應用決策樹演算法於高頻交易價格方向預測- 應用台灣指數期貨

Abstract

本研究以市場微觀分析，輔以市場輪廓理論，去分析每筆買賣報價、成交價量及價位分布等資訊，並利用決策樹分類法，訓練連續區間之tick資料，進而生成分類模型，幫助交易者預測未來價格變動的機率，試圖找尋最佳下單價格，增加獲利，也可以降低在買賣時價格滑落或無法成交所造成的損失，讓策略順利執行。

In this study, market microstructure analysis, supplemented by market contour theory to analyze each bid and offer quotations, price distribution volume and price information, and using decision tree classification, tick data of continuous interval training, thereby generating a classification model to help trade who predict the probability of future price movements, trying to find the best single price increase profits, can also reduce the losses in the trading price when the transaction caused by slipping or not, so that the successful implementation of the policy.