基於結構關聯分析之模糊資料型的財經時間數列相關程度探討

Abstract

本研究計畫首先探討關聯結構理論之技術工具，如關聯結構對財金經濟資料聯合分配偏態的影響，關聯結構模型的估計，關聯結構模型選擇等。接著探討關聯結構理論與投資組合管理關聯，最佳模式選取策略與穩定參數之估計。並發展出一套能對於財金經濟資料或資產價格聯合分配合理假設的方法。最後以實證資料(數據及非數據)分析，提出模糊相關係數的估算。並比較關連結構與模糊相關係數的優缺點，對相關係數資產分配有更妥適的描述，如投資組合管理、風險管理及衍生性金融商品訂價策略，以及提升財務分析的可靠度。

This research proposal aims at investigating structure dependence for financial/economic time series with emphasis on imprecise data such as fuzzy data. The current practice in, say, forecasting economic phenomena from observed time series data, is based upon the methods of copulas and their optimization techniques such as maximum entropy. This can be achieved when data are precise. However, observed time series data in econometrics/finance are often coarse, i.e. of low quality, due to errors in measurements, missing data, sample selection, imprecision in observations. We propose to investigate the use of copula techniques for coarse data on this research project. And compare off even the structure and the advantages and disadvantages of the fuzzy correlation coefficient, the asset allocation of the correlation coefficient is more properly described, such as portfolio management, risk management and financial derivatives pricing strategies, as well as to enhance the reliability of financial analysis. Application domains: Risk management in financial econometrics, actuarial sciences, credit management decision.