Evolutionary rule-based system for IPO underpricing prediction

Abstract

Academic literature has documented for a long time the existence of important price gains in the first trading day of initial public offerings (IPOs).Most of the empirical analysis that has been carried out to date to explain underpricing through the offering structure is based on multiple linear regression. The alternative that we suggest is a rule-based system defined by a genetic algorithm using a Michigan approach. The system offers significant advantages in two areas, 1) a higher predictive performance, and 2) robustness to outlier patterns. The importance of the latter should be emphasized since the non-trivial task of selecting the patterns to be excluded from the training sample severely affects the results.We compare the predictions provided by the algorithm to those obtained from linear models frequently used in the IPO literature. The predictions are based on seven classic variables. The results suggest that there is a clear correlation between the selected variables and the initial return, therefore making possible to predict, to a certain extent, the closing price.

Random Forest Prediction of IPO Underpricing

“…As we can see, it shows a positive mean return, high kurtosis, and the classic right-skewness corresponding to several hot IPOs with extremely high initial returns (in this case, the most extreme case almost got to quadruple the offering price with a 372% price increase). The mentioned basic characteristics are consistent with the structure of the sample used in previous studies, such as [1,7,18,19,21], among many others. …”

GPPE: a method to generate ad-hoc feature extractors for prediction in financial domains

“…Once the issue price is selected, the company is committed to maintain this price for the entire offering. If the fina price is signifi cantly over or under the IPO issue price, company will suffer important losses in the form of underpricing or overpricing (for more information about underpricing/overpricing losses, reader can refer to [6]). Thus, it is vital that the company and its investment bankers price the IPO as closely as possible to the fina price.…”

“…The IPO sample is composed of 1000 companies going into the US Stock Market for the firs time, between April 1996 and November 1999 [6]. Each company is characterized by seven explicative variables: underwriter prestige, price range width, price adjustment, offer price, retained stock, offer size and relation to tech sector.…”

“…An IPO is the firs public stock offering of a company [5,6]. The existence of important variations on the price of the IPOs in the firs day of trading has been documented for many years.…”

Time series prediction evolving Voronoi regions

“…Packard used genetic algorithms [9] to face the problem of predicting dynamic systems: in [15,16,18], a new approach is suggested, selecting subsets of the input space by means of conditional rules for time series prediction. We used Packard's idea to improve his approach in previous works [12,22].…”