Paper: arXiv 2308.14235
Authors: Haochen Li, Yi Cao, Maria Polukarov, Carmine Ventre
Abstract
In this study, we introduce a physical model inspired by statistical physics for predicting price volatility and expected returns by leveraging Level 3 order book data. By drawing parallels between orders in the limit order book and particles in a physical system, we establish unique measures for the system’s kinetic energy and momentum as a way to comprehend and evaluate the state of limit order book. Our model goes beyond examining merely the top layers of the order book by introducing the concept of ‘active depth’, a computationally-efficient approach for identifying order book levels that have impact on price dynamics. We empirically demonstrate that our model outperforms the benchmarks of traditional approaches and machine learning algorithm. Our model provides a nuanced comprehension of market microstructure and produces more accurate forecasts on volatility and expected returns. By incorporating principles of statistical physics, this research offers valuable insights on understanding the behaviours of market participants and order book dynamics.
Complexity vs Empirical Score
- Math Complexity: 7.0/10
- Empirical Rigor: 8.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced mathematical concepts from statistical physics (e.g., kinetic energy, momentum analogies) to model order book dynamics, indicating high math complexity. It is heavily empirical, utilizing high-frequency Level 3 order book data, benchmarking against traditional and machine learning methods, and demonstrating predictive accuracy with metrics like R-squared, showing high backtest-ready rigor.
Research Flowchart
flowchart TD
A["Research Goal<br>Predict Volatility & Returns<br>using Order Book Data"] --> B["Core Methodology<br>Physics-Inspired Model"]
B --> C["Key Innovation<br>Active Depth Concept"]
B --> D["Computational Process<br>Calculate Kinetic Energy & Momentum"]
C --> D
D --> E{"Benchmarking"}
E --> F["Traditional Methods"]
E --> G["Machine Learning Algorithms"]
E --> H["Proposed Physics Model"]
F --> I["Key Finding 1<br>Outperforms Benchmarks"]
G --> I
H --> I
H --> J["Key Finding 2<br>Nuanced Microstructure Insight"]
I --> K["Final Outcome<br>Accurate Forecasts &<br>Valuable Market Insights"]
J --> K
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 physics‑based model using Level 3 order‑book data to predict price volatility and expected returns.
Source sentence
In this study, we introduce a physical model inspired by statistical physics for predicting price volatility and expected returns by leveraging Level 3 order book data.
Assumes 1
- The model does not assume market efficiency. model
Source sentence
Unlike the Roll measure, this model does not assume market efficiency, focusing instead on statistically describing the state of the order book and market behaviour.
Data
- Coinbase Level 3 order book data for BTC/USD order-book public
Source sentence
We gathered Level 3 order book data for the BTC/USD and LUNA/USD cryptocurrency pairs.
- Coinbase Level 3 order book data for LUNA/USD order-book public
Source sentence
We gathered Level 3 order book data for the BTC/USD and LUNA/USD cryptocurrency pairs.
Coinbase NYSE TAQ
Costs
Not applicable
Finds 2
- The physics‑based model achieved a prediction accuracy of 60.25% for 1‑second price changes and 78.69% for 10‑second changes on the BTC/USD pair. predictive-accuracy
Source sentence
The prediction accuracy was about 60.12% for 1-second price change and 62.57% for 10-second change. In contrary, the prediction accuracy of the physics-based model is 60.25% for 1-second price change and 78.69% for 10-second on the BTC/USD pair.
- On the LUNA/USD flash‑crash dataset the physics‑based model still reached relatively high accuracies of 58.70% (1‑second) and 69.82% (10‑second). predictive-accuracy
Source sentence
the physics-based model still reaches an relative high accuracy of 58.70% and 69.82%.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 8 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.