A critical event is a type of event that has a high impact on the environment in which it happens. Traditionally, statistical and machine learning techniques have been used to know their effects. These techniques produce good results when the event under analysis is based almost exclusively on historical data. The objective of this proposal is to improve the outcomes of statistical and machine learning techniques by proposing a model that looks for the most similar previous event using Elastic Patterns and performs knowledge-oriented data processing to find specific causes that influence the outcome, getting behavioural rules. The model is applied to analyze when stock markets will recover to pre-pandemic levels in G20 countries during a pandemic such as COVID-19. Applying Elastic Patterns, we have discovered that the epidemic most similar to COVID-19 is swine flu (H1N1 virus). The recovery of stock indices due to COVID-19 is proportionally similar to that of swine flu. By applying knowledge-oriented data processing, we found that countries that injected more money into the economy (the US), initially had few deaths because it was summer (Argentina, South Africa), or took early action to contain the spread of Covid-19 (China, South Korea), saw their stock market indices recover sooner.