|
Financial and Socioeconomic Modelling
Financial Markets are hierarchical systems, with layers of complexity.
Complementary statistical measures and diverse models are necessary to
sift the available evidence on bubbles, crashes and other market movements.
In the group, we explore the potential for prediction of a multi-layered
approach, incorporating three principal strands, (using sourced data).
We aim
- To identify statistical measures of market co-movement by analysis of
eigenvalues of the variance-covariance matrix of daily price indices and
associated time series.
- To identify noise elements of this matrix, by use of a novel fractional
calculus approach. This will provide a decision-tool to pinpoint future real
market changes.
- To explore the impact of preliminary indications on movement, by means
of statistical physics models, which emphasise connected (or co-operative)
behaviour
These may be described in terms of agent interaction or herd influence in
market trading but, while some limited exposure has been given to these
ideas, our focus is on the granularity of the market response to measurable
change in coherence.
The approach proposed is highly inter-disciplinary and will provide for
analysis of comparative market behaviour across different industrial sectors
both nationally and internationally. As a prototype method for early risk
assessment, we anticipate consequent benefit to informed and strategic
decision-making, together with potential for wider applicability to other
complex social systems. Furthermore, the integrated nature of the approach
will facilitate assessment of intervention strategies under negative market
co-movements.
Researchers:Liam Tuohey,
Martin Crane, Heather J. Ruskin,
Adel Sharkasi, Justin Daly, Ruili Wang, Puspita Deo
|
|
