High-frequency Finance Research Platform,
A Wiki-style Global Project

Abstract:

Our Proposal:

The system that we propose to build will be a web-based, open-source system . Users may upload their modules, and construct interactions between modules. Modules may vary from simple data-retrieval modules to sophisticated financial models. They may also be analysis tools or graphic interfaces. Crucial to the platform is a set of databases, which we intend to collect . By building and connecting modules, the user may use the data to verify or experiment with different financial models. Human interaction is inevitably laborious. Interface will be designed to allow the platform to interact with computer programs. This will allow us to make use of optimization tools to find models. The diagram below shows the overall architecture of the system.

The platform is intended to be open-source. This will encourage researchers all over the world to participate. The project involves expertise from different disciplines, including finance, economics, computational intelligence (for model building), mathematics, computer and behavioural sciences. Following are some of the ways that this platform can be used:

• Regulators will pick and choose the modules in the system (which could be developed by top experts in their fields) to generate market reports, like what we are accustomed to in weather reports. These reports may indicate risk of financial turmoil. Through evolutionary computation, one may attempt to automate the process of finding conditions under which crashes will occur . This will allow us to build a financial early warning system.
• An economist may be interest in developing a new model on volatility. She has no expertise, nor interest, in building computer interfaces. Neither does she have data to test her new model. She can pick up one of the existing volatility modules deposited in the platform and replace it by her own. Then she can test her model with data deposited in the platform, which is something that she was unable to do without the platform.
• A computational intelligence expert may pick up a module on volatility that contains many parameters, and use evolutionary computation to search for parameters under which the volatility model fits the data. Model-fitting is very laborious if done manually. This computational intelligence expert may not have sufficient knowledge to develop new models on volatility, but, with the platform, he can make substantial contribution to perfecting volatility models.
• A researcher may want to develop quantitative trading strategies to stabilize financial markets. What he could do is to develop a module based on his algorithm and test it in the platform. There he will be able to test the algorithm without having to possess data himself.
• A visualization expert may want to use the platform to demonstrate her tools for visualizing complex simulation results. Her contribution will benefit other users of the system.

Related Projects:

• Olsenscale from Olsen Ltd provides indications on foreign exchange fluctuation analogous to weather forecast.
• Olsen Routes from Olsen Ltd demonstrates the idea of computer-aided wind-tunnel testing in high-frequency finance.
• Economic Wind-tunnel Project explains the general idea of economic wind-tunnel testing.
• Computer Aided Constraint Programming Project demonstrates an architecture that could be used by the proposed platform.
• Wikipedia demonstrates the impact of collective intelligence. We hope our system will be a rich repository of computational finance and economic modules.

Good Reads:

• Richard Olsen and Clive Cookson, "How Science can prevent the next bubble", FT.COM 2009.02.12
• Richard Olsen, "How to Trade", THINK ABOUT Press, Olsen Ltd (ISBN: 978-3-9523708-0-3), 2010
• "A Model Approach - more development work is needed to help computer simulations inform economic policy", Nature, Vol.460, Issue 7256, 6 August 2009