||Composite indicators are a widespread method for business cycle analysis, especially because they can be easily interpreted although they summarize multidimensional relationships between individual economic time series. The composite indicators can be divided into leading, coincident and lagging ones with regard to the reference time series (usually GDP or industrial production index). The composite leading indicators (CLIs) are the most frequently constructed type of these indicators as they can predict the future states of the economic activity.The methodology of the composite indicators construction is described in detail by several organizations and it is always based on many subjective expert choices and decisions. This thesis proposes a new algorithmic approach which enables to fully automate the whole computational process. This method allows to create the composite indicators faster than any other available technique and it replaces the subjectivity of choices and increases tractability of the calculations.This thesis, based on computational statistics, aims to describe the modifications that are needed to fully automate the construction of the composite indicators. It compares the results with the CLIs which are regularly published by the Organization for Economic Co-operation and Development (OECD) and shows, that the process can be automated. A simple set of rules is designed to enable objective comparison between two composite indicators. This thesis also suggests how to extend the OECD methodology to improve the CLIs performance and to analyse the relationships between the business cycles of multiple countries.