最適指數複製法之自動化建置：以ETF50為例; Automatic construction of the best index duplicates Approach : Take ETF50 as an example

NCUIR > School of Management at National Central University > Executive Master of Finance Management > Electronic Thesis & Dissertation > Item 987654321/12328

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/12328

Title:	最適指數複製法之自動化建置：以ETF50為例;Automatic construction of the best index duplicates Approach : Take ETF50 as an example
Authors:	洪真逢;Chen-Feng Hung
Contributors:	財務金融學系碩士在職專班
Keywords:	程式交易;ETF50指數;指數複製;二次規化演算法;最適投資組合;index duplicates;optimal;quadractic optimization programming
Date:	2005-07-05
Issue Date:	2009-09-22 14:45:12 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	摘要本研究主要目的在探討台灣50指數成分股現貨、台灣50指數期貨二市場間之套利機制，以台灣50指數為標的，研究了從2003年7月1日上台灣50指數首次發佈之日起，一直到2004年12月31日為止的這個時間段中的「台灣50指數模擬最適投資組合」的追蹤問題。文中應用平均數-變異數分析(Mean-Variance Analysis)，透過區分個股及整個投資組合的報酬與風險，來建構「指數模擬最適投資組合(index tracking optimal investment portfolio」。因為一個投資組合的風險性，取決於所持有的投資標的的共變異數，而非個別風險性的平均值，一群高風險性的股票仍有可能建構出一個低風險的投資組合。並研究指數複製、參數設計、模型及運用，而且為了比較，以移動視窗(moving window)模擬方法同時檢核了四種組合（不同之建構期間與持有期間）的預測能力。針對模擬指數最適投資組合追蹤的問題，運用二次規劃演算法（quadractic optimization programming）在MATLAB7.01環境下進行最佳化求解。 1. 「二次規劃最適化模型(最適化法)」與「產業限制(分層選取法)＋二次規劃最適化模型(最適化法)」無論在樣本內還是在樣本外的追蹤效果都相當不錯。 2. 「產業限制(分層選取法)＋二次規劃最適化模型(最適化法)」的統計檢定結果及報酬率差異優於「二次規劃最適化模型(最適化法)」，顯示經過產業別篩選的投資組合在複製「台灣50指數模擬最適投資組合」的能力更佳。 3. 其他條件不變的情況下，樣本內的時間窗越長，追蹤效果越好，而且樣本內追蹤效果越好，樣本外的表現也越好。 4. 研究中最佳化後的「台灣50指數模擬最適投資組合」的報酬率比台灣50期貨指數的報酬率稍差，但「台灣50指數模擬最適投資組合」的流動性相當優秀，從側面支持了本研究具有一定的實際應用價值。 5. 為加快實驗時間暨提升實驗準確性，除二次規劃模型最佳化求解外，尚在系統開發過程中，增加實驗自動化及彈性化模組功能，使整個實驗過程完全自動化。因此實務上若配合市場相關的套利交易機制與即時交易資料，本研究所開發之程式可應用於電腦程式交易。 The purpose of this experimental study was to examine arbitrage mechanism of Taiwan50 index and Taiwan 50 index futureses , regard Taiwan 50 indexes as the target, from study the issue day for the first time of Taiwan 50 indexes of since July 1 , 2003ing, until the track effect of ' the optimal investment portfolio of Taiwan 50 index simulations ' in this time slot on December 31 , 2004. In the article applied ' Mean-Variance Analysis' , through distinguishing the rate of returns and risk of stock and the whole investment portfolio, to build constructing ' optimization portfolio of index tracking simulation . Because a risk of making the investment portfolio , the becoming different altogether and counting of the ones that depend on the investment portfolio’s covariance, but not the average of the specific risk, the stock of a group of high risk may still build the investment portfolio which construct out a low risk . And study index duplicating , parameter design , model and using , and in order to compare, in order to moveing window simulation method examine core prediction ability of four association (building during constructing different and when holding ) at the same time. Direct against the question that the optimization portfolio of the simulation index is followed the trail of , use the planning two times to perform the algorithm (quadractic optimization programming ) to ask and solve the optimization under the enviroment of MATLAB7.01. 1. The 'quadractic optimization model' and ' industry's restriction (stratification) + the quadractic optimization model' is no matter in the sample or the track result out off the sample is pretty good . 2. The 'quadractic optimization model' and ' industry's restriction (stratification) + the quadractic optimization model' statistics assay result and rate of returns difference superior to ' quadractic optimization model', show that the ability to duplicate ' the optimal investment portfolio of Taiwan 50 index simulations ' in investment portfolio not screened through the industry is better. 3. Under the situation that other conditions do not change, the longer the time window in-sample, track effect of in-sample is better than track effect of out-of-sample. And track effect in-sample that the better the result is, the better the behavior outside-of-sample is. 4. In this studying, the rate of returns of ' the optimal investment portfolio of Taiwan 50 index simulations ' is slightly worse than the rate of returns of Taiwan 50 futures indexes, but the liquidity of ' the optimal investment portfolio of Taiwan 50 index simulations' is quite outstanding , have supported this research to have certain actual using value from the side. 5. In order to accelerate experiment time and improve experiment accuracy, except that the planning model optimization two times is asked and solved , still increase experiment automation and elasticity mould group's function in the course of developing systematically, make the whole experiment course totally automized. So cooperates with arbitrage trade mechanism and real-time market trade data, the procedure that this research institute develops can apply to the computer program trade.
Appears in Collections:	[Executive Master of Finance Management] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...