Correlating Market Movements With Sentiments: A Longitudinal Study

An adaptation of two well-established measures of changes in financial markets return and volatility is presented for the analysis of changes: (a) in consumer confidence about national economies, and (b) in sentiments articulated in financial news stories text and extracted automatically by counting the frequency of polarity words. This analysis is then compared and contrasted with the changes in key stock market indices by looking at the actual distribution of returns of the sentiment, consumer confidence and stock market indices. We have looked at three stock market indices (S&P 500, ISEQ, and Nikkei) for periods of two years or more. For consumer confidence we have analysed the Michigan Consumer Confidence Index and the Irish Economic and Social Research Institute’s Consumer Confidence survey. Sentiment analysis was carried out on a ten-year archive of stories published in the Irish Times (19952005); The distributions of each of the three returns, all the stock market indices, sentiment time series, and to a lesser extent the two consumer confidence surveys, appear not to follow the normal distribution of a random variable.