Anticipating a Stock Market Crash - Evidence from the S&P 500
Rintala, Panu (2012)
2012
Kuvaus
Opinnäytetyö kokotekstinä PDF-muodossa.
Tiivistelmä
The main objective of this study is to find out if it is possible to create a simple model that anticipates the upcoming large-scale negative movements in the stock market values, i.e. the stock market crashes. The study uses the S&P 500 index as the benchmark for stock markets. Four variables were chosen to represent the anticipatory factors: term spread, change in U.S. money supply, change in gold price and P/E ratio. Those four variables were chosen because they are easily obtainable and have solid connections with the stock prices.
The observation period of the study covers more than four decades of monthly data and ranges from March 1970 to October 2010. Multiple logistic regression model was chosen for the statistical analysis. In addition a simple trading test will be conducted between a trading portfolio and a buy-and-hold portfolio in order to see the applicability of the model. Based on the results of the trading test, Sharpe measures will be calculated in order to evaluate the risk-adjusted performance between the two portfolios.
According to the results of the logistic regression analysis, the term spread and the logistic transformation of the P/E ratio are statistically significant variables when anticipating a stock market crash. The results of the simple trading test were also promising and with the use of the “crash model”, the final value of the trading portfolio was over two times the value of the buy-and-hold portfolio. The Sharpe measures were 0,532 for the trading portfolio and 0,287 for the buy-and-hold portfolio, indicating better total as well as risk-adjusted returns for the trading portfolio.
The observation period of the study covers more than four decades of monthly data and ranges from March 1970 to October 2010. Multiple logistic regression model was chosen for the statistical analysis. In addition a simple trading test will be conducted between a trading portfolio and a buy-and-hold portfolio in order to see the applicability of the model. Based on the results of the trading test, Sharpe measures will be calculated in order to evaluate the risk-adjusted performance between the two portfolios.
According to the results of the logistic regression analysis, the term spread and the logistic transformation of the P/E ratio are statistically significant variables when anticipating a stock market crash. The results of the simple trading test were also promising and with the use of the “crash model”, the final value of the trading portfolio was over two times the value of the buy-and-hold portfolio. The Sharpe measures were 0,532 for the trading portfolio and 0,287 for the buy-and-hold portfolio, indicating better total as well as risk-adjusted returns for the trading portfolio.