一個以資產配置策略為基礎的動態智慧型投資組合保險管理系統

投資組合的建構需要有好的時間點進場、妥善的資金分配與完整的投資組合管理，才能獲得良好投資的績效。而目前市面上卻無兼具此三重要元素的投資系統，學術界多數研究也僅針對其中一個項目進行深度探討，但整合性的應用研究卻不多。故本研究最主要的目的為針對適合投資大眾的投資理財系統進行探討，設計一套具有完整投資流程的系統並利用近五年的歷史資料進行驗證，使其更符合實務上的操作情況。本研究提出以資產配置為基礎建置動態智慧型投資組合保險管理系統(DPIMS)，使用倒傳遞類神經網路預測交易訊號、時間不變性投資組合保險策略(TIPP)調整資金配置與使用平均/變異數模型(MV Model)來管理投資組合，並嘗試修正TIPP與MV Model於實務操作時遭遇的問題，且以台灣50成份股為例進行模擬實驗。實驗結果顯示不論是否有加入本研究所提出的成本考量機制改良MV Model與資金回饋機制修正TIPP，以原始理論基礎建立之模型皆具有良好的獲利能力，證實此三個模型組合而成之系統確實為一可有效獲利的投資工具。實驗結果亦顯示加入成本考量機制與資金回饋機制較利於投資偏好為中等以下者使用；風險偏好較高者則是以僅加入成本考量機制後的投資成果最佳，原因可能為資金過度調配導致避險功能喪失，故建議後續研究者可於資金回饋機制的運算加以改良，使其應用於不同風險偏好時皆能有良好的避險與獲利能力。

關鍵字

倒傳遞類神經網路；時間不變型投資組合保險；投資組合選擇；資金回饋

並列摘要

In order to obtain more return on investment , what we needed are three critical factors.There are “market timing”.”asset allocation” and “portfolio selection”. At present, there is no integration investment system on the market currently, and most of the past researches only focus one of this issue to study. Therefore, the purpose of this study is to design a dynamic intelligent portfolio insurance management system for the common investor. For the purpose of making it more consistent the operating conditions on practical, we used the last five years historical data to verify perfomance. This research proposes to build a dynamic intelligent portfolio insurance management system (DPIMS), applying Back-Propagation Neural Network (BPNN) to predict trading signals, Time- Invariant Portfolio Protection Insurance (TIPP) to allocate asset and MV model to manage the investment portfolio. This research also make an attempt to fix TIPP and MV Model problems of practical operation time and exploiting a constituent stock of the TWN 50 Index as a simulation example. Experiment results show the original model which does not join cost improvement and feedback mechanism also has good performance. Therefore, the system design with the theoretical foundation of BPNN.TIPP .MV Model is an effective instrument for investor. Moreover, the result also show the complete system doesn’t suitable for high risk tolerance to use. The reson may be that risk asset overfull allocate leading to lose hedging function.Thence, the proposed for subsequent researchers could be improved in feedback mechanisms and experiment its effectiveness on hedge and profitability performance when applied to different risk preferences.

並列關鍵字

Back-Propagation Neural Network ； Time-Invariant Portfolio Protection Insurance ； Portfolio Selection ； Cash Feedback

參考文獻

15.薛品嶸(2009)，設定停損或停利對投資者行為之影響，國立台灣大學財務金融學研究所，碩士論文。

17.Andre F.-P., William F.-S (1988). Dynamic Strategies for Asset Allocation. Financial Analysts Journal, 44(1), 16-27.

18.Black, F., & Jones, R. W. (1987). Simplifying portfolio insurance. Journal of Portfolio Management, 14(1), 48-51.

19.Brinson, G. P., L. R. Hood, & G. L. Beebower, (1991). Determinants of Portfolio Performance II: An Update. Financial Analyst Journal, 47(3), 40-48.

20.Brinson, G. P., L. R. Hood, & G. L. Beebower (1986). Determinants of

國際替代計量

一個以資產配置策略為基礎的動態智慧型投資組合保險管理系統

全文下載

主題瀏覽