Created by Paul Adams and Jeff Nguyen.
Market Trading Investments Forecasting Strategy with ARIMA, Neural Networks using Multi-Layered Perceptrons, Signal-Plus-Noise, Vector Autoregressive (VAR) and composite ensemble models.
In this project, we analyze 3,202 stock price and volume data time series traded on the NASDAQ exchange between May 30th and October 30th, 2019. This date range was selected for its distance from recent, significant market disruption in order to build a proof of concept model for selecting and forecasting prices. Data was sourced as comma-separated values through API from AlphaVantage.
We applied automated, iterative pre-processing to analyze all stocks and provide candidate models for direct analysis. Through this analysis, we identified one stock and modeled its time series using ARIMA, Neural Networks using Multi-Layered Perceptrons, Signal-Plus-Noise, Vector Autoregressive (VAR), and composite ensemble models. Forecasting accuracy was assessed using the Average Squared Error (ASE).
