The Hot Potato Effect Within a Single Tick: Evidence from NASDAQ ITCH Order Book Data

This paper documents and analyzes a canonical instance of the hot potato effect occurring within a single price tick in a Central Limit Order Book (CLOB), using high-frequency NASDAQ ITCH order book data.

While hot potato–type inventory handoffs are well known to market practitioners and central to many high-frequency trading strategies, they are difficult to identify unambiguously in real data due to aggregation, anonymization, and the prevalence of common order sizes. The core contribution of this study is the identification of a mechanically generated odd-lot residual (107 shares) that allows the fill–flip sequence to be traced causally at the queue level.

We show how a passive limit order is partially filled as a result of FIFO queue depletion, generating a residual odd-lot position that is immediately neutralized via a market order of the exact same size. The unusual quantity and tight temporal proximity make alternative explanations—such as independent order placement—economically implausible, allowing for high-confidence identification of the same reacting agent.

Building on this empirical sequence, the paper links the observed behavior to a stylized hot potato framework within a single-tick CLOB, using an A–B–C–D agent labeling and drawing on classical dealer inventory models (Ho–Stoll; Glosten–Milgrom). The resulting analysis highlights how rapid role switching between liquidity provision and liquidity taking can generate volume without directional conviction, and why such mechanisms are largely invisible in aggregated market data.

The findings have implications for the empirical study of toxic flow, execution quality, and best execution, and illustrate how practitioner-informed modeling can bridge the gap between microstructural intuition and observable market data.