Konkordance mezi časovými řadami kurzů měn, statisticko-pravděpodobnostní pohled.

The paper presents one of the possible methodologies for comparing two time series by estimating the probability that both series will increase or decrease (in probability) synchronously. This procedure is analogous to the specific measures of dependence (concordance) between random variables with non-parametric approach (Kendal tau, Blomqvist beta, ...). There is no problem with the probability model. This problem occurs in statistical inference. For observed data (transformed into increasing and decreasing movement), the assumption of independent observations will be easily reversed (with rare exceptions). Therefore, we present a simple Markov model of transitions between states (increasing-increasing, increasing-decreasing, decreasing-increasing, decreasing-decreasing). The assumption of the independence of the observed states is then replaced by the assumption of independent observed transitions.