Two major obstacles prevent the large-scale uptake of emerging software technologies and non-linear modelling techniques such as chaos theory, fractals, non-parametric statistics, neural nets and genetic algorithms within the financial community. Firstly, non linear models based on emerging software technologies do not provide the diagnostic statistics to which practitioners are accustomed when using competing statistical techniques. Secondly, modelling complex inter-relationships between financial and economic factors with non-linear techniques, a data rich approach instead of a model rich approach is mostly employed. In an area such as finance, which can be labelled "not yet fully understood", such an approach is always susceptible.
Within the Esprit project HANSA (P6369), ARTS GmbH has developed limited research prototype versions of three different fund management schemes and a trading and forecasting system generator making extensive use of emerging software technologies. The schemes and the generator have partially been tested within HANSA, but need to go through extensive trials and further testing (due diligence procedure) before they can be deployed on a larger scale. Also in HANSA, Mimetics has developed MimeNice, a PC-based, general purpose neural network tool and environment which may serve as an implementation tool for the investment models. The London Business School, Decision Science Department is running the NeuroForecasting Centre, a centre for applied research in the use of advanced, non-linear forecasting techniques in finance.
ESTIMATION is a "live and on-line" application trial and is designed to fully evaluate the risks and benefits of intelligent, emerging software technologies in trading and fund management. This is carried out by the three ESTIMATION partners, assisted by the five industrial pilot users, Hermes, CitiBank, Kleinwort Benson/Dresdner Bank, Société Générale, and Deutsche Bank London, independently assessing and evaluating the different fund management schemes and the different aspects and risks involved in the overall money management process.
Specifically, simulation, field and acid tests are to be carried out, encompassing the development of diagnostic tests for non-linear modelling and risk management, testing whether tactical asset allocation and derivatives trading can be improved using emerging software technologies, comparing standard statistical methods with advanced non-linear models and testing one of the key assumptions in the uptake of the new technologies - the assumption that markets are partially predictable, exhibiting recurring trends and patterns, this time in the form of the much heralded Coherent Market Hypothesis.
The expected results from the application trial are to overcome the major obstacles in using emerging software technologies in finance as mentioned above, thus vastly reducing the risk of a larger scale deployment and demonstrating to a range of potential European users that AI-technologies can actively assist in solving a vast array of financial problems. Furthermore, during the course of the trial, risks still involved in employing these techniques can be better assessed and quantified and strategies for cushioning them will be developed.
After completion of ESTIMATION, a fund, based on the prototype models that were tested and risk adjusted in the ESTIMATION application trial might be launched. Furthermore, it should be possible to market a software environment composed of different AI-techniques, namely neural nets, standard financial analysis techniques and integrated diagnostics, specifically for non-linear modelling purposes. These diagnostics should be comparable to those, practitioners are accustomed to when using conventional linear techniques. Within that frame, ARTS will be responsible for marketing the fund management models, Mimetics will lead the software tool efforts as far as the neural net component is concerned and London Business School will assist both industrial SME's through offering consultancy and spreading the results of the ESTIMATION application trial.
Funding SchemeEST - Marie Curie actions-Early-stage Training