统计物理学对经济科学的应用 I.ppt

统计物理学对经济科学的应用 (I) 陈志 4/11/2010 Why do we need statistical physics? Article on the 23 August 1997 issue of The Economist: “The Puzzling Failure of Economics” the fundamental principles governing the complex system called economics are not completely uncovered. Wall Street has been the biggest single recruiter of physics PhDs. Many physicists work as quants (quantitative analysts) --- very common in large investment banks and other financial firms. Physicists welcome the research related finance topics Roughly 10 physics PhDs each year have addressed finance topics with their doctoral research. Why? Jean-Phillipe Bouchaud (a pioneer in this area and the advisor of several such students) said: “Somebody has to train all the physics graduates going into banking and finance, and we want it to be us, not people from other disciplines. To do this we need to establish a scientific presence in the field.” Tools from statistical physics provides a fresh point of view to understanding finance topics. Materials can be found on the web Econophysics : /~hes/econophysics/ Basic knowledge can be found at Wiki: en.W What we look at? For example, we study a time series the SP 500 index (an index of the New York Stock Exchange that consists of 500 companies representative of the US economy) Z(t), the time lag for recordings is Δt. We usually look at the increments and the log return: G(t,Δt)= returns: Z(t+Δt)/Z(t)-1, volatility: standard deviation of time series over a suitable time interval. Why? The fluctuations of Z(t) are usually proportional to Z(t). log return satisfies the addition law: G(t,Δt1+Δt2)=G(t,Δt1)+G(t+Δt1,Δt2) SP 500 index from 1984-1997. An example (I)Distributions: increments Traditional view: The increments are independent and identically distributed (IID) with a well defined second moment. From the central limit theorem: the distribution of G(t) is Gaussian. What we found, for increments of both stock indices and foreign exchange rates: distribu