Stock Trading: Prediction of Auction Volumes

Abstract

With data analysis methods comes the ability to make predictions about the future behavior of numerous types of data. This project seeks to analyze stock market closing auctions. Stock exchanges hold closing auctions at the end of a trading day to determine the closing price for each stock. We analyzed volume, the total value of stock exchanged, during the closing auctions from stock exchanges covering a set of 900 stocks over 800 days. In order to complete this project, we built a model to predict stock trade volumes utilizing data provided on challengedata.ens.fr. A benchmark was created by Capital Fund Management, the providers of the challenge, that utilizes a linear fit method for our model to be compared against. We have interpreted and cleaned the data in order to predict exchanged volume of a given stock during closing auctions.