Relation between analyst reports and stock market prices

Abstract

This project aims to examine the correlation between analyst predictions and stock market prices. Analysts publish reports at regular intervals for many regularly traded stocks on the market. The stock prices can then follow the predictions of the report or move in a completely opposite direction. This project first tries to and out important points of change in the stock price data using a probabilistic change point method given by adams et. al [2]. The change points are then used as indicators of the analyst reports that need to be examined. The analyst reports are examined using a text analysis technique called Latent Dirichlet Allocation[3] that gets topics(a group of words with associated probabilities) from the reports. Finally, the similarity in these topics is analyzed to and out the extent to which analyst reports are able to predict the future performance of a stock.