- 學位論文
建立COVID-19高風險環境下的台指期貨量能理論模型
Building a Volume Theory Model Under the High-Risk COVID-19 Environment for Taiwan Stock Index Futures
摘要
本研究針對COVID-19發生期間的台指期貨的資料作為主要研究對象，研究期間為2019年1月到5月以及2020年1月到5月，以每5分鐘之開盤價、最高價、最低價、收盤價、交易量等日內資料進行研究。由於過去研究多為以價格為基礎去進行研究，因此本研究以價量關係為理論基礎結合最大交易量的概念，提出一個以交易量為理論基礎的量能理論模型。且本研究使用可以處理時間序列相關的類神經網路機器學習工具分別為長短期記憶(Long Short-Term Memory, LSTM)模型以及門控循環單元(Gated Recurrent Unit, GRU)模型，並以二元分類的方式預測台指期貨的價格走勢。實驗結果各類模型以交易量為基礎的量能理論模型都有高於50%準確率，本研究所提出的量能理論模型，具有參考價值。
並列摘要
This research focuses on the impact of COVID-19 on Taiwan stock index futures based on the change of trading volume. The research period is from January to May 2019 and January to May 2020. The data input is based on the daily data such as opening price, highest price, lowest price, closing price, and trading volume every 5 minutes for research. Since most of the past researches is based on price, this research uses the relationship between price and volume as the theoretical basis and combines the concept of maximum transaction volume to propose a theoretical model of volume and energy based on transaction volume. And this research uses neural network-like machine learning tools that can deal with time-series correlations, namely Long Short-Term Memory (LSTM) model and Gated Recurrent Unit (GRU) model. And predict the price trend of the Taiwan Index futures in a binary classification method. In this experiment, the experimental results of various models are based on transaction volume. The theoretical models of energy and quantity have an accuracy rate higher than 50%. The theoretical model of energy and quantity proposed in this research has reference value.