This work had as objective the construction of indicators that anticipate the movements of economic cycles, and
thereby anticipate their onset through the dynamics of the economies of the 27 countries responsible for more than
84% of the global economy and the dynamics of the prices of the main commodities responsible for most of world
trade. To this end, Several steps were necessary until the operationalization of the study. At first, it was necessary to
collect all available data to all countries initially selected in the database (St. Louis FED) of the American Central
Bank - achieved through an automation for the creation of a glossary containing all the details for more than 39
thousand series. Subsequently, the filters with the imposed requirements were applied, leaving just over 5 thousand.
There was a need to reduce the number of series due to the limited computational power, therefore, the use
of the model for variable selection, Elastic-Net, which selected the most relevant to the quarterly GDP of each
country. From there, the data were grouped in blocks and sub-blocks organized by the dynamics of the
economies. The model used to construct the indicator was the Factor Model. Hierarchical Dynamics that
aggregates series with common factors, by blocks sectoral, subsectoral and/or idiosyncratic components.
The resulting factor had good adherence to tests, showing good ability to predict crises. When generating the
probabilities softened, it was possible to foresee the crisis of the Coronavirus already in January 2020, for the
Subprime crisis had a more modest performance, can be explained by the fact that it contains the weight of all
other economies in addition to the American one, and also due to the shortened time horizon. The factor of
Financial Market generated very efficient probabilities, anticipating all major events, but also issuing alerts for
periods that do not there were crises, but being more localized recessions.