The aim of this thesis is to study and identify time periods of high activity in commodity and stock market sentiment based on a data mining approach. The method is to develop tools to extract relevant information from web searches and Twitter feeds based on the tally of certain keywords and their combinations at regular intervals. Periods of high activity are identified by a measure of complexity developed for analysis of living systems. Experiments were conducted to see if the measure of activity could be applied as a predictor of changes in stock market and commodity prices.