US/Mountain, 28 May - 2 June 2017
- Published on Friday, 10 August 2012 14:56
Scientists aim to forecast natural or economic disasters by identifying statistical anomalies.
Professional Dragon King hunter Didier Sornette from the Department of Management, Technology and Economics, ETH Zurich, Switzerland, together with his colleague Guy Ouillon, present the many facets of Dragon Kings in a review just published in EPJST. Their work has just appeared alongside nineteen other contributions exploring the ways in which this emerging field of statistical analysis could become further established.
Dragon Kings are events akin to catastrophes. They don’t belong to the same power law regime as the more standard events. For example, they can be found in financial market bubbles ending in crashes, neuron-firing cascades leading to epileptic seizures, forest fires, the distribution of city sizes, insurance claims, and even in seemingly more mundane systems such as a stick balancing on a fingertip that eventually falls to one side or the other. Their name refers to the extreme behaviour of dragons stemming from their supernatural powers.
This review focuses on elucidating how Dragon Kings are created and can be detected. It also gives an overview of their empirical evidence in abnormal rainfall, hurricanes, and sudden events such as landslides and snow avalanches. The authors also outline the limitations of this sort of statistical analysis. For example, despite being sometimes interpreted as featuring characteristic events of Dragon Kings, great earthquakes may not be formally confirmed as such.
Finally, the authors share their views on the importance of devising prediction models that could become the basis for Dragon Kings simulators. These could be designed to help interpret the warning signs of complex systems evolving from their safe equilibrium into extreme events such as the subprime crisis, and to steer them into sustainability and ultimately avoid such crisis.
Dragon-kings : mechanisms, statistical methods and empirical evidence
D. Sornette and G. Ouillon, Eur. Phys. J. Special topics 205 1-26 (2012), DOI: 10.1140/epjst/e2012-01559-5