An Efficient Learning System for Knowledge of Asset Management

Abstract

This paper examines if it is possible to obtain valuable knowledge for asset management by performing text mining on an enormous volume of analyst reports. Analyst reports are the evaluation reports of firms that are published by securities analysts. These reports describe the business conditions of firms mainly with a large amount of text information. However, it is impossible for a human being to read and understand all of the reports within a limited amount of time. To address this problem, we extract information from analyst reports automatically using text mining methods and analyze the influences of the reports. As a result of analyses, we confirm that the analyst reports contain valuable information that affect to stock prices. We also find that the stock prices react to the information before the report is published, which indicates that analysts are affected by the opinions of the other analysts.