Risk Management With S&P 500 Stock Price Regression Forecast

Modeling the Apple Stock with Stepwise and Multiple Regression

Abstract

For short-term stock market traders who trade with large amounts of capital within a short time frame, having the analytical tool to predict stock price is crucial for the risk management and planning of entry and exit. While retail traders rely on the visualizations of technical indicators to anticipate general price direction and trend lines to estimate areas of concentrated demand or supply, multiple regression modeling uses known quantifiable indices whose functions are based on historical prices of an equity to yield a concise price prediction. This study aims to build an accessible and accurate model for retail traders without advanced software. In order to do that, we must optimize the quality of forecast while minimizing the number of independent variables. Through multiple linear regression and forward stepwise regression, we isolated 3 key variables – the daily closing price, Nasdaq 10-day moving average, and the Dow intraday average – which, when combined again in multiple regression, would produce a model with a multiple-R of 0.9995 and a low standard error of 2.44556268.