The Lunar Algorithm That Predicts the Future

A new extension of an algorithm that helped put a man on the moon can help us to predict the advent of a crisis and other binary variables that evolve over time. The algorithm is called Kalman Filter. In 1969, the on-board computer that guided moon landing had one. It evaluated information coming from the module and filtered measurement errors and random components in order to estimate the actual position and to predict the future position.

Fifty years later, two academics (Daniele Durante and Sonia Petrone) and two PhD students (Augusto Fasano and Giovanni Rebaudo) in the Department of Decision Sciences of Bocconi University have developed a method that efficiently extends the Kalman Filter to non-continuous binary variables that indicate the occurrence or not of an event over time."Until now, any application of the Kalman Filter to binary variables was only approximate", Daniele Durante says. "Now, we can predict whether the stock market will open up or down based on the occurrence of certain conditions in the past, or who will win the next boat race between Cambridge and Oxford based on past competitions".
The method makes use of online updating: knowledge about the occurrence of a given event is updated sequentially with each new observation. The broad range of applications goes from weather events to cyber attacks. "Our formula applies to 0/1 variables. We intend to extend it to non-continuous variables with more than two values, for instance the occurrence of a strong, medium or weak crisis".

Read more about this topic:
Financial Crises: Who Can Predict Them and How. By Carlo Favero
The Ratio of Uncertainty
The Controllers That Report the Risk of a Crisis
How Companies Overcome a Crisis