Queue Imbalance as a One-Tick-Ahead Price Predictor in a Limit Order Book

We investigate whether the bid/ask queue imbalance in a limit order book (LOB) provides significant predictive power for the direction of the next mid-price movement. We consider this question both in the context of a simple binary classifier, which seeks to predict the direction of the next mid-price movement, and a probabilistic classifier, which seeks to predict the probability that the next mid-price movement will be upwards. To implement these classifiers, we fit logistic regressions between the queue imbalance and the direction of the subsequent mid-price movement for each of 10 liquid stocks on Nasdaq. In each case, we find a strongly statistically significant relationship between these variables. Compared to a simple null model, which assumes that the direction of mid-price changes is uncorrelated with the queue imbalance, we find that our logistic regression fits provide a considerable improvement in binary and probabilistic classification for large-tick stocks, and provide a moderate improvement in binary and probabilistic classification for small-tick stocks. We also perform local logistic regression fits on the same data, and find that this semi-parametric approach slightly outperform our logistic regression fits, at the expense of being more computationally intensive to implement.