Estimating Insider Involvement in Cryptocurrency Exchange Thefts

This working paper introduces the Erez Tall's Insider Attribution Model (EIAM), an exploratory framework for estimating deliberate insider involvement following reported cryptocurrency exchange thefts. The framework combines a reference probability with adjustments for accessible funds, external-security controls, and safeguards against insider abuse, followed by a conditional Bayesian evidence update. An alternative specification examines privileged-access concentration, contractor access, offboarding weaknesses, early anomalies, separation of duties, and monitoring. The draft includes explicit equations, variable definitions, sensitivity examples, a register of eight incident families, linked sources, reproduced numerical checks, and a proposed empirical validation protocol. Python source and reproduced results are embedded as PDF attachments. Status: work in progress, version 0.1. Not peer reviewed. Model coefficients and evidence likelihood ratios remain untrained assumptions. Numerical verification establishes implementation consistency, not predictive accuracy.