Search

This paper studies the topological properties of the International Trade Network (ITN) among world countries using a network analysis. We explore the distribu- tions of the most important network statistics measuring connectivity, assortativ- ity and clustering. We show that the topological properties of the weighted rep- resentation of the ITN are very different from those obtained by a binary network approach. In particular, we find that: (i) the majority of countries are character- ized by weak trade relationships, (ii) well connected countries tend to trade with poorly connected partners and (iii) countries holding more intense trade relation- ships are more clustered. Finally, we display that all structural properties of the ITN have remained remarkably stable over time.

This Bachelor thesis joins two areas which currently attracted increased attention --- Behavioral Finance theory and agent based models. Goal of this work was to overcome deficiencies of known agent based models and to create a model focusing on decision-making of individual agents, optimally using findings of Behavioral finance. We managed to find such a model and proved that it is capable of reproducing basic statistical properties of market prices, which is a basic test of the quality of the model. Its further use in investigation of Behavioral phenomena is currently impossible, because we are not able to sufficiently constraint the space of parameters. At the end of the thesis we discuss how it would be possible to overcome this obstacle.

The main objective of this thesis is to inspect the abilities of the Ising model to exhibit selected statistical properties, or stylized facts, that are common to a wide range of financial assets. The investigated properties are heteroskedasticity of returns, rapidly decaying linear autocorrelation, volatility clustering, heavy tails, negative skewness and non-Gaussianity of the return distribution. In the first part of the thesis, we test the presence of these stylized facts in S&P 500 daily returns over the last 30 years. The main part of the thesis is dedicated to the Ising model-based simulations and to discussion of the results. New features such as Poisson process governed lag or magnetisation dependent trading activity are incorporated in the model. We conclude that the Ising model is able to convincingly replicate most of the examined statistical properties while even more satisfactory results can be obtained with appropriate tuning. 1