Parameter Optimization for Combining Lognormal Antithetic Time Series

Abstract

Time series models of serially correlated variables yield biased ï¬tted values and forecasts. Antithetic time series can be created so as to be negatively correlated with the original time series. The original and antithetic time series can be combined so as to eliminate said bias. A system of equations
that deï¬ne the optimal combining parameters is derived.