Paper: arXiv 2406.19402

Authors: Creighton Heaukulani, Abhinav Pandey, Lancelot F. James

Abstract

Modeling the trading volume curves of financial instruments throughout the day is of key interest in financial trading applications. Predictions of these so-called volume profiles guide trade execution strategies, for example, a common strategy is to trade a desired quantity across many orders in line with the expected volume curve throughout the day so as not to impact the price of the instrument. The volume curves (for each day) are naturally grouped by stock and can be further gathered into higher-level groupings, such as by industry. In order to model such admixtures of volume curves, we introduce a hierarchical Poisson process model for the intensity functions of admixtures of inhomogenous Poisson processes, which represent the trading times of the stock throughout the day. The model is based on the hierarchical Dirichlet process, and an efficient Markov Chain Monte Carlo (MCMC) algorithm is derived following the slice sampling framework for Bayesian nonparametric mixture models. We demonstrate the method on datasets of different stocks from the Trade and Quote repository maintained by Wharton Research Data Services, including the most liquid stock on the NASDAQ stock exchange, Apple, demonstrating the scalability of the approach.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper uses advanced hierarchical Bayesian nonparametrics (Dirichlet process mixtures, slice sampling MCMC) requiring significant mathematical sophistication, but also demonstrates practical application on real financial data (Wharton TAQ dataset) with scalable implementation details.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Model intraday trading volume curves"] --> B{"Data Source<br>Trade and Quote repository<br>Wharton Research Data Services"}
  B --> C["Key Methodology<br>Hierarchical Poisson Process Model"]
  C --> D["Computational Process<br>Bayesian nonparametric mixture<br>via MCMC Slice Sampling"]
  D --> E["Key Finding 1<br>Scalability on liquid stocks<br>e.g., Apple"]
  D --> F["Key Finding 2<br>Effective for trade execution<br>strategies"]

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 introduces a hierarchical Poisson process model for intensity functions of admixtures of inhomogeneous Poisson processes to model daily financial volume curves.

Source sentence
In order to model such admixtures of volume curves, we introduce a hierarchical Poisson process model for the intensity functions of admixtures of inhomogenous Poisson processes, which represent the trading times of the stock throughout the day.

Assumes 3

  • Each day's trading times are modeled as a Poisson point process with a nonnegative, locally integrable intensity function λd(t). model
    Source sentence
    Fix d ≤D. Let λd(t) be a nonnegative, locally integrable function on (0, 1) and model the counting process Id(t) given by Eq. (2) as a Poisson point process on (0, 1) with intensity function λd(t).
  • The total intensity Λd follows a Gamma prior and the normalized density fd(t) is modeled as a Dirichlet‑process mixture of beta densities. statistical
    Source sentence
    let Λd ∼ gamma(aΛ, bΛ) for some aΛ, bΛ > 0, and let fd(t) be a nonparametric mixture of beta density functions. In particular, this mixture model has an infinite number of components, and the mixing weights are given by a Dirichlet process.
  • Stocks can be grouped hierarchically (e.g., by industry or market‑cap) and the model can share mixture components across these groups. other
    Source sentence
    Our hierarchy may be straightforwardly extended to naturally capture broader groupings of the dataset, for example, across different stocks. Furthermore, financial instruments are sensibly grouped at even coarser levels, for example, into small-, medium-, and large-cap stocks, or grouped by the sector of the company.

Data

  • AAPL · Wharton Research Data Services (Trade and Quote repository) order-book public
    Apple Inc. (most liquid NASDAQ stock)
    Source sentence
    including the most liquid stock on the NASDAQ stock exchange, Apple, demonstrating the scalability of the approach.
  • IRBT · Wharton Research Data Services (Trade and Quote repository) order-book public
    IRBT (medium‑cap stock)
    Source sentence
    the medium‑cap stock (IRBT) is about 3.2x as liquid as the small‑cap stock (PZZA)

Recognised sources: WRDS

Costs

Not applicable The paper reports no trading or execution result.

Finds 2

  • The hierarchical Poisson process model is an appropriate choice for daily financial volume trading curves and learns a smooth inhomogeneous Poisson process intensity function that shares components across different days. theoretical
    Source sentence
    The hierarchical Poisson process model is an appropriate choice for daily financial volume trading curves and learns a smooth inhomogenous Poisson process intensity function that shares components across different days.
  • Inference via MCMC is fast even for the largest stock (Apple) using the Dirichlet‑process slice sampling framework. performance
    Source sentence
    Inference via MCMC is fast for even the largest of stocks (like Apple) using the Dirichlet process slice sampling framework.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 11 related papers). Coverage is partial, so this is neither replication nor refutation.

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.