GAUK 588912 - Empirical Validation of Heterogeneous Agent Models
|Principal investigator:||PhDr. Jiří Kukačka Ph.D.|
doc. PhDr. Jozef Baruník Ph.D.
|Description:||In recent financial literature, the Representative Agent Approach and the Efficient Market Hypothesis, dominating in the past, are being replaced by more realistic agent based computational approaches. Moreover, recent financial markets events pointed at the deficiency of knowledge about functioning of this crucial segment of global economy. Although consequences of market fluctuations are worldwide, the essence of problems remains on the level of individual market agents with their heterogeneous beliefs and expectations.
Core idea of the heterogeneous agent modelling is the well documented departure from agents' full rationality. HAMs employ interacting groups of boundedly rational agents to model the financial world and this reasonably realistic methodology appears very successful in replicating observed stylized facts of financial time series.
The aim of the project is to develop algorithms for empirical validation of particular HAM designs. Resulting model will additionally be examined via the out-of-sample verification and contrasted to other 'competing' approaches (VAR, GARCH 'family'...) to assess whether HAMs outperform these well-known models in terms of fitting real financial data and the forecasting performance power. So far only a few proper attempts have been made in this pioneering and highly challenging research field, requiring innovative algorithmic solutions combined with extensive computational capacity.
|Work in grant:||submitted to GAUK in November 2011|
|Finance:||accepted for financing in April 2012
prolonged for financing for 2013
prolonged for financing for 2014