Processing of high frequency data with risk aversion

This paper suggests the calculation methodology of univariate and multivariate absolute risk aversion based on asymptotic analysis of conditional expectation and future excess return variance. In the paper we provide modification of the multivariate econometric algorithm on the assumption of weakly time-varying correlation matrices for which the conditions of positive definiteness were received. We prove theorems on relation between risk aversion of shares and futures. Next, concerning risk aversion of various financial instruments with common underlying asset, we defined a time-varying stochastic no-arbitrage interest rate used in Black-Cox credit risk model.