Numerical inference using stochastic volatility and time-series prediction on stock market prices

2^nd International Conference and Exhibition on Industrial Engineering

November 16-18, 2015 Dubai, UAE

Shoaib Shaikh

Florida Polytechnic University, USA

Posters-Accepted Abstracts: Ind Eng Manage

Abstract :

In a unique approach, advanced statistical modeling techniques coupled with predictive machine-learning are blended to forecast future stock price movement. From the outset, the stock data is analyzed and â�?�?cleansedâ�? for non-stationarity, including any underlying seasonal or trend components. The methods for characterizing the data include detrending and differentiation for normalization purposes. Next, using statistical inference techniques, hypothesis tests and confidence intervals are calculated to provide insight on potential future movement in stock price. This is further enhanced by determining volatility bands above and below the current stock price. These bands measure the variabiltiy of price movement based on historical data. For prediction purposes, two methods applied for this research include auto regressive integrated moving average (ARIMA) and artificial neural networks (ANN). Trained ARIMA models provide greater insight between relationships in the historical stock prices and future stock price movement. ANN will be used as a verificiation and confirming quality measure to the ARIMA forecast.

Biography :

Email: sshaikh@flpoly.org