In the aftermath of high-profile collapses such as the FTX exchange and a staggering $24.2 billion in reported cryptocurrency crime in 2023, the digital asset market is a volatile frontier. However, new research from the Alberta School of Business is introducing advanced data analytics frameworks designed to bring transparency and stability to these complex financial ecosystems.
Ivor Cribben, a renowned statistician and researcher at the Alberta School of Business, has co-authored two studies that leverage "tensors" — multi-dimensional data — to identify hidden market risks and detect sophisticated fraud. The research arrives at a critical time as regulators and the business community struggle to keep pace with the "Wild West" of digital finance.
Traditional financial models often fail in the cryptocurrency space because they treat different exchanges and time periods as independent. This was a fatal flaw during the FTX collapse, where undisclosed trading relationships and hidden leverage across platforms created systemic risks that caught the world off guard.
Cribben’s first article introduces a model called "WeDTLasso," which captures how risks propagate through networks of exchanges over time. By accounting for these "temporal dependencies" the researchers found they could identify systemic risks before they materialize. For the business community, the practical application is significant: the model delivered up to 50% improvements in risk-adjusted returns (Sharpe ratios) compared to existing portfolio construction methods.
“Neglecting temporal dependence among observations can lead to incorrect or even false conclusions and result in an underestimation of risk, potentially undermining regulatory oversight efforts," says Cribben.
While one statistical tool helps investors manage risk, the second, a method labeled "TenSeg," focuses on security. As market manipulators move funds across multiple tokens and ledgers to hide their tracks, they create a "network of networks" that is notoriously difficult to monitor.
TenSeg acts as an early warning system by detecting "change-points" — sudden shifts in the structure of trading data that often signal activities such as pump-and-dump schemes (a form of securities fraud that involves artificially inflating the price of an owned stock or asset through false, misleading, or greatly exaggerated positive statements). When applied to data from the Ethereum blockchain (the digital footprint), the method successfully identified changes across multiple trading networks simultaneously, outperforming current state-of-the-art techniques.
"Such tools are essential for both pre-failure intervention and post-failure policy responses, representing a crucial step toward realizing cryptocurrency’s innovative potential while ensuring adequate protection for retail investors and maintaining market stability," says Cribben.
In order for the industry to move beyond its current volatility, Cribben argues it must adopt statistical methods capable of identifying systemic risks before they lead to large-scale collapses or successful market manipulation.
This research offers clear directives for two primary groups:
Read the full research articles published in the Journal of the Royal Statistical Society Series A:Statistics in Society at DOI: 10.1093/jrsssa/qnag028 and DOI: 10.1093/jrsssa/qnaf180.
Bold ideas in action! From reimagining industries to building sustainable infrastructure, the U of A is helping to achieve the UN's goal of creating a more resilient, tech-powered world.
See How We're Igniting InnovationBecome part of our community. Get the latest news and event information from the Alberta School of Business in your inbox every month.