Complexity in Financial Markets: Models and empirical evidence at the micro and macro level

In this thesis I have used tools and methods lent from Statistical Physics to build models or directly analyze some empirical evidence of Financial Markets. In particular, I analyzed the correlation and distribution properties of the price variations of various stocks, finding a new property which links extreme price fluctuations. This evidence can be quantitatively explained with a simple model. I have conceived and developed an Agent Based Model to explain the main statistical properties of financial time series, addressing in particular the issue of Self-Organization. From a market microstructure perspective, I have studied the problem of the emergence of liquidity crises and the connection between traders' strategies, the spread relaxation after a shock and volatility clustering.