Queue Imbalance as a One-Tick-Ahead Price Predictor in a Limit Order Book (1512.03492v1)

Abstract: 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.

• The paper demonstrates that queue imbalance enhances one-tick-ahead price prediction using both binary and probabilistic logistic regression models.
• The study reveals a 50–60% improvement in binary classification for large-tick stocks, emphasizing the impact of market microstructure on prediction accuracy.
• Employing local logistic regression captures non-linearities in order book dynamics, suggesting practical applications for high-frequency trading strategies.

An Analysis of Queue Imbalance for Predicting One-Tick-Ahead Price Movements in a Limit Order Book

This paper by Gould and Bonart provides a comprehensive empirical paper of whether queue imbalance in a Limit Order Book (LOB) can significantly predict the direction of the subsequent mid-price movement. The focus is primarily on both binary and probabilistic classifiers, which aim to project whether the next price move will be upwards or downwards, and the probability that it will be upwards, respectively.

Methodological Approach

The authors utilize data on ten liquid stocks from Nasdaq over a one-year period to assess the predictive power of queue imbalance, a metric reflecting the imbalance between bid and ask queue sizes. They employ logistic regression as the primary analytical tool. One of the strengths of employing logistic regression here is its ability to cohesively handle binary outcomes linked to imbalanced data. The model's output is then benchmarked against a null model that assumes no relationship between queue imbalance and price direction, thereby serving as a robustness check for the observed predictability.

Key Findings

For large-tick stocks, the findings are robust: imbalance sharply correlates with out-of-sample predictive improvements for predicting future price movements, outperforming the null model significantly. Specifically, binary classification improves by about 50–60%, and probabilistic classification by 20–30%. Small-tick stocks demonstrate moderate predictive improvements, specifically around 10–30% in binary classification. These quantitative insights underline the nuanced impact of market microstructure on price predictability.

The authors further substantiate their findings by employing local logistic regression, thereby investigating more intricate structures potentially underlying the observed relationships and capturing non-linearities obscured by simpler parametric models. Although slightly more computationally intensive, these semi-parametric models slightly outperform their standard counterparts, highlighting subtle market behaviors possibly related to microstructural factors like queue formation at different price levels.

Theoretical and Practical Implications

One of the central contributions is the potential application of queue imbalance in crafting predictive algorithms for high-frequency trading strategies. Considering the focus on improving prediction efficiencies, even marginal gains in accuracy can yield substantial economic benefits considering market dynamics. The ability to predict price movements based on simple derivatives of LOB activity has pivotal importance in executing optimal trading strategies and algorithmic decision-making in real-time.

From a theoretical perspective, the paper provides a refined empirical domain for testing existing models and suggests avenues for the enhancement of microstructural models by including queue dynamics as a core component. Moreover, the paper opens a dialogue on nuanced behaviors between various stock types (large-tick vs. small-tick) stressing the necessity to consider tick size and LOB states' specific configurations during model formulation.

Conclusion and Future Directions

The insights offered by Gould and Bonart draw attention to the multifaceted dynamics within LOBs that influence price predictability. Future research could enlighten how variations in imbalance at deeper LOB levels or alternative imbalance metrics could further enhance prediction accuracy. Given this foundational work, investigations into longer time frames and broader sets of stocks might uncover generalizable principles in price formation across varying market conditions. Moreover, integrating this model with machine learning techniques could push the boundaries further in predictive analytics within market microstructures.