Abstract
Shortcomings of continuous and static microstructure models are noted with motivation provided by data from the NASDAQ. The influence of volume order imbalance on microstructure dynamics is incorporated into a model which allows the agent to adjust their strategy based on an easily observable quantity. The predictive power of order imbalance allows the agent to decide when they should trade more aggressively to take advantage of beneficial price movements, and when they should trade more conservatively to protect against adverse selection effects. High imbalance results in a stronger inclination to place limit buy orders with the opposite effect on limit sell orders. We attempt to quantify the value of accurately observing imbalance through simulations.
[PDF] Slides of the talk.