Up until the financial crisis of 2008, financial models were simplified by incorporating assumptions known to be relatively unrealistic. The trade-off was in the generation of models which were easier and faster to use but which contained data, which was approximate rather than wholly accurate. Applying statistical and probability theory to aspects of the financial crisis made it clear that events such as this – considered extreme and therefore unlikely – are actually more common than previous models had assumed. A consequence is that more recent financial mathematical modelling has increased in computational complexity. Making the most of these models was the key driver for the EU-funded STRIKE project. STRIKE sought to do so through the creation of a specially trained network of young European scientists. Beyond number crunching One of the key challenges that STRIKE set out to overcome was that of combining various methodologies and approaches, such that they complimented each other. This curated training incorporated mathematical analysis, sophisticated numerical methods, stochastic simulation methods, financial modelling (with deep qualitative and quantitative financial market data), optimal control techniques and model validation techniques. It also went beyond an understanding of the numbers, towards a better consideration of their potential social impact. The theoretical framework for the research was the European response to the financial crisis evidencing characteristics described as ‘contagion’ and ‘herding’, beyond standard models (such as Black-Scholes-Merton's model used as an investment tool, especially for derivatives). To develop a new, more robust non-linear (or non-Gaussian) model relied on meaningful inputs and so data for the project was provided by companies and banks. As the project coordinator Prof. Matthias Ehrhardt illuminates, ‘This allowed us to compare simulation outcomes with real data from history. The data also of course helped us to calibrate models.’ These models were then collated into the STRIKE Computational Finance Toolbox. Prof. Ehrhardt further explains, ‘This enables example runs and documentation with background information, allowing for input changes and then the observation of the resulting impact, caused by these settings. Several implementations also exploited the computational power of graphical parallel processing units (such as those used for computer gaming) to speed-up simulations.’ What Czexit can learn from quantum theory? Further elaborating on the utility of the toolkit, Prof. Ehrhardt calls for the understanding of financial systems to follow that of the evolution of physics, which has successfully moved beyond linear causality assumptions. As he summarises, ‘With STRIKE, outcomes hedging and risk analysis can be better trusted or assumptions adapted in a more, timely manner.’ Adhering to its aim of providing useful decision making tools and a knowledge centred collaborative network, STRIKE has developed models applicable to real situations and problems. Prof. Ehrhardt highlights one example of very timely significance when recollecting that, ‘A special example for our modelling was taking the situation where a country wanted to join the EU, when we were able to properly estimate the temporal evolution of the interest rate.’ He elaborates that, ‘Now a similar technique can be used when a country leaves the EU or decouples its currency from the Euro. The later situation fits to ‘Czexit’, whereby in April 2017, the Czech Republic decided to decouple the Czech Koruna from the Euro.’ There are wide-ranging real world applications for the algorithms STRIKE developed, including apps providing customers with stock market advice or energy market pricing information, to help citizens make consumption decisions. For now, as well as maintaining the research consortium, the project will also publish a book of research outcomes and further develop its biennial International Conference on Computational Finance series.
STRIKE, financial modelling, financial crisis, currency decoupling, interest rates, financial algorithm, risk analysis, risk assessment, decision making toolkit