GP age-layer and crossover effects in bid-offer spread prediction

The bid-offer spread on equity options is a key source of profits for market makers, and a key cost for those trading in the options. Spreads are influenced by dynamic market factors, but is there also a predictable element and can Genetic Programming be used for such prediction? We investigate a standard GP approach and two optimisations — age-layering and a novel crossover operator. If both are beneficial as independent optimisations, will they be mutually beneficial when applied simultaneously? Our experiments show a degree of success in predicting spreads, we demonstrate significant benefits for each optimisation technique used individually, and we show that when both are used together significant detrimental over-fitting can occur. Categories and Subject Descriptors I.2.M [Artificial Intelligence]: Miscellaneous General Terms Algorithms, Experimentation Keywords GP, Finance, Options, Spreads, Age Layers, ALPS, Crossover

• Type: Conference
• Year: 2008
• Where: GECCO
• Authors: Amy Willis, Suneer Patel, Christopher D. Clack
• Keywords: Dynamic Market Factors, Equity Options, GECCO 2008, Optimization, Standard Gp Approach