Quality over Quantity: Semi-Supervised Detection of Illicit Bitcoin Flows via Feature Engineering

Yekaterina Smolenkova; Nickolay Larionov; Nikolay Ivanov; Yury Yanovich

cs.LGAug 22, 2026

Quality over Quantity: Semi-Supervised Detection of Illicit Bitcoin Flows via Feature Engineering

Authors: Yekaterina Smolenkova, Nickolay Larionov, Nikolay Ivanov, Yury Yanovich

Organizations: Skolkovo Institute of Science and Technology, Moscow, Russia · Moscow Institute of Physics and Technology, Moscow, Russia

Abstract

Detecting illicit cryptocurrency transactions is hampered by extreme class imbalance, adversarial obfuscation, and a scarcity of reliable labels. While semi-supervised learning (SSL) offers a promising solution by leveraging unlabeled data, we show that its success is not guaranteed by data volume alone but is contingent on data quality. We introduce an SSL framework for detecting illicit Bitcoin flows in Shared Send Mixers (SSM) transactions, built on a comprehensive historical dataset comprising 163 million transactions. Our main conclusion is that the success of SSL depends on data quality rather than volume: high-fidelity features such as KeyLinker address clustering and Shared Send Untangling (SSU) complexity metrics achieve an F1 score of 0.84 on unlabeled data. Finally, we empirically show that common heuristics like One-Time Change (OTC), though abundant, introduce noise, while strategic reliance on higher-fidelity features like KeyLinker is essential. Our work establishes that in blockchain forensics, the path to better performance lies in smarter feature engineering for data quality, not just larger datasets.

Figures & tables

Figure 1: Bitcoin UTXO transaction model: each transaction consumes previous outputs and creates new ones.

Figure 2: Example of address clustering using CS heuristic. 1st address and 2nd address are owned by a single wallet

Figure 3: Example of address clustering using OTC heuristic.

Transactions		Addresses
Total	1,150.9M	Total	1,370.1M
With any label	161.2M	With any label	39.0M
CoinJoin	163.4M
Labeled CJ	4.6M
Clusterized transactions		Clusterizes addresses
OTC	188.9M	KeyLinker	131.4K

Table 1: Comprehensive dataset statistics (up to block 882,421).

Model	Hyperparameters
CatBoost	learning_rate, l2_leaf_reg, iterations,
	depth, border_count
XGBoost	subsample, n_estimators, max_depth,
	learning_rate, gamma
RandomForest	n_estimators, max_depth,
	min_samples_leaf

Table 2: Hyperparameter search spaces

Model	Features					Metrics
	DEFAULT	REUSE	CS	OTC	SSU	Precision	Recall	F1-score	ROC AUC
CatBoost	✓					0.929	0.689	0.791	0.958
	✓	✓				0.929	0.730	0.818	0.966
	✓	✓	✓			0.930	0.740	0.824	0.969
	✓	✓	✓	✓		0.928	0.740	0.823	0.967
	✓				✓	0.926	0.705	0.800	0.960

Table 3: Metrics by feature set for all models

Model	Features					Metrics
	DEFAULT	REUSE	CS	OTC	SSU	Precision	Recall	F1-score	ROC AUC
CatBoost	✓					0.848	0.759	0.801	0.956
	✓	✓				0.873	0.775	0.821	0.964
	✓	✓	✓			0.866	0.795	0.829	0.966
	✓	✓	✓	✓		0.866	0.788	0.825	0.964
	✓				✓	0.856	0.764	0.807	0.958

Table 4: Metrics by feature set on pseudo stage

Model	$\tau^{+}$	Pseudo-labeled pos	$\tau^{-}$	Pseudo-labeled neg
CatBoost	0.904	1,360,746	0.006	8,600,795
XGBoost	0.982	1,364,991	0.022	8,627,624
RandomForest	0.933	774,035	0.015	5,203,072

Table 5: Pseudo-labeling thresholds and the number of added pseudo-labeled samples for the default+reuse+cs+ssu setting.

Financial Fraud Detection

Financial fraud detection identifies suspicious transactions, accounts, claims, and behavior to prevent losses and guide investigation.

See all 139 papers

Extracting Money Laundering Transactions from Quasi-Temporal Graph Representation
Apr 3, 2026Haseeb Tariq, Marwan HassaniAnti-Money Laundering Detection Anti-Money Laundering
ING Bank, Amsterdam, The Netherlands · Eindhoven University of Technology, Eindhoven, The Netherlands
Large Language Models for Cryptocurrency Transaction Analysis: A Bitcoin Case Study
Jan 30, 2025Yuchen Lei, Yuexin Xiang, Rafael Dowsley +5LLM Evaluation LLMs for Cybersecurity
School of Cyber Science and Engineering, Wuhan University, Wuhan 430072, China · Faculty of Information Technology, Monash University, Clayton, VIC 3800, Australia · CSIRO’s Data61, Eveleigh, NSW 2015, Australia +2
USE: Uncertainty Structure Estimation for Robust Semi-Supervised Learning
Feb 28, 2026Tsao-Lun Chen, Chien-Liang Liu, Tzu-Ming Harry Hsu +4Semi-Supervised Learning
National Taiwan University of Science and Technology, Taipei, Taiwan · National Taiwan University, Taipei, Taiwan · Chang Gung University, Taoyuan, Taiwan +3

Apr 3, 2026J/K move · Enter open · S save

Extracting Money Laundering Transactions from Quasi-Temporal Graph Representation

Haseeb Tariq, Marwan Hassani

ING Bank, Amsterdam, The Netherlands · Eindhoven University of Technology, Eindhoven, The Netherlands

Anti-Money Laundering Detection Anti-Money Laundering Financial Fraud Detection

Open paper AskPDF arXiv

Money laundering presents a persistent challenge for financial institutions worldwide, while criminal organizations constantly evolve their tactics to bypass detection systems. Traditional anti-money laundering approaches mainly rely on predefined risk-based rules, leading to resource-intensive investigations and high numbers of false positive alerts. In order to restrict operational costs from exploding, while billions of transactions are being processed every day, financial institutions are investing in more sophisticated mechanisms to improve existing systems. In this paper, we present ExSTraQt (EXtract Suspicious TRAnsactions from Quasi-Temporal graph representation), an advanced supervised learning approach to detect money laundering (or suspicious) transactions in financial datasets. Our proposed framework excels in performance, when compared to the state-of-the-art AML (Anti Money Laundering) detection models. The key strengths of our framework are sheer simplicity, in terms of design and number of parameters; and scalability, in terms of the computing and memory requirements. We evaluated our framework on transaction-level detection accuracy using a real dataset; and a set of synthetic financial transaction datasets. We consistently achieve an uplift in the F1 score for most datasets, up to 1% for the real dataset; and more than 8% for one of the synthetic datasets. We also claim that our framework could seamlessly complement existing AML detection systems in banks. Our code and datasets are available at https://github.com/mhaseebtariq/exstraqt.

Abstract

Financial Fraud Detection

Explore similar work

Extracting Money Laundering Transactions from Quasi-Temporal Graph Representation

Large Language Models for Cryptocurrency Transaction Analysis: A Bitcoin Case Study

USE: Uncertainty Structure Estimation for Robust Semi-Supervised Learning