Prediction of initial public offering short-term performance using nearest neighbor and support vector machine models

The first public stock released by a company is defined as initial public offering. In recent years, there have been many research on IPO short-term performance. The present research aims to test different classification models to find model having great efficiency in Prediction of initial public offering short-term performance. The study included 60 IPO in Tehran Stock Exchange during the period 2005-2015. In the proposed framework, average surplus return for first three days of IPO has a positive value and equals to 1.3% although this value is not high same as developed markets. 10-fold cross-validation method was used for evaluating and monitoring nearest neighbor, support vector machines, decision trees and naive-Bayes and results showed that among monitoring nearest neighbor and support vector machines models has high accuracy in predicting IPO short-term performance.