EDDIE for investment opportunities forecasting: Extending the search space of the GP

2011 IEEE Symposium on Computational Intelligence for Financial Engineering and Economics (CIFEr)

Chen

Tsang

2011

Self Cite

Abstract-This paper extends a previous market microstructure model, where we used Genetic Programming (GP) as an inference engine for trading rules, and Self Organizing Maps as a clustering machine for those rules. Experiments in that work took place under a single financial market and investigated whether its behavior is non-stationary or cyclic. Results showed that the market's behavior was constantly changing and strategies that would not adapt to these changes, would become obsolete, and their performance would thus decrease over time. However, because experiments in that work were based on a specific GP algorithm, we are interested in this paper to prove that those results are independent of the choice of such algorithms. We thus repeat our previous tests under two more GP frameworks. In addition, while our previous work surveyed only a single market, in this paper we run tests under 10 markets, for generalization purposes. Finally, we deepen our analysis and investigate whether the performance of strategies, which have not co-evolved with the market, follows a continuous decrease, as it has been previously suggested in the agent-based artificial stock market literature. Results show that our previous results are not sensitive to the choice of GP. Strategies that do not co-evolve with the market, become ineffective. However, we do not find evidence for a continuous performance decrease of these strategies. I. INTRODUCTIONThere are several types of models in the agent-based financial markets literature. One way of categorizing them is to divide them into the N -type models and the Santa-Fe Institute (SFI) like ones (see [1] for details). The former type of models focuses on the mesoscopic level of markets, by allowing agents to choose among different types of strategies. A typical example is the fundamentalist-chartists model. Agents in this model are presented with these two strategy types and at any given time they have to choose between these two. Other examples of such N -type models in the literature are [2], [3], [4]. A typical area of investigation of these models is fraction dynamics, i.e. how the fractions of the different strategy types change over time. However, what is not presented in most of these models is novelty-discovering agents. For instance, in the fundamentalist-chartists example, agents can only choose between these two types; they cannot create new strategies that do not fall into these two types. On the other hand,

Section: ) Simple Gp Algorithmmentioning

confidence: 99%

Section: ) Simple Gp Algorithmmentioning

confidence: 99%

Investigating the effect of different GP algorithms on the non-stationary behavior of financial markets

2011 IEEE Symposium on Computational Intelligence for Financial Engineering and Economics (CIFEr)

Chen

Tsang

2011

Self Cite

“…Hence, it could be argued that the higher rate of the "<VarConstructor>< RelationOperation> <VarConstructor>" production attributed to the superiority of the first tree. Lastly, Table III presents the average computational time of a single run 6 of each algorithm for the Hammerson dataset. As we can observe, ED8-ATTR is only 1.5 seconds slower than ED8, which is 11 seconds slower than ED7.…”

Section: R Esultsmentioning

confidence: 99%

Using Attribute Construction to Improve the Predictability of a GP Financial Forecasting Algorithm

2013 Conference on Technologies and Applications of Artificial Intelligence

Otero

2013

Self Cite

Abstract-Financial forecasting is an important area in computational finance. EDDIE 8 is an established Genetic Programming financial forecasting algorithm, which has successfully been applied to a number of international datasets. The purpose of this paper is to further increase the algorithm's predictive performance, by improving its data space representation. In order to achieve this, we use attribute construction to create new (high-level) attributes from the original (low-level) attributes. To examine the effectiveness of the above method, we test the extended EDDIE's predictive performance across 25 datasets and compare it to the performance of two previous EDDIE algorithms. Results show that the introduction of attribute construction benefits the algorithm, allowing EDDIE to explore the use of new attributes to improve its predictive accuracy.

“…However, what is evident is that both academics and people who work in the industry tend to use very specific L lengths for the indicators; for instance, 20 days is a common short-length period and 50 days is a common long-term period. In [7], it was argued that this method is not very flexible and cannot guarantee that specific pre-specified indicators are necessarily the best ones. For example, nobody can guarantee that a 20 days MA is definitely more effective than a 25 days MA, under all possible datasets.…”

Section: Financial Forecastingmentioning

confidence: 99%

“…5 and RIPPER. For the purposes of the GP, we will be using EDDIE [7], an algorithm especially designed for classification financial forecasting problems. For the purposes of ACO, we will be using the Unordered cAnt-Miner PB [10], which is a classification algorithm.…”

Section: Introductionmentioning

confidence: 99%

A Comparative Study on the Use of Classification Algorithms in Financial Forecasting

Otero

Applications of Evolutionary Computation

2014

Self Cite

Abstract. 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.