In this paper we study the problems of pricing and optimizing sidecar and collateralized reinsurance portfolios. The academic literature on sidecar portfolio optimization that takes into account the institutional aspects of the creation of a sidecar is somewhat limited. These aspects include the full collateralization of risk and the separation of the sponsor of risks and the sidecar, which requires the creation of a special purpose insurer. We employ a stochastic programming approach to solve these problems, which has advantages over other methods and captures the underlying institutional structure of the sidecar. We detail the method and numerical results for compound Poisson-distributed risks, showing how optimal portfolios are built by maximizing return while controlling for tail risk.