Market Mill Dependence Pattern in the Stock Market: Individual Portraits

This paper continues a series of studies of dependence patterns following from properties of the bivariate probability distribution P(x,y) of two consecutive price increments x (push) and y (response). The paper focuses on individual differences of the P(x,y) for 2000 stocks using a methodology of identification of asymmetric market mill patterns developed in [1,2]. We show that individual asymmetry patterns (portraits) are remarkably stable over time and can be classified into three major groups - correlation, anticorrelation and market mill. We analyze the conditional dynamics resulting from the properties of P(x,y) for all groups and demonstrate that it is trend-following at small push magnitudes and contrarian at large ones