[fod] Seminar on machine learning techniques in the capital market

[Janet Wiles <janetw@itee.uq.edu.au>]

This seminar describes work that Rolf has completed during his PhD.
Everyone is welcome.

Title:   Predicting First-Day Returns of Cross-Industry Initial Public
	 Offerings in the US Stock Market
Speaker: Rolf Mitsdorffer <RMitsdor@bigpond.net.au>
Date: 	 Thursday 15th November, 11am
Room: 	 420 GP-south (bldg 78) UQ, St. Lucia Campus


Abstract

The raising of $US340 billion through IPOs between 1970 and 1997
demonstrates the importance of IPOs for the US capital market. IPOs on the
first trading day often rise to artificially high levels.

Against the background of the Capital Asset Pricing Model (CAPCM), which
is based on the Random Walk hypothesis, the hypothesis that market
sentiment and IPO specific attributes are responsible for first-day IPO
returns in the US stock market is tested using machine learning methods.

Methods used include Bayesian classifications, support vector machines,
decision tree techniques, rule learners and artificial neural networks.
The outcomes are predictions and rules associated with first-day returns
of historical cross-industry IPOs.

This seminar will give an introduction to current research on machine
learning techniques in the capital market.