Support Vector Regression with A Priori Knowledge Used in Order Execution Strategies Based on VWAP

In this article, we propose a novel application for Support Vector Regression (SVR) for strategies of executing orders on stock exchanges. We use SVR for predicting volume participation function in execution strategies which try to achieve Volume Weighted Average Price (VWAP) measure of quality of the execution. Moreover, we use SVR with a priori knowledge about stock prices in order to further improve the cost of order execution. The main result is that SVR outperforms tested null hypotheses such as prediction based on average from historical data. SVR with additional knowledge about prices improves the final execution error. The tests were performed on real stock data from NASDAQ exchange.