Paper: arXiv 2411.13594

Authors: Christian D. Blakely

Abstract

We propose an error-correcting model for the microprice, a high-frequency estimator of future prices given higher order information of imbalances in the orderbook. The model takes into account a current microprice estimate given the spread and best bid to ask imbalance, and adjusts the microprice based on recent dynamics of higher price rank imbalances. We introduce a computationally fast estimator using a recently proposed hyperdimensional vector Tsetlin machine framework and demonstrate empirically that this estimator can provide a robust estimate of future prices in the orderbook.

Complexity vs Empirical Score

  • Math Complexity: 8.0/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper presents advanced mathematical structures including recursive definitions of higher-order adjustments, Markov chain transition matrices, and a complex theoretical framework for microprice convergence, resulting in a high math complexity score. It provides empirical validation using high-frequency level 3 orderbook data from a real provider (Databento) and discusses practical implementation speed, which supports a high empirical rigor score.

Research Flowchart

  flowchart TD
  A["Research Goal: High-Fres Microprice<br>Estimation from Limit Orderbook"] --> B["Input Data:<br>Limit Orderbook Dynamics"]
  B --> C["Methodology:<br>Hyperdimensional Vector<br>Tsetlin Machine"]
  C --> D["Computational Process:<br>Error-Correcting Model<br>Adjusts Microprice based on<br>High-Rank Orderbook Imbalances"]
  D --> E["Outcome:<br>Fast & Robust<br>High-Frequency<br>Price Prediction"]

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 an error‑correcting model for the microprice that uses higher‑order order‑book imbalance information and a hyperdimensional vector Tsetlin machine for fast estimation.

Source sentence
We propose an error-correcting model for the microprice, a high-frequency estimator of future prices given higher order information of imbalances in the orderbook.

Assumes 3

  • The features extracted from higher‑order depth must be computed fast enough to remain relevant for the next order‑book event. execution
    Source sentence
    these features need to be extracted and evaluated in a fast enough manner to be still relevant during the next event of the orderbook.
  • A fully trained Tsetlin machine is available at prediction time to adjust the microprice. model
    Source sentence
    we assume a fully trained TM in order make the predictions in how to adjust the microprice.
  • Level‑3 order‑book data from Databento accurately represents the market microstructure for the equities studied. data
    Source sentence
    using orderbook data from level 3 orderbook data provided by Databento

Data

  • Level‑3 order‑book data · Databento order-book commercial
    TEM (small‑cap) and TSLA (blue‑chip) equities
    Source sentence
    Data is provided by Databento.

Costs

Not applicable The paper reports no trading or execution result.

Finds 3

  • The error‑correcting estimator provides a more robust forecast of future prices compared with the baseline microprice. predictive-accuracy
    Source sentence
    demonstrate empirically that this estimator can provide a robust estimate of future prices in the orderbook.
  • Adjusted microprice shows slightly better accuracy when spreads are tighter and imbalance is near zero, especially for blue‑chip equities like TSLA. performance
    Source sentence
    The empirical results so far have demonstrated slightly better accuracy at measuring a microprice adjustment when the spreads are tighter and there the imbalance is centered around zero which is the case often with blue chip equities such as TSLA.
  • When spreads are tight, the adjusted microprice performs similarly to the original microprice during later trading sessions. performance
    Source sentence
    We also see that the adjusted microprice tends to perform the same as the microprice when spreads are more tight as trading continues during the afternoon trading session as volatility dies down.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 14 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.

