Journal article: This paper focuses on the optimal program of a non life insurance company which maximizes an expected risk ajusted return on capital (RAROC) criterium under a conditional value at risk (CVaR) shortfall constraint, by simultaneously choosing the best allocation of her portfolio and the best level of her capital. A solution is provided by refering to the framework of Concave Convex Fractional programs ( CCFP). The risk the company is facing with jointly comes from the financial assets and from the potential losses, whose dependence structure is taken into account through copulas. The methodology is applied to real data of a French non life insurance company. Using Monte Carlo simulations, we determine the optimal solution and we implement sensitivity analyses.