Paper: arXiv 2507.14960
Authors: Ivan Letteri
Abstract
The detection of outliers within cryptocurrency limit order books (LOBs) is of paramount importance for comprehending market dynamics, particularly in highly volatile and nascent regulatory environments. This study conducts a comprehensive comparative analysis of robust statistical methods and advanced machine learning techniques for real-time anomaly identification in cryptocurrency LOBs. Within a unified testing environment, named AITA Order Book Signal (AITA-OBS), we evaluate the efficacy of thirteen diverse models to identify which approaches are most suitable for detecting potentially manipulative trading behaviours. An empirical evaluation, conducted via backtesting on a dataset of 26,204 records from a major exchange, demonstrates that the top-performing model, Empirical Covariance (EC), achieves a 6.70% gain, significantly outperforming a standard Buy-and-Hold benchmark. These findings underscore the effectiveness of outlier-driven strategies and provide insights into the trade-offs between model complexity, trade frequency, and performance. This study contributes to the growing corpus of research on cryptocurrency market microstructure by furnishing a rigorous benchmark of anomaly detection models and highlighting their potential for augmenting algorithmic trading and risk management.
Complexity vs Empirical Score
- Math Complexity: 7.0/10
- Empirical Rigor: 6.0/10 (capped from 6.5: transaction costs only partially modelled; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 6.5)
Research Flowchart
flowchart TD
A["Research Goal:<br>Detect Outliers in BTC LOBs"] --> B["Data Input:<br>26,204 LOB Records"]
B --> C["Methodology:<br>AITA-OBS Framework"]
C --> D["Model Evaluation:<br>13 Statistical & ML Models"]
D --> E["Backtesting:<br>Performance Benchmarking"]
E --> F{"Key Outcomes"}
F --> G["Top Model: Empirical<br>Covariance (EC)"]
F --> H["Result: 6.70% Gain vs.<br>Buy-and-Hold"]
F --> I["Insight: Effective for<br>Risk Management"]
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 unified backtesting environment (AITA-OBS) to evaluate anomaly detection models in Bitcoin limit order books and identifies the Empirical Covariance model as the top-performing approach.
Source sentence
The primary contributions of this study are threefold: (i) the development of a unified and transparent backtesting environment (AITA-OBS) for evaluating anomaly detection models in LOBs; (ii) a rigorous comparative analysis of thirteen statistical and machine learning models for outlier detection; and (iii) an empiric
Assumes 4
- The data follow a multivariate Gaussian distribution for parametric statistical models. model
Source sentence
We assume the data follow a multivariate Gaussian distribution or are based on a robust estimate of the covariance.
- Outlier scores can be converted into trading signals using a mean‑reversion strategy. execution
Source sentence
we provided a transparent and reproducible evaluation of each model’s ability to generate profitable trading signals based on a mean-reversion strategy.
- Feature dimensions are independent for the Histogram‑Based Outlier Score (HBOS) method. statistical
Source sentence
HBOS score for a data point x is the sum of the inverse logarithms of the estimated densities for each of its feature values, assuming feature independence.
- Transaction costs are approximated by a flat market‑maker fee of 0.08% per trade. execution
Source sentence
Figure 6 illustrates the estimated total trading fees for each model, assuming a market maker fee of 0.08% per trade.
Data
Bitcoin limit order books · crypto
- Dataset of 26,204 records from a major exchange order-book unknown
Source sentence
An empirical evaluation, conducted via backtesting on a dataset of 26,204 records from a major exchange, demonstrates that the top-performing model, Empirical Covariance (EC), achieves a 6.70% gain
Costs
Partial
Costs are estimated using a flat market‑maker fee of 0.08% per trade, shown in a separate cost analysis figure.
Source sentence
Figure 6 illustrates the estimated total trading fees for each model, assuming a market maker fee of 0.08% per trade.
Finds 5
total return 6.7%
- Empirical Covariance (EC) achieved the highest cumulative profit of $100.47, a 6.70% gain, outperforming the Buy‑and‑Hold benchmark. performance
Source sentence
Table 2 shows that Empirical Covariance (EC) was the most effective statistical model, achieving a cumulative profit of $100.47 (a 6.70% gain).
- CBLOF was the best machine‑learning model with a profit of $75.48 (5.03% gain) but executed nearly double the number of trades of EC. performance
Source sentence
Among the machine learning models (Table 3), CBLOF yielded the highest cumulative profit at $75.48 (a 5.03% gain), though it executed a high number of trades.
