This dissertation analyzes risk measures and wealth allocation under a multi-period perspective. We propose a general wealth allocation model and present the classic multi-period risk measures under this perspective, when asset prices time series are modeled by GARCH processes or Geometric Brownian Motions. We further propose an alternative multi-period risk measure, called the Multi-period Relative Value-at-Risk (MRVaR), aiming to correct some undesired features of measures based on absolute values of cash-flows. We then present an analytic bound for the MRVaR and study the consequences of using the bound itself as a risk metric for portfolio optimization. Finally, we study numerically the error we incur when using the bound as risk measure and develop a case study, using assets from BM&FBovespa, to validate the proposed methodology.