Share market acts as one of the key role for building economy of any developing nation. It is highly complex to determine booms and crashes in the market, so it became one of the interesting and challenging topics for all the business researchers to predict an accurate share value. In this paper we propose an algorithm based on Support Vector Machine (SVM) and k-Nearest Neighbor (k-NN) algorithms to predict next day's closing price share value of a company. Here we take historical data of company i.e. past prices to predict future prices by training them. We use Relief-F feature selection technique used for parameter selection. Performance of SVM and k-NN is calculated using Root mean squared error (RMSE) and Mean absolute percentage error (MAPE). © Research India Publications.