Historically, financial variables’ volatilities are drastically affected during economical crisis periods. In particular, this statement is valid for the exchange rate between brazilian and north american currencies, the São Paulo Stock Exchange Index (Ibovespa) and the Emerging Market Bond Index + Brazil, usually interpreted as a measure of brazilian sovereign risk. Therefore, the presence of a common factor affecting the volatilities of these three variables becomes a plausible assumption. This work intends to extract this latent factor applying the quasi-maximum likelihood estimator into an adapted stochastic volatility model. The estimation requires the Kalman filtering on its diffuse version, once it’s supposed that the volatilities follow a non stationary process. The results indicated the presence of one common factor driving the mentioned volatilities, confirming the previous expectations.