Order Flows and Limit Order Book Resiliency on the Meso-Scale
Order Flows and Limit Order Book Resiliency on the Meso-Scale
This paper investigates the behavior of limit order books (LOBs) on the meso-scale (minutes), which is crucial for order execution scheduling. Using Nasdaq ITCH data for six large-tick stocks, the authors propose volume-based bucketing to aggregate tick-by-tick data into buckets of fixed executed market volume. This method reduces intra-day seasonality and provides a natural normalization for comparing order flows across periods.
The main empirical findings include an S-shaped nonlinear relationship between trade imbalance (TI) and mid-price change (ΔP), which becomes linear when using net liquidity flow—a weighted average of market and limit order flows. This highlights the importance of limit order flows, which are found to be at least as statistically significant as market order flows for predicting price changes. Additionally, deeper LOB shape metrics (e.g., the slope of the price impact curve) are more relevant than top-level queue depths for quantifying liquidity and price impact. The authors also document a hockey-stick dependence between trade imbalance and one-sided limit order flows, revealing asymmetric effects and scarce liquidity during extreme order imbalances.
The study provides a phenomenological description of LOB resilience and price formation on the meso-scale, offering insights for algorithmic trading and market-making. The results suggest that adaptive order scheduling should incorporate both market and limit order flows, as well as deeper LOB shape information, to better predict price impact and liquidity conditions.
Highlights
- 1Documents an S-shaped nonlinear relationship between trade imbalance and mid-price change, which becomes linear when using a weighted average of market and limit order flows.
- 2Shows that limit order flows are at least as statistically significant as market order flows for predicting price changes on the meso-scale.
- 3Finds that deeper LOB shape metrics (e.g., slope) are more relevant than top-level queue depths for predicting price impact and liquidity.
- 4Identifies asymmetric effects between the active and passive sides of the LOB, including a hockey-stick dependence between trade imbalance and one-sided limit order flows.
- 5Proposes volume-based bucketing as a natural aggregation method for meso-scale analysis, reducing intra-day seasonality and improving statistical properties.
Methods
- MVolume-based bucketing: Aggregating tick-by-tick LOB data into buckets of fixed executed market volume (e.g., 0.25%, 1%, 2% of average daily volume).
- MNon-parametric statistical analysis: Using data-driven tools to explore relationships between order flows, price changes, and LOB metrics without assuming linearity.
- MLinear regression models: Estimating price impact slopes from LOB depth and testing predictive power of various liquidity predictors.
- MTime-series analysis: Examining autocorrelations and cross-correlations of order flow components.
Results
- RThe relationship between trade imbalance (TI) and mid-price change (ΔP) is S-shaped, but becomes linear when using net liquidity flow (a weighted combination of market and limit order flows).
- RLimit order flows (additions and cancellations at the touch) have stronger predictive power for price changes than market order flows alone.
- RDeeper LOB shape metrics (e.g., slope S) are more informative than top-level queue depths for predicting price impact and liquidity on the meso-scale.
- RThere is a hockey-stick dependence: when trade imbalance is extreme, one-sided limit order flows become highly nonlinear, indicating scarce liquidity.
- RVolume-based bucketing reduces intra-day seasonality and yields more stable statistical properties compared to time-based aggregation.
Analyze Paper
Generate insights from "Order Flows and Limit Order Book Resiliency on the Meso-Scal...".