Cited References (44)

  1. Self-Organizing Feature Maps Identify Proteins Critical to Learning in a Mouse Model of Down Syndrome (2015) — link
  2. Exploring Effects of Hyperdimensional Vectors for Tsetlin Machines (2024) — link
  3. Coalesced Multi-Output Tsetlin Machines with Clause Sharing (2021) — link
  4. HDC-MiniROCKET: Explicit Time Encoding in Time Series Classification with Hyperdimensional Computing (2022) — link
  5. Recursive Binding for Similarity-Preserving Hypervector Representations of Sequences (2022) — link
  6. Extending the Tsetlin Machine With Integer-Weighted Clauses for Increased Interpretability (2020) — link
  7. Theory of games and economic behavior (commemorative edition) (2007)
  8. Learning Automata: An Introduction (1989)
and 36 more
  1. On behaviour of finite automata in random medium (1961)
  2. The Weighted Tsetlin Machine: Compressed Representations with Weighted Clauses (2019) — link
  3. The Tsetlin Machine - A Game Theoretic Bandit Driven Approach to Optimal Pattern Recognition with Propositional Logic (2018) — link
  4. Adaptive Machine Learning for Credit Card Fraud Detection (2015)
  5. A Tsetlin Machine Approach for Learning and Prediction in Multivariate Time Series (2020)
  6. A Scheme for Continuous Input to the Tsetlin Machine with Applications to Forecasting Disease Outbreaks (2019) — link
  7. A Novel Multi-Step Finite-State Automaton for Arbitrarily Deterministic Tsetlin Machine Learning (2020) — link
  8. Increasing the Inference and Learning Speed of Tsetlin Machines with Clause Indexing (2020) — link
  9. A Tsetlin Machine with Multigranular Clauses (2019) — link
  10. The Convolutional Tsetlin Machine (2019) — link
  11. Sparse spatial autoregressions (1997)
  12. A two-armed bandit collective for hierarchical examplar based mining of frequent itemsets with applications to intrusion detection (2014)
  13. Learning Automata-based Solutions to the Nonlinear Fractional Knapsack Problem with Applications to Optimal Resource Allocation (2007)
  14. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead (2019)
  15. Stochastic searching on the line and its applications to parameter learning in nonlinear optimization (1997)
  16. A Tsetlin Machine with Multigranular Clauses (2019)
  17. Explainability and Dependability Analysis of Learning Automata based AI Hardware (2020)
  18. Increasing the Inference and Learning Speed of Tsetlin Machines with Clause Indexing (2020)
  19. Learning Automata based Energy-efficient AI Hardware Design for IoT (2020)
  20. The Regression Tsetlin Machine - A Novel Approach to Interpretable Non-Linear Regression (2019)
  21. Using the Tsetlin Machine to Learn Human-Interpretable Rules for High-Accuracy Text Categorization with Medical Applications (2019) — link
  22. Hyperdimensional Vector Tsetlin Machines with Applications to Sequence Learning and Generation (2024) — link
  23. Accelerating Hyperdimensional Computing on FPGAs by Exploiting Computational Reuse (2020) — link
  24. Neural Additive Models: Interpretable Machine Learning with Neural Nets (2020) — link
  25. The Micro-Price: A High Frequency Estimator of Future Prices (DOI: 10.1080/14697688.2018.1489139)
  26. Using SHAP values for XGboost interpretability (2018 (accessed May 20, 2020)) — link
  27. Five practical steps to make Artificial Intelligence (AI) interpretable (2019 (accessed July 27, 2020)) — link
  28. COMPAS Analysis (2016 (accessed May 20, 2020)) — link
  29. A Unified Approach to Interpreting Model Predictions (2017) — link
  30. The UCR Time Series Classification Archive (2018)
  31. Contrastive Principal Component Analysis (2017)
  32. The Sparse Tsetlin Machine: Sparse Representation with Active Literals (2024) — link
  33. A Survey on Hyperdimensional Computing aka Vector Symbolic Architectures, Part I: Models and Data Transformations (2022) — link
  34. ``Why Should I Trust You?’’: Explaining the Predictions of Any Classifier (2016) — link
  35. Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methods (2019) — link
  36. An FPGA Architecture for Online Learning using the Tsetlin Machine (2023) — link

0 of 44 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.