Insider Trading Patterns: A Price Indicator For US Equities

Open Access
- Author:
- Boehringer, Paul H
- Area of Honors:
- Industrial Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Dr. Guodong Pang, Thesis Supervisor
Dr. Catherine Mary Harmonosky, Thesis Honors Advisor - Keywords:
- Insider Trading
Equities
data scraping
Data
finance - Abstract:
- This paper investigates the potential of insider trading as a price indicator for US equities along with a non-stochastic relationship between insider transactions and prices. Further tests are done to see if there is a difference in the quality of information coming from insiders buying compared to selling, as the majority of the current literature tends to discredit the use of insider selling as an input in any investment strategy. Analyzing the returns of a company’s stock for a series of holding periods ranging from five days to four years after each insider buy and sell shows insiders have a significant advantage on timing regarding when to buy and sell their own stocks. After confirming a difference in the behavior of buys and sells a price forecasting model was built. Including either buying or selling information during model training increase prediction accuracy compared to a model without any insider trading information. Further, the most accurate models include both buy and sell information. However, there was no significant difference between the quality of buying and selling info. There is also significant evidence that price formation is non-stochastic based on the financial state and insider trading patterns for a company over time. Lastly, both a linear regression and neural network were used to make pricing predictions, while the linear regression was more accurate for pricing predictions the activation function in the network models allowed for significantly more accurate timing predictions, that is when the high and low prices will occur in a quarter. The insights from this study are found using the price, insider trading, and quarterly financial histories from 3,505 companies with 3,021,444 insider transactions and 139,986 financial quarters shared between these companies.