A virtual power plant (VPP) is analogous to a microgrid. However, a VPP is designed to participate in the electricity markets, such as the day-ahead (DA) and the real-time (RT) markets, while a microgrid does not. In order to participate in the electricity markets, historical data evaluation is necessary to predict market behavior. There are three main parameters to predict: the day-ahead market price, the real-time market price, and the renewable energy sources data such as the wind speed. One of the popular techniques used for forecasting is the autoregressive integrated moving average (ARIMA) model. In this paper, an improved ARIMA model is introduced. Using the proposed sequential ARIMA model, the VPP operator can forecast the uncertain parameters with higher accuracy, improving the optimization decisions taken by the VPP. This will reduce the risk of failure to participate in the electricity markets and give a better opportunity to achieve better results in terms of the optimization objectives such as cost minimization and profit maximization.