The goal of getting computers to automatically solve problems is central to artificial intelligence, machine learning, and the broad area encompassed by what Turing called “machine intelligence ” [161, 162].

Abstract. A developmental co-evolutionary genetic programming approach (PAM DGP) is compared to a standard linear genetic programming (LGP) implementation for trading of stocks across market sectors. Both implementations were found to be impressively robust to market fluctuations while reacting efficiently to opportunities for profit, where PAM DGP proved slightly more reactive to market changes than LGP. PAM DGP outperformed, or was competitive with, LGP for all stocks tested. Both implementations had very impressive accuracy in choosing both profitable buy trades and sells that prevented losses, where this occurred in the context of moderately active trading for all stocks. The algorithms also appropriately maintained maximal investment in order to profit from sustained market upswings.

Summary. A developmental co-evolutionary genetic programming approach (PAM DGP) is compared to a standard linear genetic programming (LGP) implementation for trading of stocks in the technology sector. Both interday and intraday data for these stocks were analyzed, where both implementations were found to be impressively robust to market fluctuations while reacting efficiently to opportunities for profit. PAM DGP proved slightly more reactive to market changes compared to LGP for intraday data, where the converse held true for interday data. Both implementations had very impressive accuracy in choosing both profitable buy trades and sells that prevented losses for both interday and intraday stock data. These successful trades occurred in the context of moderately active trading for interday prices and lower levels of trading for intraday prices. Technical analysis of the stock market involves attempts to examine the past effects of market movements in order to anticipate what traders will do next to affect the market. Such analysis involves the use of technical indicators to examine price trends and trading volume in order to identify the likely future trading activity and change in price

Abstract. Many automatically-synthesized programs have, like their hand-made counterparts, numerical parameters that need to be set properly before they can show an acceptable performance. Hence, any approach to the automatic synthesis of programs needs the ability to tune numerical parameters efficiently. Grammatical Evolution (GE) is a promising grammar-based genetic programming technique that synthesizes numbers by concatenating digits. In this paper, we show that a naive application of this approach can lead to a serious number length bias that in turn affects efficiency. The root of the problem is the way the context-free grammar used by GE is defined. A simple, yet effective, solution to this problem is proposed. 1

The information provided is the sole responsibility of the authors and does not necessarily reflect the opinion of the members of IRIDIA. The authors take full responsability for any copyright breaches that may result from publication of this paper in the IRIDIA – Technical Report Series. IRIDIA is not responsible for any use that might be made of data appearing in this publication. Exposing a Bias Toward Short-Length Numbers

Genetic Programming (GP) is an automated computational programming methodology, inspired by the workings of natural evolution techniques. It has been applied to solve complex problems in multiple domains including finance. This paper illustrates the application of an adaptive form of GP, where the probability of crossover and mutation is adapted dynamically during the GP run, to the important real-world problem of options pricing. The tests are carried out using market option price data and the results illustrate that the new method yields better results than are obtained from GP with fixed crossover and mutation rates. The developed method has potential for implementation across a range of dynamic problem environments. Categories and Subject Descriptors

GEVA is an open source implementation of Grammatical Evolution in Java developed at UCD’s Natural Computing Research & Applications group. As well as providing the characteristic genotype-phenotype mapper of GE a search algorithm engine and a simple GUI are also provided. A number of sample problems and tutorials on how to use and adapt GEVA

Abstract. Genetic Programming (GP) is an automated computational programming methodology which is inspired by the workings of natural evolution techniques. It has been applied to solve complex problems in multiple application domains. This paper investigates the application of a dynamic form of GP in which the probability of crossover and mutation adapts during the GP run. This allows GP to adapt its diversitygenerating process during a run in response to feedback from the fitness function. A proof of concept study is then undertaken on the important real-world problem of options pricing. The results indicate that the dynamic form of GP yields better results than are obtained from canonical GP with fixed crossover and mutation rates. The developed method has potential for implementation across a range of dynamic problem environments. 1

calls & calendarEDITORIAL

Recently, a form of memory usage was introduced for genetic programming (GP) called “soft memory. ” Rather than have a new value completely overwrite the old value in a register, soft memory combines the new and old register values. This work examines the performance of a soft memory linear GP and developmental GP implementation for stock trading. Soft memory is known to more slowly adapt solutions compared to traditional GP. Thus, it was expected to perform well on stock data which typically exhibit local turbulence in combination with an overall longer term trend. While soft memory and standard memory were both found to provide similar impressive accuracy in buys that produced profit and sells that prevented losses, the softer memory settings traded more actively. The trading of the softer memory systems produced less substantial cumulative gains than traditional memory settings for the stocks tested with climbing share price trends. However, the trading activity of the softer memory settings had moderate benefits in terms of cumulative profit compared to buy-and-hold strategy for share price trends involving a drop in prices followed later by gains.