Prediction Based On Support Vector Machine Price Reversal Point

Stock markets change rapidly, thus the stock price reversal points play a vital role in investment decisions. For a short term case, technical analysis can reveal some features of stock price reversal. But when there are so many technical indicators, it is a problem for the investors to determine which one is the best. Moreover, using a sole technical indicator to predict the reversal points has very low recall and precision rates.In order to solve the problems mentioned above, we propose a novel method which is called stock price reversal points prediction mechanism base on Support Vector Machines (SVM). Ours method first choose five indicators which are MACD, KDJ, RSI, CCI and BIAS as candidates, by calculating the recall and precision rates of these indicators, we choose three out of the five, whose recall and precision rates perform better. They are MACD, CCI and BIAS. Secondly, we combine the three indicators as a reversal point prediction vector, train and test on this vector using SVM to get a better prediction result.The experiment results show our method has higher recall and precision rates than any sole indicator. Trading the stock using our method and regulations will gain higher investment profile than any sole indicator.