market,self-excitation effects of limit book events is much stronger than co-excitation effect, and these effects vary significantly under different market states. Classifying market events according to their aggressiveness is much better in reflecting the information feedback loops and trading behavior of traders.

Key words: limit order book; state-dependent Hawkes processes; limit order; excitation effect; market state