Forecasting the critical points of stock markets’ indices using log-periodic power law

This article presents Log-Periodic Power Law and considers its usefulness as a forecasting tool on the financial markets. One of the estimation methods of this function was presented and six models were built, based on time series of the DJIA and the WIG20. Estimated models were utilized to predict crashes of those indices. The variations between the actual values of analyzed indices observed in the forecasted period and values observed in the actual period of their downturn were assembled to assess the results. In three cases, relative errors were below 5%; and in three cases, they were higher than 15%