- OC‑SVM delivered a respectable gain of $43.59 (2.91%) with only 140 trades, indicating higher profit‑to‑trade efficiency. performance
Source sentence
OC‑SVM provided a respectable gain of $43.59 (2.91%) with only 140 trades, suggesting a more efficient strategy when considering transaction costs.
- The Buy‑and‑Hold benchmark incurred a loss of $37.06 over the same period. performance
Source sentence
the B&H benchmark, which incurred a loss of -$37.06 over the same period.
- High‑frequency models such as K‑Means, CBLOF, and KNN generate substantial trading fees that would erode profits. performance
Source sentence
The high trade frequency of models like K‑Means, CBLOF, and KNN results in substantial costs that would erode profits.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 40 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
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 (41)
- Manipulation of the Bitcoin market: an agent-based study (2022) — link
- Identifying and treating outliers in finance (2019) — link
- Price Discovery in Cryptocurrency Markets (2019) — link
- Return and volatility spillovers among cryptocurrencies (2018) — link
- Trading Strategy Validation Using Forwardtesting with Deep Neural Networks : (2023) — link
- DNN-ForwardTesting: A New Trading Strategy Validation using Statistical Timeseries Analysis and Deep Neural Networks (2022) — link
- A Stock Trading System for a Medium Volatile Asset using Multi Layer Perceptron (2022) — link
- Stock Market Forecasting Using Machine Learning Models Through Volatility-Driven Trading Strategies (2024) — link
and 33 more
- AITA: A new framework for Trading Forward Testing with an Artificial Intelligence Engine (2023) — link
- VolTS: A Volatility-based Trading System to forecast Stock Markets Trend using Statistics and Machine Learning — → 2307.13422
- A Robust Multivariate Outlier Detection Method for Detection of Securities Fraud : (2020) — link
- LOF : identifying density-based local outliers (2000) — link
- The Elliptical Envelope (2007) — link
- Bitcoins and Bank Runs : Analysis of Market Imperfections and Investor Hysterics (2017) — link
- Spoofing the Limit Order Book: An Agent-Based Model (2017) — link
- Scalable and Interpretable One-Class SVMs with Deep Learning and Random Fourier Features (2019)
- The mathematics of money management: risk analysis techniques for traders (1992)
- Minimum covariance determinant (2010) — link
- Outlier Detection in Axis - Parallel Subspaces of High Dimensional Data (2009) — link
- The Mahalanobis distance (2000) — link
- Intelligent Identification and Order - Sensitive Correction Method of Outliers from Multi - Data Source Based on Historical Data Mining (2022) — link
- Identifying Outliers in Finance (2017) — link
- AITA: A new framework for Trading Forward Testing with an Artificial Intelligence Engine (2023)
- Convergence in cryptocurrency prices? the role of market microstructure (2021) — link
- Nothing but noise? Price discovery across cryptocurrency exchanges (2021) — link
- Dataset Optimization Strategies for MalwareTraffic Detection (2020) — link
- New Optimization Approaches in Malware Traffic Analysis (2021) — link
- Security in the internet of things: botnet detection in software-defined networks by deep learning techniques (2019) — link
- Feature Selection Strategies for HTTP Botnet Traffic Detection (2019) — link
- Logic-based Machine Learning for Transparent Ethical Agents (2020) — link
- Extension of constraint-procedural logic-generated environments for deep Q-learning agent training and benchmarking (2023) — link
- A Logic-based Multi-agent System for Ethical Monitoring and Evaluation of Dialogues (2021) — link
- Care Robots Learning Rules of Ethical Behavior Under the Supervision of an Ethical Teacher (Short paper) (2022) — link
- First Evaluation of an Adaptive Tool Supporting Formative Assessment in Data Science Courses (2023) — link
- Exploring the Impact of LLM-Generated Feedback: Evaluation from Professors and Students in Data Science Courses (2024) — link
- Evaluation of dynamic cointegration-based pairs trading strategy in the cryptocurrency market (2021) — link
- Modelling high-frequency limit order book dynamics with support vector machines (2015) — link
- An automated cryptocurrency trading system based on the detection of unusual price movements with a Time - Series Clustering - Based approach (2022) — link
- Abnormal Transaction Node Detection on Bitcoin (2024) — link
- Analyzing outliers activity from the time-series transaction pattern of bitcoin blockchain (2019) — link
- The informational dimensions of the Amihud (2002) illiquidity measure: Evidence from the M&A market (2019) — link
1 of 41 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.