Stochastic nature of wind power generation creates techno-economic challenges to power system operations. Ancillary services such as operational reserves are used to address such challenges. Ancillary services increase the operating cost of utility companies and increased operating cost is suggested to be paid by Wind Power Producers (WPP). Penalizing wind power deviation from the forecast is one of the effective ways to shift increased operating costs from utility companies to WPPs. WPPs profit is getting reduced due to high deviation charges. Deviation charges can be reduced by enhancing wind power forecasting accuracy. Several models such as statistical methods and artificial neural networks are available for wind power forecasting. However, there is a scope of accuracy improvement, because forecasting errors result in significant techno-economic challenges. Also, existing forecasting approaches focus on error matrices rather than economic factors such as deviation charges. Expressing forecasting performance in terms of money will motivate WPPs to improve the forecasting accuracy to reduce deviation charges. The deviation charge estimation of an individual plant is obsolete due to the occurrence of negatively correlated forecasting errors of different plants. Therefore, this paper proposes a novel Support Vector Regression (SVR) based aggregated wind forecasting model for deviation charge reduction. SVR models are widely used in the last few years, due to its superior performance over ANN and statistical methods. Hyper-parameters of proposed SVR model is intelligently tuned using Particle Swarm Optimization (PSO) to ensure optimum model performance. Results show that the proposed model has low deviation charges compared to reference models.