Since　load　forecasting　plays　an　important　role　in　the　planning　and　operation　of　power　industry,　substantial　efforts　are　made　in　improving　the　accuracy　and　reliability　of　load　forecasting.　In　this　paper,　we　develop　a　novel　hybrid　approach　based　on　phase　space　reconstruction　and　least　square　support　vector　for　the　short-term　load　forecasting.　However,　the　proper　parameters　in　phase　space　reconstruction　and　least　square　vector　machine　have　a　significant　effect　on　the　forecasting　performance,　and　there　is　no　standard　solution　for　the　parameter　estimation　problem.　Therefore,　in　this　paper,　the　genetic　algorithm　(GA)　approach　is　employed　to　optimize　the　parameters　of　both　phase　space　reconstruction　and　least　square　support　vector　machine　together.　The　experimental　results　suggest　that　the　joint　optimization　parameter　is　superior　to　the　separate　optimization　solutions.