Under the new regulations in Turkey, due to the electricity regulations the electricity supply chain has changed. Eligible customers have the privilege to buy electricity from different suppliers. Electricity prices are determined in a dynamic market based on the consumption forecasts and production plans. The intermediaries carry a financial risk since they buy the electricity from the market at a dynamic price but sell to their customers based on a constant price. This study aims to determine the value at risk (VaR) calculated due to forecasting errors and compare the forecasting techniques. Forecasting methods including; ARIMA, Grey Prediction with Rolling Mechanism (GPRM), Artificial Neural Networks (ANN), Support Vector Machine (SVM), and Holt's model are used for predicting electricity consumption of a factory and the techniques are compared based on the risk they cause regarding historical VaR. Results show ANN and SVM are the leading forecasting techniques cause a minimum VaR values.