A Hybrid, Transparent Trust and Risk Assessment Framework for Cryptocurrency Exchanges

Authors

  • Bongani Mawhayi University of the Western Cape
  • Johnny Botha
  • Louise Leenen University of the Western Cape

DOI:

https://doi.org/10.34190/eccws.25.1.4840

Keywords:

Cryptocurrency exchanges, Trust assessment, Risk scoring, Sentiment analysis

Abstract

Cryptocurrency exchanges act as critical intermediaries within the digital asset ecosystem, yet users currently rely on largely opaque, platform-defined trust scores to assess their reliability and risk. Existing industry frameworks, notably those produced by CoinGecko and CoinMarketCap, provide useful signals related to liquidity and volume integrity but suffer from limited transparency, fixed weighting schemes, and the absence of sentiment-based assessment. This paper presents HTREx (Hybrid Trust and Risk Evaluation framework for exchanges), a semi-automated, modular trust and risk assessment framework for cryptocurrency exchanges that addresses these limitations. The framework integrates five dimensions of exchange integrity: user sentiment, regulatory compliance, technical security, transparency, and incident history. Sentiment is quantified using transformer-based natural language processing applied to user-generated content from mobile application reviews and online forums. Compliance is assessed through structured extraction of regulatory and operational disclosures from Terms of Service documents using large language models. Security, transparency, and incident history are evaluated through a combination of publicly verifiable indicators, third-party assessments, and a recency-weighted incident scoring model. All components are normalised and aggregated into a composite score using user-adjustable weights, enabling personalised risk prioritisation while retaining a defensible default configuration for comparative analysis. The framework is demonstrated using four prominent exchanges—Kraken, Coinbase, Binance, and Uniswap—highlighting clear differences between centralised and decentralised platforms and illustrating how sentiment, compliance, and historical incidents materially influence overall trust assessments. The results suggest that transparent, extensible, and user-configurable scoring models can provide a more interpretable and context-sensitive evaluation of exchange risk than existing monolithic trust scores, with direct relevance for both retail and institutional participants.

Author Biographies

Bongani Mawhayi, University of the Western Cape

Bongani Mawhayi is pursuing a Master's degree in Computer Science at the University of the Western Cape. His research builds upon his honours project, which delved into investigating crypto scams, particularly examining the practice of web scraping wallet addresses from social media platforms. In addition to his academic pursuits, Bongani is an active participant in Capture The Flag (CTF)-style challenges, sharpening his problem-solving skills. He is also a proud Mastercard Foundation Scholar.

Johnny Botha

Johnny Botha is currently working on a PhD in Computer Science at the University of Western Cape. His topic is “A process for Blockchain Crime Investigations Based on Open-Source Intelligence”. He is a principal researcher and project manager at the CSIR. Botha holds a Master’s degree in Information Technology from the University of South Africa (UNISA), an NDip and BTech degree in Computer Systems Engineering from the Tswane University of Technology (TUT)

Louise Leenen, University of the Western Cape

Louise Leenen is an Associate Professor in the Computer Science Department at the University of the Western Cape (UWC) in South Africa. Louise completed her PhD in Artificial Intelligence (AI) at the University of Wollongong in Australia. Her research focus areas are AI applications in cybersecurity, and problem formulation and modelling. She joined UWC in 2019 after working as a Principal Researcher at the CSIR in the Cyber Defence Research Group. Leenen is a member of the Centre for Artificial Intelligence (CAIR) and leads the research group on AI and Cybersecurity. She regularly serves in various positions on committees of national and international bodies, such as the NRF, BRICS, and IFIP. She has authored or co-authored several journal publications, book chapters, and conference papers.

Downloads

Published

2026-06-15