A Breakthrough of Digital Assets Security in Crypto Insurance: Cyber Risk Prediction Models

We use unsupervised cluster analysis to divide the world into geographically distinct groups and then use our proposed model to analyze the Privacy Right Clearinghouse (PRC) data breach chronology. We model zero losses using a covariate-dependent probability, moderate losses using a finite mixture distribution, and large losses using an extreme value distribution to capture the heavy-tailed nature of the loss data. The risks and opportunities that digital technologies, devices and media bring us are manifest. Cyber risk is never a matter purely for the IT team. An organisation's risk management function needs a thorough understanding of the constantly evolving risks, as well as the practical tools and techniques available to address them. It is challenging to model the whole range of losses using a typical loss distribution when considering cyber losses in terms of the number of records exposed as a result of cyber events since these losses frequently include a significant share of zeros, distinct characteristics of mid-range losses, and high losses. By suggesting a three-component splicing regression model that can concurrently simulate zeros, moderate, and substantial losses as well as take into account heterogeneous effects in mixture components, we attempt to solve this modeling dificulty. Parameters and coeffcients are estimated using the Expectation Maximization (EM) algorithm. Combining with our frequency model (generalized linear mixed model) for data breaches, aggregate loss distributions are investigated and applications on cyber insurance pricing and risk management are discussed.