EvoFIN Logo

Detailed Programme

Natural Computing Methods in Finance and Economics

EvoFIN is the only European event specifically dedicated to the application of Evolutionary Computation, and other Natural Computing methodologies to finance and economics. It gives researchers in those fields, as well as people from industry, an opportunity to present their latest research and to discuss current developments and applications.

Topics of interest include (but not limited to):

Programme Committee

EvoFIN Programme

Wed 1120-1300  EvoFIN 1 
Chair: Ahmed Kattan

On PBIL, DE and PSO for Optimization of Reinsurance Contracts    Omar Andres Carmona Cortes, Andrew Rau-Chaplin, Duane Wilson, Jürgen Gaiser-Porter
In this paper, we study from the perspective of an insurance company the Reinsurance Contract Placement problem. Given a reinsurance contract consisting of a fixed number of layers and a set of expected loss distributions (one per layer) as produced by a Catastrophe Model, plus a model of current costs in the global reinsurance market, identifying optimal combinations of placements (percent shares of sub-contracts) such that for a given expected return the associated risk value is minimized. Our approach is to explore the use evolutionary algorithms with the goal of determining which evolutionary optimization approach leads to the best results for this problem, while being executable in a reasonable amount of time of realistic industrial sized problems. Our approach evaluates the performance of the algorithms solving larger "real world" problem instances than previous methods.

Algebraic level-set approach for the segmentation of financial time series     Rita Palivonaite, Kristina Lukoseviciute, Minvydas Ragulskis
Adaptive algebraic level-set time segmentation algorithm of financial time series is presented in this paper. The proposed algorithm is based on the algebraic one step-forward predictor with internal smoothing, which is used to identify a near optimal algebraic model. Particle swarm optimization algorithm is exploited for the detection of a base algebraic fragment of the time series. A combinatorial algorithm is used to detect intervals where predictions are lower than a predefined level. Moreover, the combinatorial algorithm does assess the simplicity of the identified near optimal algebraic model. Automatic adaptive identification of quasi-stationary segments can be employed for complex financial time series.

Dynamic Index Trading using a Gene Regulatory Network Model     Miguel Nicolau, Michael O'Neill, Anthony Brabazon
This paper presents a realistic study of applying a gene regulatory model to financial prediction. The combined adaptation of evolutionary and developmental processes used in the model highlight its suitability to dynamic domains, and the results obtained show the potential of this approach for real-world trading.

Geometric Semantic Genetic Programming for Financial Data     James McDermott, Alexandros Agapitos, Anthony Brabazon, Michael O'Neill
We cast financial trading as a symbolic regression problem on the lagged time series, and test a state of the art symbolic regression method on it. The system is geometric semantic genetic programming, which achieves good performance by converting the fitness landscape to a cone landscape which can be searched by hill-climbing. Two novel variants are introduced and tested also, as well as a standard hill-climbing genetic programming method. Baselines are provided by buy-and-hold and ARIMA. Results are promising for the novel methods, which produce smaller trees than the existing geometric semantic method. Results are also surprisingly good for standard genetic programming. New insights into the behaviour of geometric semantic genetic programming are also generated.

Wed 1430-1610  EvoFIN 2 
Chair: Michael Kampouridis

Analysis of Dynamic Properties of Stock Market Trading Experts Optimized with Evolutionary Algorithm     Krzysztof Michalak
This paper concerns optimization of trading experts that are used for generating investment decisions. A population of trading experts is optimized using dynamic evolutionary algorithm. In the paper a new method is proposed which allows analyzing and visualizing the behaviour of optimized trading experts over a period of time. The application of this method resulted in an observation that during certain intervals of time the behaviour of the optimized trading experts becomes more stable. The trading experts that remain unchanged for a long time seem to be well adapted and their fitness does not deteriorate with time.

A Comparative Study on the Use of Classification Algorithms in Financial Forecasting     Fernando Otero, Michael Kampouridis
Financial forecasting is a vital area in computational finance, where several studies have taken place over the years. One way of viewing financial forecasting is as a classification problem, where the goal is to find a model that represents the predictive relationships between predictor attribute values and class attribute values. In this paper we present a comparative study between two bio-inspired classification algorithms, a genetic programming algorithm especially designed for financial forecasting, and an ant colony optimization one, which is designed for classification problems. In addition, we compare the above algorithms with two other state-of-the-art classification algorithms, namely C4.5 and RIPPER. Results show that the ant colony optimization classification algorithm is very successful, significantly outperforming all other algorithms in the given classification problems, which provides insights for improving the design of specific financial forecasting algorithms.

Pattern Mining in Ultra-High Frequency Order Books with Self-Organizing Maps     Piotr Lipinski, Anthony Brabazon
This paper addresses the issue of discovering frequent patterns in order book shapes, in the context of the stock market depth, for ultra-high frequency data. It proposes a computational intelligence approach to building frequent patterns by clustering order book shapes with Self-Organizing Maps. An experimental evaluation of the approach proposed on the London Stock Exchange Rebuild Order Book database succeeded with providing a number of characteristic shape patterns and also with estimating probabilities of some typical transitions between shape patterns in the order book.

On evolving multi-agent FX traders    Alexander Loginov, Malcolm Heywood
Current frameworks for identifying trading agents using machine learning are able to simultaneously address the characterization of both technical indicator and decision tree. Moreover, multi-agent frameworks have also been proposed with the goal of improving the reliability and trust in the agent policy identified. Such advances need weighing against the computational overhead of assuming such flexibility. In this work a framework for evolutionary multi-agent trading is introduced and systematically benchmarked for FX currency trading; including the impact of FX trading spread. It is demonstrated that simplifications can be made to the `base' trading agent that do not impact on the quality of solutions, but provide considerable computational speedups. The resulting evolutionary multi-agent architecture is demonstrated to provide significant benefits to the profitability and improve the reliability with which profitable policies are returned.