Paper: arXiv 2407.16527

Authors: Timothy DeLise

Abstract

Market making refers to a form of trading in financial markets characterized by passive orders which add liquidity to limit order books. Market makers are important for the proper functioning of financial markets worldwide. Given the importance, financial mathematics has endeavored to derive optimal strategies for placing limit orders in this context. This paper identifies a key discrepancy between popular model assumptions and the realities of real markets, specifically regarding the dynamics around limit order fills. Traditionally, market making models rely on an assumption of low-cost random fills, when in reality we observe a high-cost non-random fill behavior. Namely, limit order fills are caused by and coincide with adverse price movements, which create a drag on the market maker’s profit and loss. We refer to this phenomenon as “the negative drift” associated with limit order fills. We describe a discrete market model and prove theoretically that the negative drift exists. We also provide a detailed empirical simulation using one of the most traded financial instruments in the world, the 10 Year US Treasury Bond futures, which also confirms its existence. To our knowledge, this is the first paper to describe and prove this phenomenon in such detail.

Complexity vs Empirical Score

  • Math Complexity: 7.0/10
  • Empirical Rigor: 4.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper presents a theoretical proof using a discrete market model and references stochastic processes (SDEs, Poisson rates), indicating advanced math, but the empirical component is a single-asset simulation rather than a multi-asset backtest or implementation-heavy analysis.

Research Flowchart

  flowchart TD
  A["Research Goal: Investigate discrepancy between idealized models<br>and actual limit order fill dynamics"] --> B["Method: Discrete Market Model"]
  A --> C["Method: Empirical Simulation"]
  
  B --> D["Theoretical Proof: Demonstrated existence<br>of adverse price movement upon fill"]
  
  C --> E["Data: 10 Year US Treasury Bond Futures<br>(One of world's most traded instruments)"]
  E --> F["Simulation Process: Modeled real market execution<br>with accurate fill timing"]
  
  D --> G["Key Finding: Confirmed 'Negative Drift'<br>Limit orders fill during adverse price movements"]
  F --> G
  
  G --> H["Outcome: Models must account for<br>high-cost non-random fill behavior<br>vs. low-cost random fill assumption"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes a discrete market model that incorporates a consistent negative drift associated with limit order fills and validates it with empirical simulation on 10‑Year US Treasury futures.

Source sentence
We describe a discrete market model and prove theoretically that the negative drift exists.

Assumes 6

  • Prices move in discrete full tick intervals. model
    Source sentence
    prices move in discrete full tick intervals
  • There is a consistent future negative drift in the mid‑price movement with respect to every limit order fill. model
    Source sentence
    there is a consistent future negative drift in the mid price movement with respect to every limit order fill
  • The fill rate of a limit order in the face of an adverse price movement is assumed to be 100% in the discrete model. model
    Source sentence
    The fill rate of a limit order in the face of an adverse price movement is assumed to be 100% in our discrete model.
  • Rf is the Bernoulli rate at which orders are filled when the price movement is not adverse. model
    Source sentence
    And Rf is assumed to be the rate at which the orders are filled in the cases when the price movement was not adverse.
Show 2 more
  • Market order slippage cost equals half a tick (½ the bid‑ask spread). execution
    Source sentence
    the slippage cost involved with a market order is equal to 1/2 the distance between the best bid and ask, which is usually equal to 1/2 tick
  • Raw tick data are used without down‑sampling for simulation. data
    Source sentence
    we do not down sample the data at all, but use the raw tick data, event by event

Data

Backtest sample: 10-Year US Treasury Bond futures · futures · intraday

  • 10 Year US Treasury Bond futures (TY) · Trading Technologies, Inc order-book proprietary
    8AM - 5PM EST, 11/30/2023 · TY Mar24 contract
    Source sentence
    Contract TY Mar24 Time Period 8AM - 5PM EST, 11/30/2023

Costs

Partial Costs appear only as a robustness check, a flat fee without spread or impact, or outside the headline numbers.

Market orders incur a slippage of ½ tick; limit order fills incur a negative drift cost.

Source sentence
the slippage cost involved with a market order is equal to 1/2 the distance between the best bid and ask, which is usually equal to 1/2 tick

Finds 5

  • The average mid‑price drift after a limit order fill is approximately –0.0065 ticks. negative
    Source sentence
    On top of this we have computed the average drift in each case, which comes out to approximately −0.0065.
  • Buy order fills are followed by significant downward price drift; sell order fills are followed by significant upward drift. statistical
    Source sentence
    There is a significant downward price drift following buy order fills and an equally significant upward drift following sell order fills.
  • The empirical fill probability is much lower than the assumed 100% in many prior models. statistical
    Source sentence
    We will show later that we find this probability to be much lower than 1 in practice.
  • The discrete market model theoretically proves that limit order fills are on average accompanied by a price move opposite to the order direction. theoretical
    Source sentence
    We describe a discrete market model and prove theoretically that the negative drift exists.
  • The simulated back‑test aims to match the live TT simulator performance as closely as possible. performance
    Source sentence
    The goal is for our simulated back test to match the live TT simulator as well as possible.

Checked by others

Supported by

  • Market Simulation under Adverse Selection (2024)
    On the question: How does the assumed fill probability impact observed strategy performance?
    Compare the findings

    This paper: The empirical fill probability is much lower than the assumed 100% in many prior models.

    That paper: Excluding adverse fills and using unrealistic non‑adverse fill probabilities can significantly inflate strategy performance.

    Both note that using unrealistic 100% fill rates (A) inflates performance; B shows excluding adverse fills (non‑adverse probabilities) leads to over‑optimistic results, confirming A's lower empirical fill probability.

Cited by 5 papers in the archive (5 by independent authors): The Market Maker's Dilemma: Navigating the Fill Probability … (discusses its findings); Event-Based Limit Order Book Simulation under a Neural … (discusses its findings)

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (10)

  1. Algorithmic and High-Frequency Trading (2015) — link
  2. High Frequency Trading in a Limit Order Book (2008) — link
  3. The fractal geometry of nature (1983)
  4. General Compound Hawkes Processes in Limit Order Books (2018) — link
  5. General Compound Hawkes Processes for Mid-Price Prediction (2021) — link
  6. DeepLOB: Deep Convolutional Neural Networks for Limit Order Books (2019) — link
  7. Algorithmic Trading, Stochastic Control, and Mutually Exciting Processes (2018) — link
  8. Limit Order Books (2013) — link
and 2 more
  1. Point Spectra of Some Mutually Exciting Point Processes (1971)
  2. Hawkes model specification for limit order books (2020) — link

0 of 10 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.