Topological Isomorphisms of Human Brain and Financial Market Networks
Topological Isomorphisms of Human Brain and Financial Market Networks
复制标题
人脑与金融市场网络的拓扑同构
DOI：
--
复制
发表时间：
2011
影响因子：
3
通讯作者：
E. Bullmore
中科院分区：
文献类型：
--
作者：
P. Vértes;R. M. Nicol;S. Chapman;N. Watkins;Duncan A. Robertson;E. Bullmore
尽管人类大脑和金融市场之间经常有隐喻和概念上的联系，但这种类比的严格物理或数学基础在很大程度上仍未得到探索。在这里，我们应用统计和图理论方法来研究两个数据集-纽约证券交易所3年期间90只股票的时间序列，以及健康志愿者在休息时大脑功能的10分钟功能性MRI扫描过程中从90个大脑区域获得的fmri衍生时间序列。尽管这两个数据集之间存在许多明显的实质性差异，但图形分析在全局网络拓扑特性方面显示出惊人的共性。人类大脑和市场网络都是非随机的、小世界的、模块化的、分层的系统，它们的厚尾度分布表明存在高度连接的枢纽。这些性质不能简单地用股票价格收益的单变量时间序列统计来解释。这种程度的拓扑同构表明，大脑和市场可以被广泛地视为同一网络家族的成员。然而，这两个系统在拓扑结构上并不相同。金融市场比大脑网络更高效、更模块化——对信息处理的优化程度更高；但由于轮毂缺失，对系统解体的抵抗力也较弱。我们的结论是，大脑和市场之间的概念联系不仅仅是隐喻性的；相反，这两种信息处理系统可以用相同的数学语言进行严格比较，并且在某种程度上经常具有共同的重要拓扑特性。在系统神经科学和金融市场统计物理学之间的图形理论中介界面的进一步工作中，将会有有趣的科学套利机会。
Although metaphorical and conceptual connections between the human brain and the financial markets have often been drawn, rigorous physical or mathematical underpinnings of this analogy remain largely unexplored. Here, we apply a statistical and graph theoretic approach to the study of two datasets – the time series of 90 stocks from the New York stock exchange over a 3-year period, and the fMRI-derived time series acquired from 90 brain regions over the course of a 10-min-long functional MRI scan of resting brain function in healthy volunteers. Despite the many obvious substantive differences between these two datasets, graphical analysis demonstrated striking commonalities in terms of global network topological properties. Both the human brain and the market networks were non-random, small-world, modular, hierarchical systems with fat-tailed degree distributions indicating the presence of highly connected hubs. These properties could not be trivially explained by the univariate time series statistics of stock price returns. This degree of topological isomorphism suggests that brains and markets can be regarded broadly as members of the same family of networks. The two systems, however, were not topologically identical. The financial market was more efficient and more modular – more highly optimized for information processing – than the brain networks; but also less robust to systemic disintegration as a result of hub deletion. We conclude that the conceptual connections between brains and markets are not merely metaphorical; rather these two information processing systems can be rigorously compared in the same mathematical language and turn out often to share important topological properties in common to some degree. There will be interesting scientific arbitrage opportunities in further work at the graph-theoretically mediated interface between systems neuroscience and the statistical physics of financial markets.