At Fleming, Kirby e Ostdiek (2001), evidences are found that volatility timming models, have signicant economic value when comparing with the simple unconditional variance matrix, in a framework of portfolio optimization. Going further, this work analyze if the more complex Constant (CCC) and Dynamic (DCC) Conditional Corrrelation models, suggested respectivily by Bollerslev (1990) and Engle (2002) can have a higher performance. The results found depend on the investor´s preference. A more risk averse investor has a higher utility level employing the DCC and CCC models when comparing with the simple exponencial moving avarage model, popularized by RiskMetrics. This happens because the DCC and CCC models usually have smaller standard deviation and return. Futhermore, it is not possible to assert, like at Fleming, Kirby e Ostdiek (2001), that volatility timming models have higher economic value.