In this draft, we consider a hedging strategy concerning only the continuous parts of two asset price processes which have jumps. Two consistent estimators of the hedging strategy, ρ̂ and ρ̃, are presented in terms of realized bipower variation and threshold quadratic variation, respectively. Based on ρ̂, estimators for operational risk, market risk (risk due to jumps) and total risk are investigated. It turns out that the variance of ρ̂ enters into the bias of the operational risk estimator, whereas the variance is mainly due to jump influenced bipower estimation error. The convergence rate of the operational risk estimator (properly centralized) is O ((δt)̄. The convergence rate of the market risk is however O ((δt)̄. Based on ρ̃, the total risk is also studied, and it has the same convergence rate as that based on ρ̂. Besides the interest in financial econometrics, it is also of significance in a statistical sense when we are interested in estimating the quadratic variation of the corresponding unhedgeable residual process.