The document presents a Bayesian sensitivity analysis methodology for estimating Value at Risk (VaR) and Tail Value at Risk (TVaR). The methodology aims to explain the variability of VaR and TVaR as functions of the parameter θ, which is estimated using maximum likelihood. It allows considering not only variation in the model but also variation in the estimation of the model parameters. The analysis guarantees risk measures that are marginally higher than the real values, ensuring sufficient funds to cover potential extreme losses. In conclusion, the technique provides a relevant way to assess risk while accounting for uncertainty in both the model and parameter estimates.