結構性改變下的多元配對交易

Abstract

本論文旨為於共整合理論框架下，提出一種自動搜尋無系統性風險的多元配對投資組合方法，並檢驗其結構性改變性質、數據模型配適適性與統計套利機率。本論文亦使用 S&P 500 進行實證分析與檢驗。

This thesis proposes a statistical procedure for searching for ``multivariate'' pairs trading strategies. This procedure generates market-neutral portfolios (or pairs) with stationary price processes generated from the cointegration properties. In addition, the procedure tests for the existence of structural change points and information criteria in order to ensure the quality of the time series model selections. Using the proposed procedure, we can filter eligible pairs with win probabilities higher than a predetermined threshold. We use S&P 500 minute data for the period 2008-2016 to justify the robustness of the proposed trading procedure.